92166
UNDERSTANDING RISK
IN AN EVOLVING WORLD
Emerging Best Practices in Natural Disaster Risk Assessment
GLOBAL FACILITY FOR DISASTER REDUCTION AND RECOVERY
UNDERSTANDING RISK
IN AN EVOLVING WORLD
Emerging Best Practices in Natural Disaster Risk Assessment
GLOBAL FACILITY FOR DISASTER REDUCTION AND RECOVERY
© 2014 International Bank for Reconstruction and
Development / International Development Association or
The World Bank
1818 H Street NW
Washington DC 20433
Telephone: 202-473-1000
Internet: www.worldbank.org
This work is a product of the staff of The World Bank with
external contributions. The findings, interpretations, and
conclusions expressed in this work do not necessarily
reflect the views of The World Bank, its Board of Executive
Directors, or the governments they represent.
The World Bank does not guarantee the accuracy of
the data included in this work. The boundaries, colors,
denominations, and other information shown on any map
in this work do not imply any judgment on the part of The
World Bank concerning the legal status of any territory or
the endorsement or acceptance of such boundaries.
Rights and Permissions
The material in this work is subject to copyright. Because
The World Bank encourages dissemination of its
knowledge, this work may be reproduced, in whole or in
part, for noncommercial purposes as long as full attribution
to this work is given.
Any queries on rights and licenses, including subsidiary
rights, should be addressed to the Office of the Publisher,
The World Bank, 1818 H Street NW, Washington, DC 20433,
USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org.
UNDERSTANDING RISK IN AN EVOLVING WORLD
TABLE OF CONTENTS
13 Foreword
14 Acknowledgments
16 Abbreviations
18 Overview
20 Risk Information as the Basis for Decision Making
22 A Framework for Quantifying and Understanding Risk
23 Advances in Disaster Risk Assessment and Key Remaining Challenges
27 Recommendations for Future Risk Assessments.
30 Recommendations toward the Next Hyogo Framework for Action
31 Endnotes
31 References
Chapter 01
32 Introduction
34 About This Publication
35 A Brief History of Risk Assessment
38 The Rise of Open Models and Data: The Changing Risk Assessment Paradigm
41 Aligning and Targeting Risk Assessments
43 Endnotes
43 References
Chapter 02
44 Progress, Achievements, & Remaining Challenges
in Risk Assessment
45 Hazard Assessment
53 Exposure
59 Vulnerability and Loss
64 Hazard and Risk Assessment Tools
68 Creating Platforms and Partnerships to Enable the Development of Risk Assessments
71 Endnotes
71 References
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
Case Study Color Key TABLE OF CONTENTS
Chapter 03
74 Case Studies Highlighting Emerging Best Practices
76 Open Data for Resilience Initiative (OpenDRI)
Data for Risk
80 Open Cities: Application of the Open Data for Resilience Initiative in South Asia and the
Lessons Learned
86 Preliminary Survey of Government Engagement with Volunteered
Geographic Information
91 Collection of Exposure Data to Underpin Natural Hazard Risk Assessments in Indonesia
and the Philippines
95 International Collaboration of Space Agencies to Support Disaster Risk Management
Using Satellite Earth Observation
98 Global Earthquake Model
101 Global Probabilistic Risk Assessment: A Key Input into Analysis for the 2013 and 2015
Modelling Global Assessment Reports
Developments
107 Global Water-related Disaster Risk Indicators Assessing Real Phenomena of Flood
Disasters: Think Locally, Act Globally
112 Government-to-Government Risk Assessment Capacity Building in Australasia
Risk Assessment
Case Studies 120 Informing Disaster Risk Management Plans in Aqaba, Jordan, through Urban Seismic Risk
Mapping
123 Tsunami Risk Reduction: Are We Better Prepared Today Than in 2004?
127 World Bank Probabilistic Risk Assessment (CAPRA) Program for Latin America and the
Caribbean: Experiences and Lessons Learned
132 Detailed Island Risk Assessment in Maldives to Inform Disaster Risk Reduction and
Climate Change Adaptation
136 Malawi: How Risk Information Guides an Integrated Flood Management Action Plan
141 Reducing Seismic Risk to Public Buildings in Turkey
145 Applying Multi-Hazard Risk Assessment to the Development of a Seismic Retrofit
Program for Public Schools in Metro Manila, Philippines
149 Morocco Comprehensive Risk Assessment Study
153 Risk Assessment for Financial Resilience: The Approach of the World Bank
160 The Pacific Catastrophe Risk Assessment Initiative
TABLE OF CONTENTS Case Study Color Key
163 From Multi-Risk Assessment to Multi-Risk Governance: Recommendations for
Future Directions Participation,
Collaboration, and
168 Build Back Better: Where Knowledge Is Not Enough Communication
172 InaSAFE: Preparing Communities to Be a Step Ahead
177 Global River Flood Risk Assessments
185 Delivering Risk Information for a Future Climate in the Pacific Future of Risk
191 A Framework for Modelling Future Urban Disaster Risk
198 Endnotes
201 References
Chapter 04
210 Recommendations
219 Endnotes
220 Photo Credits
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
LIST OF FIGURES
18 Overview
21 Figure O—1 The components for assessing risk and the difference between “impact”
and “risk.”
23 Figure O—2 Risk as a function of hazard, exposure, and vulnerability.
Chapter 01
32 Introduction
39 Figure 01—1 What makes data “open.”
Chapter 02
44 Progress, Achievements, & Remaining Challenges
in Risk Assessment
47 Figure 02—1 Hypothetical drought index showing periods of extreme dryness (above
the dotted red line) and periods of extreme wetness (below the dotted blue line); the
historical record does not capture extreme dry and wet periods experienced prior to its
start in 1900.
59 Figure 02—2 The relationship between hazard intensity and damage to structures.
65 Figure 02—3 Sample software package review.
Chapter 03
74 Case Studies Highlighting Emerging Best Practices
77 Figure 03—1 Examples of locations of GeoNodes supported by the World Bank
and GFDRR.
87 Figure 03—2 Change detection using OSM.
92 Figure 03—3 LiDAR provides an opportunity to map and visualize in detail the highly
urbanized environment of Manila.
93 Figure 03—4 Application of aerial imagery, LiDAR data, and land-use mapping to
develop exposure database (Taguig City).
94 Figure 03—5 Growth in exposure data through crowdsourced (OSM) mapping of
buildings and infrastructure in three locations in Indonesia.
99 Figure 03—6 A fuller picture of seismic history is obtained when instrumentally recorded
events are combined with events from historical records (in pink).
LIST OF FIGURES
103 Figure 03—7 Example of the 5km x 5km grid constituting the exposure database
for GAR13.
109 Figure 03—8 Effects of water infrastructure in reducing flood inundation depths for 50-
year floods.
113 Figure 03—9 Earthquake hazard map of central Sulawesi Province, developed
collaboratively by Badan Geologi and AIFDR.
114 Figure 03—10 Badan Geologi and Geoscience Australia staff working collaboratively on
probabilistic seismic hazard maps for Indonesia.
115 Figure 03—11 Badan Geologi and Geoscience Australia staff collect volcanic ash samples
from a roadside agricultural plot of land approximately 10km from the summit of Ciremai
volcano, West Java, in 2010.
116 Figure 03—12 The dispersal of volcanic ash from the last historical eruption of Guntur in
1840, as produced by the Volcanology and Geological Disaster Mitigation Centre.
117 Figure 03—13 Modelled depths for a flood equivalent to that experienced in Manila
during Typhoon Ketsana in 2009.
120 Figure 03—14 Historical seismicity in Jordan.
121 Figure 03—15 Jordan’s fault system.
133 Figure 03—16 The safe island concept.
134 Figure 03—17 The islands selected for detailed multi-hazard risk assessment.
138 Figure 03—18 1-in-100-year flood extent (in pale blue) around the Elephant Marshes of
the Lower Shire Valley, Malawi.
138 Figure 03—19 Flood zoning in the area of the Elephant Marshes based on different
return period flood events.
142 Figure 03—20 Prioritization methodology for high seismic risk public buildings.
147 Figure 03—21 Estimated Metro Manila student fatalities per school building for a
magnitude 7.2 West Valley fault scenario earthquake occurring in the daytime.
154 Figure 03—22 Flowchart for financial decision making.
169 Figure 03—23 Two common housing types in Padang: Unreinforced masonry
construction using river stone and mortar with no reinforcement (left) versus confined
masonry construction using steel-reinforced concrete columns in the corners and tops
of walls (right).
173 Figure 03—24 InaSAFE can be used to improve understanding of the impact of disaster
events, such as floods in Jakarta.
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
LIST OF FIGURES
176 Figure 03—25 QGIS2.0 with the InaSAFE 2.0 dock showing a map and indicative results
for an assessment of the impact of flooding on roads in Jakarta.
178 Figure 03—26 Observed flood extents in Bangladesh during July and August 2004:
Dartmouth Flood Observatory database versus GLOFRIS model.
179 Figure 03—27 Map of modelled inundation extent and depth in Nigeria using GLOFRIS.
Maps of this type can be used to assess which areas are exposed to flooding.
180 Figure 03—28 Maps of Nigeria showing the modelled results of the number of people
affected per state (expressed as a percentage of the total population per state) for
floods of different severities. Maps of this type can be used for identifying risk hot spots.
181 Figure 03—29 People living in flood-prone areas in different regions, 2010–2050.
182 Figure 03—30 Annual exposed GDP to flooding in 2010 and 2050, under different
assumptions of flood protection standards.
184 Figure 03—31 Historical tropical cyclone tracks for the period 1981–2000 (top) and
tropical-cyclone-like vortices extracted from a 20-year simulation using a general
circulation model (bottom).
187 Figure 03—32 Ensemble mean proportion of cyclones for current and future climate in
the Northern Hemisphere (left) and Southern Hemisphere (right).
188 Figure 03—33 Individual regional end-of-century exceedance probability curves for
ensemble members (blue) compared to the current climate exceedance probability
curve (green).
189 Figure 03—34 Ensemble mean 250-year losses across the Pacific as a proportion of
Pacific Island countries’ GDP for current climate conditions (1981–2000).
189 Figure 03—35 Ensemble mean change in 250-year return period loss.
192 Figure 03—36 The three components of risk and their time dependence.
193 Figure 03—37 Incrementally expanding buildings and corresponding changes
in vulnerability.
194 Figure 03—38 Number of buildings sustaining heavy damage or collapse from a single
ground motion field, at six different time periods.
195 Figure 03—39 Full distribution of the number of buildings sustaining heavy damage or
collapse, for six different time frames.
196 Figure 03—40 Expected number of buildings sustaining heavy damage or collapse as a
function of time, with confidence interval.
LIST OF BOXES
Chapter 01
32 Introduction
34 Box 01—1 How Risk Information Contributes to Mainstreaming of DRM in World Bank
Group Operations
38 Box 01—2 OpenStreetMap
39 Box 01—3 Community Mapping in Indonesia
41 Box 01—4 Defining “Open”
Chapter 02
44 Progress, Achievements, & Remaining Challenges
in Risk Assessment
46 Box 02—1 Multi-Peril Risk Assessment: An Overview
47 Box 02—2 Assessing Damage and Loss Caused by Drought: Example of a
Deterministic Assessment
48 Box 02—3 A Cost-Benefit Analysis of Livestock Protection in Disaster Risk Management
51 Box 02—4 The Importance of Accurate Elevation Data for Understanding
Tsunami Hazard
54 Box 02—5 Global Exposure Data Sets
57 Box 02—6 Indirect Characterization of Exposure
58 Box 02—7 How Study Scale Drives Exposure Data Collection Methods
60 Box 02—8 The Uses of Loss Inventories
61 Box 02—9 Incorporating Disaster Resilience into Cultural Heritage Buildings in Bhutan
66 Box 02—10 Training in Use of Risk Models: The GEM Perspective
67 Box 02—11 The Understanding Risk Community
69 Box 02—12 Willis Research Network
70 Box 02—13 Participatory Earthquake Risk Assessment in Dhaka
Chapter 03
74 Case Studies Highlighting Emerging Best Practices
78 Box 03—1 Typhoon Yolanda GeoNode: An Example of the Collaborative Effort Possible
under OpenDRI
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
LIST OF BOXES
96 Box 03—2 International Charter Space and Major Disasters
97 Box 03—3 Innovations in Earth Observation over the Coming Decade
118 Box 03—4 Factors Leading to Successful Technical Capacity Building
124 Box 03—5 The Challenge of Multiple Tsunami Hazard Maps in Padang, Indonesia
150 Box 03—6 Risk Assessments as an Advocacy Tool for DRM in the Middle East
and North Africa
157 Box 03—7 R-FONDEN: The Financial Catastrophe Risk Model of the Ministry of Finance
and Public Credit in Mexico
159 Box 03—8 Southeast Europe and Caucasus Catastrophe Risk Insurance Facility
LIST OF TABLES
Chapter 01
32 Introduction
42 Table 01—1 Sample Risk Assessment Products and Their Attributes
Chapter 02
44 Progress, Achievements, & Remaining Challenges
in Risk Assessment
49 Table 02—1 Examples of Globally Available Hazard-related Data
55 Table 02—2 Categories of a Comprehensive Exposure Model
63 Table 02—3 Sources of Disaster Loss Data
Chapter 03
74 Case Studies Highlighting Emerging Best Practices
107 Table 03—1 Basic Characteristics of the Three River Basins
108 Table 03—2 Historical Flood Disasters in the Three River Basins
110 Table 03—3 Potential Flood Inundation Areas in the Three River Basins (considering or
omitting dams and flood protection)
110 Table 03—4 People Potentially Affected by Flood Inundation (considering or omitting
dams and flood protection)
121 Table 03—5 Seismic Risk Scenario for Aqaba (maximum magnitude 7.5 earthquake)
122 Table 03—6 Economic and Financial Impacts of Earthquake Scenario (magnitude
7.5 earthquake)
143 Table 03—7 Building Classifications Used in Prioritization Methodology
144 Table 03—8 Prioritization for Reconstruction and Rebuilding
174 Table 03—9 Hazard Data Accepted in InaSAFE 2.0
174 Table 03—10 Exposure Data Accepted in InaSAFE 2.0
174 Table 03—11 Sample Impact Functions
186 Table 03—12 Changes in Key Tropical Cyclone–related Parameters for the Five-
member Ensemble
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
FOREWORD
In 1999, the most powerful tropical cyclone ever effect the social and political will necessary to build
recorded in the North Indian Ocean made landfall resilience before disasters occur, sparing countless
in the state of Odisha, India, bringing catastrophic lives and better preserving the fragile prosperity
losses in human life and property. With nearly gains of the world’s most vulnerable communities.
10,000 fatalities and US$5 billion in damages,
Today, another powerful storm is forming at
the tragedy revealed a stark need for disaster risk
the intersection of population growth, rapid
reduction and preparedness.
urbanization, and climate change—one that
The following decade saw an impressive and threatens to undo decades of progress toward
sustained effort by the government of Odisha development goals. To prepare communities to
and partners to identify and mitigate cyclone weather its impact, we will need to prioritize
risk, resulting in the construction of emergency disaster risk assessment to inform our collective
roadways, reinforced bridges, shelters, improved resources, and enable risk management with
coastal embankments, and extensive early warning unprecedented levels of innovation, cooperation,
systems. When the similarly intense Cyclone Phailin and scale.
made landfall in Odisha late last year, fatalities were
Underpinning successes like these is accurate
minimal: the region experienced a 99.6 percent
and actionable risk information. This publication
reduction from the 1999 storm, in large part due to
highlights some of the influential efforts—by
these effective disaster risk management initiatives.
technical specialists, institutions, and governments
Case studies like this clearly show the potential of around the world—to create and communicate risk
targeted interventions to reduce human suffering information faster and at lower cost, to improve the
and lessen the impact of major natural disasters. quality and transparency of risk information, and to
What makes these and other efforts possible, enable more local engagement in the production of
however, is accurate and actionable risk assessment. authoritative risk information than ever before.
Far too often, tragedies like the 1999 Odisha cyclone This publication is a small but valuable contribution
are the drivers for change, but the future can be toward that effort. We hope you will work alongside
different. If brought to scale and embedded within us as we seek to better understand risk in a
development efforts, disaster risk assessment can changing world.
Francis Ghesquiere
Head, GFDRR Secretariat, Manager DRM Practice Group, The World Bank
13
14
ACKNOWLEDGMENTS
This publication was prepared by a team led by Alanna (GFDRR); M. V. De Guzman (Department of Foreign
Simpson and consisting of Rick Murnane, Keiko Saito, Affairs and Trade, Manila); Manuela Di Mauro (UNISDR);
Emma Phillips, Rob Reid, and Anne Himmelfarb. Angela Di Ruocco (Analisi e Monitoraggio del Rischio
Ambientale); Andrew Eddy (Athena Global); Christopher
This publication benefited from inputs and contributions
T. Emrich (Hazards and Vulnerability Research Institute,
from the following: Jeroen C. J. H. Aerts (Institute for
University of South Carolina); Nicole Fassina (World
Environmental Studies and Amsterdam Global Change
Society for the Protection of Animals); Nishara Fernando
Institute, VU University Amsterdam); Vyron Antoniou
(University of Colombo); Kevin Fleming (Helmholtz Centre
(Hellenic Military Geographical Service); Christoph Aubrecht Potsdam, German Research Centre for Geosciences
(World Bank); W. C. Arthur (Geoscience Australia); Elif [GFZ], Potsdam); Marc Forni (World Bank); Sergio Freire
Ayhan (World Bank); Abigail Baca (World Bank); Claudia (European Commission Joint Research Center); Stuart Frye
Bach (UNU-EHS); Axel Baeumler (World Bank); Philippe (NASA); Francesco Gaetani (Group on Earth Observations
Bally (European Space Agency); Sofia Basiouka (National Secretariat); Melanie Gall (Hazards and Vulnerability
Technical University of Athens); B. C. Bautista (Philippine Research Institute, University of South Carolina); Alexander
Institute of Volcanology and Seismology); M. L. Bautista Garcia-Aristizabal (Analisi e Monitoraggio del Rischio
(Philippine Institute of Volcanology and Seismology); Mendy Ambientale); Paolo Gasparini (Analisi e Monitoraggio
Bengoubou-Valerius (Bureau de Recherches Géologiques del Rischio Ambientale); , Amir S. J. Gilani (Miyamoto
et Minières); Marc F. P. Bierkens (Department of Physical International); Jonathan Griffin (Geoscience Australia);
Geography, Utrecht University); Joern Birkmann (UNU- Rashmin Gunasekera (World Bank); Muki Haklay (University
EHS); Michael Bonte (World Bank); Arno Bouwman (PBL College London); Carl B. Harbitz (Norwegian Geotechnical
Netherlands Environmental Assessment Agency); Jason Institute); Sven Harig (Alfred Wegener Institute); S.
Brown (Australia-Indonesia Facility for Disaster Reduction); Hidayati (Badan Geologi); Niels B. Holm-Nielsen (World
Philip Bubeck (Adelphi); Nama Raj Budhathoki (World Bank); Nick Horspool (Geoscience Australia); Steven
Hosford (Centre National d’Etudes Spatiales); Chu Ishida
Bank); K. Chapman (Humanitarian OpenStreetMap Team);
(Japan Aerospace Exploration Agency); Oscar Ishizawa
John Crowley (GFDRR); Helen Crowley (GEM Foundation);
(World Bank); M. Jakab (Geoscience Australia); A. T. Jones
P. Cummins (Geoscience Australia); Susan L. Cutter
(Geoscience Australia); Brenden Jongman (Institute for
(Hazards and Vulnerability Research Institute, University
Environmental Studies and Amsterdam Global Change
of South Carolina); P. Dailey (AIR Worldwide); James Daniell
Institute, VU University Amsterdam); Swarna Kazi (World
(Karlsruhe Institute of Technology); Vivien Deparday
Bank); Nicole Keller (GEM Foundation); Anne Kiremidjian
(Stanford University); Kamal Kishore (UNDP); Nadejda
Komendantova (Institute for Environmental Decisions,
ETH Zurich; Risk, Policy and Vulnerability Program,
International Institute for Applied Systems Analysis);
Widjo Kongko (Agency for Assessment and Application
of Technology, Indonesia); Heidi Kreibich (GFZ German
Research Centre for Geosciences); Jolanta Kryspin-Watson
(World Bank); Daisuke Kuribayashi (International Centre
for Water Hazard and Risk Management); Youngjoo Neysa J. Setiadi (UNU-EHS); Iain Shuker (World Bank);
Kwak (International Centre for Water Hazard and Risk Benedikt Signer (World Bank); Robert Soden (World Bank);
Management); David Lallemant (Stanford University); R. U. Solidum Jr. (Philippine Institute of Volcanology and
Hamzah Latief (Bandung Institute of Technology); Juan Seismology); Kate Stillwell (GEM Foundation); Joaquin
Carlos Lam (World Bank); Sangeun Lee (International Toro (World Bank); Dechen Tshering (World Bank); Rens
Centre for Water Hazard and Risk Management); Willem van Beek (Department of Physical Geography, Utrecht
Ligtvoet (PBL Netherlands Environmental Assessment University); K. Van Putten (Geoscience Australia); Charlotte
Agency); Finn Løvholt (Norwegian Geotechnical Institute); Vinchon (Bureau de Recherches Géologiques et Minières);
Olivier Mahul (World Bank); Gusyev Maksym (International Pieter Waalewijn (World Bank); Philip J. Ward (Institute for
Centre for Water Hazard and Risk Management); I. Environmental Studies and Amsterdam Global Change
Meilano (Bandung Institute of Technology); Erwann Institute, VU University Amsterdam); A. Wibowo (Badan
Michel-Kerjan (Wharton Business School, University of Nasional Penanggulangan Bencana); Marc Wieland
Pennsylvania); H. Kit Miyamoto (Miyamoto International); (Helmholtz Centre Potsdam, German Research Centre
Daniel Monfort (Bureau de Recherches Géologiques et for Geosciences [GFZ], Potsdam); Hessel C. Winsemius
Minières); Charlotte Morgan (Geoscience Australia); Roger (Deltares); Steven Wong (Stanford University); H. Martine
Mrzyglocki (German Committee for Disaster Reduction Woolf (Geoscience Australia); Jianping Yan (UNDP); Nario
[DKKV]); J. Murjaya (Indonesian Agency for Meteorology, Yasuda (International Centre for Water Hazard and Risk
Climatology and Geophysics); Zubair Murshed (UNDP); Management); Andrea Zanon (World Bank).
Farrokh Nadim (Norwegian Geotechnical Institute);
For comments and advice that improved this publication,
I. C. Narag (Philippine Institute of Volcanology and
the team is grateful to the following World Bank and
Seismology); Francis Nkoka (World Bank); Ariel Nunez
GFDRR colleagues: Abigail Baca, Jack Campbell, John
(World Bank); Toshio Okazumi (International Centre
Crowley, Vivien Deparday, Marc Forni, Ben Fox, Rashmin
for Water Hazard and Risk Management); A. Orquiza
Gunasekera, Oscar Ishizawa, Daniel Kull, James Newman,
(Department of Foreign Affairs and Trade, Manila);
Fernando Ramirez, Robert Soden, Annegien Tijssen,
Anthony Patt (Institute for Environmental Decisions,
Joaquin Toro, and Jon Walton.
ETH Zurich; Risk, Policy and Vulnerability Program,
International Institute for Applied Systems Analysis); Ivan Special thanks go to Kate Stillwell (Global Earthquake
Petiteville (European Space Agency, Committee on Earth Model), Kamal Kishore (UNDP), Andrew Jones (Geoscience
Observation Satellites Disasters Working Group); Emma Australia), and UNISDR for support and feedback.
Phillips (GFDRR); Massimiliano Pittore (Helmholtz Centre
Finally, the team greatly appreciates the support and
Potsdam, German Research Centre for Geosciences
guidance received from Francis Ghesquiere, Zoubida
[GFZ], Potsdam); Fernando Ramírez-Cortés (World Bank);
Allaoua, Rachel Kyte, James Close, and Ede Jorge
David Robinson (Geoscience Australia); Sahar Safaie
Ijjasz-Vasquez.
(GEM Foundation); Artessa Saldivar-Sali (World Bank);
Hisaya Sawano (International Centre for Water Hazard
and Risk Management); Kerry Sawyer (Committee on
Earth Observation Satellites); Charles Scwarthon (Kyoto
University, emeritus); Andreas Schäfer (Karlsruhe Institute
of Technology); Anna Scolobig (Institute for Environmental
Decisions, ETH Zurich; Risk, Policy and Vulnerability
Program, International Institute for Applied Systems
Analysis); Guy Seguin (International Space Consultant);
15
16
ABBREVIATIONS
AAL average annual loss CEOS Committee on Earth Observation
AIFDR Australia-Indonesia Facility for Disaster Satellites
Reduction CEPREDENAC Central American Coordination Center
ASEZA Aqaba Special Economic Zone Authority for Natural Disaster Prevention
ASI Italian Space Agency CMIP Coupled Model Intercomparison Project
AusAID Australian Agency for International CNES National Center for Space Studies
Development (France)
BCR benefit-cost ratio CSA Canadian Space Agency
BMKG Indonesian Agency for Meteorology, CSCAND Collective Strengthening of Community
Climatology and Geophysics Awareness on Natural Disasters
BNPB Badan Nasional Penanggulangan CV coefficient of variation
Bencana (Indonesian National Disaster DEM digital elevation model
Management Agency) DFAT Australian Department of Foreign
BRACE Building the Resilience and Awareness Affairs and Trade
of Metro Manila Communities to Natural DLR German Aerospace Center
Disaster and Climate Change Impacts
DRFI disaster risk financing and insurance
BTOP block-wise TOP
DRM disaster risk management
BUET Bangladesh University of Engineering
ENSO El Niño-Southern Oscillation EO earth
and Technology
observation
CAPRA Central American Probabilistic Risk
EU European Union
Assessment
Europa Re Europa Reinsurance Facility
CIMA Centro Internazionale in Monitoraggio
FEMA Federal Emergency Management
Ambientale
Agency
CIMNE International Centre for Numerical
FEWS-NET Famine Early Warning Systems Network
Methods in Engineering FONDEN Fondo Nacional de Desastres Naturales
GAR Global Assessment Report on Disaster
Risk Reduction
GCM general circulation model
GDP gross domestic product
GED Global Exposure Database
GED4GEM Global Exposure Database for GEM
GEM Global Earthquake Model
GFDRR Global Facility for Disaster Reduction
and Recovery
GHSL Global Human Settlement Layer
GIS geographic information system
GLOFRIS GLObal Flood Risk with IMAGE
Scenarios
GMMA RAP Greater Metro Manila Area Risk PIC Pacific Island country
Assessment Project PML probable maximum loss
GMPE ground motion prediction equation PTHA probabilistic tsunami hazard
GPS Global Positioning Satellite assessment
GPWv3 Gridded Population of the World RCM Radarsat Constellation Mission
GSNL Geohazard Supersites and Natural RHoK Random Hacks of Kindness
Laboratories SAR synthetic aperture radar
GRUPMv1 Global Rural-Urban Mapping Project SEEC CRIF Southeast Europe and Caucasus
GUF Global Urban Footprint Catastrophe Risk Insurance Facility
GUI graphical user interface SOPAC Secretariat of the Pacific Community
HEC Hydrologic Engineering Center Applied Geoscience Technology Division
HFA Hyogo Framework for Action SRTM Shuttle Radar Topography Mission
HOT Humanitarian OpenStreetMap Team TAP Technical Assistance Project
IDCT Inventory Data Capture Tool TCLVs tropical-cyclone-like vortices
IMD India Meteorological Department UR Understanding Risk
InaSAFE Indonesian Scenario Assessment for UNDP United Nations Development
Emergencies Programme
InSAR Interferometric SAR UNEP-GRID United Nations Environment
IPCC Intergovernmental Panel on Climate Programme–Global Resource
Change Information Database
IRM integrated risk management UNISDR United Nations Office for Disaster Risk
ISMEP Istanbul Seismic Risk Mitigation and Reduction
Emergency Preparedness Project USAID U.S. Agency for International
JAXA Japan Aerospace Exploration Agency Development
JICA Japan International Cooperation Agency VGI volunteered geographic information
LGU Local Government Unit WSPA World Society for the Protection of
MASDAP Malawi Spatial Data Portal Animals
MATRIX New Multi-HAzard and MulTi-RIsK
Assessment MethodS for Europe
MnhPRA Morocco natural hazards probabilistic
risk assessment
NASA National Aeronautics and Space
Administration
NGO nongovernmental organization
OSM OpenStreetMap
OpenDRI Open Data for Resilience Initiative
PACCSAP Pacific-Australia Climate Change
Science and Adaptation Planning
PacRIS Pacific Risk Information System
PAGER Prompt Assessment of Global
Earthquakes for Response
PCRAFI Pacific Catastrophe Risk Assessment
and Financing Initiative
17
18
OVERVIEW
T he 10-year-long Hyogo Framework for Action
(HFA) set out to substantially reduce impacts
from natural disasters by 2015. Despite efforts
Across the globe, emerging consensus is highlighting
the central importance of risk information in
disaster risk management (DRM):
toward this goal, economic losses from natural
The foundation for DRM is understanding
disasters are rising—from US$50 billion each year
the hazards, and the exposure and
in the 1980s, to just under $200 billion each year
vulnerability of people and assets to
in the last decade (World Bank and GFDRR 2013).
those hazards. By quantifying the risks
The economic losses sustained by lower- and
and anticipating the potential impacts of
middle-income countries alone over the last 30
hazards, governments, communities, and
years represent a full third of all total development
assistance in the same time period, offsetting individuals can make informed prevention
tremendous efforts by governments, multilateral decisions. Such information can be
organizations, and other actors. used to set priorities for development
and adaptation strategies, sector plans,
As the HFA period ends against a backdrop of programs, projects, and budgets. (World
challenging disaster risk trends, and consultations Bank 2012, 5)
toward a post-2015 framework move forward, it
is important to reflect on the role of disaster risk This report contains case studies spanning 40
assessments in achieving disaster and climate countries that showcase emerging best practices,
resilience, and on the contributions risk assessments demonstrate how risk assessments are being
have made over the last 10 years. Understanding used to inform DRM and broader development,
Risk in an Evolving World: Emerging Best Practices and highlight lessons learned through these
in Natural Disaster Risk Assessment, which was efforts. Taken as a group, these case studies
developed to inform post-HFA discussions and the evidence the need for continued investment in
2015 Global Assessment Report on Disaster Risk accurate and useful risk information and provide
Reduction (GAR),1 reports on the current state of recommendations for the future.
the practice of risk assessment and on advances
Experience has shown that a purely technical
made over the last decade.
assessment of risk, however sophisticated and
cutting-edge, is by itself unlikely to trigger actions
that reduce risk. Successful risk assessments
produce information that is targeted, authoritative,
understandable, and usable. Thus, the first steps
in a risk assessment include understanding why
the assessment is needed and wanted, defining
the information gaps that currently prevent DRM
actions, and identifying the end-users of the
information. These steps can be completed only if
the process of generating and using risk information
is integrated with institutional processes, and between public officials, affected communities,
if there is communication and trust among all and financial providers. Hence the importance of
involved parties: scientists, engineers, decision authoritative information, which can be fit into a
makers, governmental authorities, and community regulated framework backed by the necessary legal
representatives. A risk assessment designed and institutional context.
along these lines will enable the development of
This publication is not a "how-to" guide for
information useful for risk mitigation.
risk assessment, nor does it provide a technical
But it is also important to recognize that articulation of the risk assessment process. Rather,
understanding risk is more than just modelling risk; it provides insight into the potential richness and
it requires an understanding of the development range of risk assessment approaches and their
and social processes that underlie and drive the capacity to meet a variety of purposes and contexts
generation of disaster risk, such as the political within the same overarching framework. For
and social nature of disaster risk information and scientists, engineers, and others producing risk
its use. For example, the decision of an individual information, the publication highlights some of
or government to construct a building that is the challenges in understanding risk—beyond the
resilient to seismic events will be a result of a strictly technical aspects that are described in many
complex interplay between awareness of, belief in, other publications.
and acceptance of the potential risks; the financial
and technical capacity to design and construct the
resilient structure; and the appropriate (enforced)
legal, institutional, and regulatory framework (e.g.,
enforcement of building codes). Similarly, land
scarcity in rapidly developing urban environments
forces often uncomfortable trade-offs between the
urgent needs of today, such as the need to build
on vacant land near employment and educational
opportunities, and the potential risks of tomorrow,
such as a 1-in-20-year flood event.
Moreover, from a public policy perspective, risk
information can be sensitive information, as it
requires—government officials, private sector
companies, community, or individual—to decide
on action (or inaction) to reduce the impacts of
a potential hazardous event. The decision—for
example, to relocate communities away from high
flood risk areas—will come with explicit (e.g.,
financial/resource) costs and implicit (e.g., political
and/or social capital) costs, all of which have to
be weighed within a broader context. The chance
of risk information translating into action, then,
depends to a large extent on sensitive negotiations
19
20
Risk Information as the
1. Risk identification: Understanding,
communicating, and raising awareness
Basis for Decision Making of disaster risk. Managing disaster risk is just
one of myriad challenges faced by governments,
Risk information provides a critical foundation
communities, and individuals, and it is one that
for managing disaster risk across a wide range of
may be easy to neglect. Because the damages and
sectors. In the insurance sector, the quantification
losses caused by historical disasters are often
of disaster risk is essential, given that the solvency not widely known, and because the potential
capital of most non-life insurance companies is damages and losses that could arise from future
strongly influenced by their exposure to natural disasters (including infrequent but high-impact
catastrophe risk. In the construction sector, events) may not be known at all, DRM is given
quantifying the potential risk expected in the a low priority. Appropriate communication of
lifetime of a building, bridge, or critical facility robust risk information at the right time can raise
drives the creation and modification of building awareness and trigger action.
codes. In the land-use and urban planning sectors,
2. Risk reduction: Informing policies,
robust analysis of flood risk likewise drives
investments, and structural and
investment in flood protection and possibly effects
nonstructural measures intended to
changes in insurance as well. At the community
reduce risk. Hazard and risk information may
level, an understanding of hazard events—whether
be used to inform a broad range of activities to
from living memory or oral and written histories—
reduce risk, from improving building codes and
can inform and influence decisions on preparedness,
designing risk reduction measures (such as flood
including life-saving evacuation procedures and the
and storm surge protection), to carrying out
location of important facilities. macro-level assessments of the risks to different
types of buildings (for prioritizing investment in
Building on the DRM framework proposed in the
reconstruction and retrofitting, for example).
Sendai report (World Bank 2012), we highlight
here the role of risk identification in five key 3. Preparedness: Informing early warning
areas of decision making. Each of the case studies systems and emergency measures and
included in this publication deals with the planning, supporting preparedness and contingency
development, and application of risk information for planning at various levels. An understanding
at least one of these areas. of the geographic area affected, along with the
Ha za rd
HAZARD Exposure
EXPOSURE
The Components
The Components Ha
Haza
The zardrd probability, or chance of a
likelihood, Exposure
Exposure
The location, attributes, and values of
The
The
The Components
Components
Components
HaHa
za za
rd
The
TheThe
rd destructive
potentially
likelihood,
The likelihood,
a potentially destructive phenomenon.
potentially
likelihood,
likelihood,
potentially
phenomenon.
probability
probability
destructive
probability
destructive
orchance
,,or chanceof
ofa
, orphenomenon.
probability ,chance of aof a
or chance
phenomenon.
a Exposure
Exposure
assets
The
TheThe
that are
Thelocation,
location,
that are important to communities.
assets that
location,
location,
assets
attributes,
that
important
attributes,
are
areimportant
attributes,
attributes,
to communities.
and
and
important
valuesof
values
andand to
of
values
values
tocommunities.
of of
communities.
for
for Assessing
Assessing Risk
Risk
potentially
potentially destructive
destructive phenomenon.
phenomenon. assets
assets are are
that
that
27° 43' 0" N 85° 19' 0" E
important
important
27°
27°43'
43'0" N85°
0"N
to communities.
to communities.
19'0"
85°19' 0" EE
Risk
for
for Assessing
Assessing Risk
Risk 27°
27° 43' N 0"
43'
0" 85°N19'
85° 19'E0" E
0"
material: cinder block
roof: steel block
material:
material:cinder
cinder block
roof:
roof:steel
steel
material:
material: cinder
cinder blockblock
steelsteel
roof:roof:
Vuln er abil
VULNERABILITY it y Impact
IMPACT Risk
RISK
Vuln
Vuln
The er
er abil
abil
likelihood it
it
that y
yassets will
The likelihood that assets will be damaged be damaged or Impact
Impact
For use in preparedness, an evaluation of Risk
Risk
Is the composite of the impacts of ALL poten-
Vuln Vuln
erlikelihood
destroyed
The
The er abil
abil it
when
likelihoodyit y
thatexposed
that assets to a be
assetswill
will hazard event.
bedamaged
damaged orImpact
or Impact
what
Foruse
For might
use in happen to people
inpreparedness,
preparedness, anevaluation
an evaluation Risk
of Risk
and assets
of ALL
Is
Isthepotential
thecompositeevents
composite of (100’s
ofthe
the or 1,000’s
impacts
impacts of
of
or destroyed when exposed to might
whatfrom a happen
single to people and assets
event.
TheThedestroyed
likelihood
likelihood
destroyed when
thatthat
when exposed
assets
assets to
will60%
exposedwill
beto a
ahazard
be damaged
damaged
hazardevent.
or or For For
event. what
use use in preparedness,
in preparedness,
what might
might happen
happen to
an people
an
people and
evaluation
toevaluation
and assets
of of
assets Is of
Is the ALL
the
ALLmodels).
potential
composite
composite of events
potential of the
the impacts
events (100’s
impacts
(100’s or
or1,000’s
of of
1,000’s
a hazard event. 50%
destroyed
destroyed whenwhen exposed
exposed to
to a60%
hazard
a hazard event. what
event. from
what
might
from a single
might
a happen
happen
singleevent.to people
to people
event. andand assets
assets ALLALL of
ofmodels).
potential
potential events
models). events (100’s
(100’s or 1,000’s
or 1,000’s
50% 60%
50% from
from a single
a single event.
event. of models).
of models).
60% 60%
50% 50%
intensity and frequency of different hazard new risk information is critical. It is important Figure O—1
events, is critical for planning evacuation routes, to recognize that investment in risk information The components for
creating shelters, and running preparedness for insurance or financial purposes is typically assessing risk and the
drills. Providing a measure of the impact of resource-intensive and needs to adhere to difference between
different hazard events—potential number specific standards of analysis. "impact" and "risk."
of damaged buildings, fatalities and injuries,
Source: GFDRR 2014.
secondary hazards—makes it possible to establish 5. Resilient reconstruction: Informing
detailed and realistic plans for better response early and rapid estimates of damage
to disasters, which can ultimately reduce the and providing critical information for
severity of adverse natural events. reconstruction. Risk assessment can play a
critical role in impact modelling before an event
4. Financial protection: Developing financial
strikes (in the days leading up to a cyclone, for
applications to manage and/or transfer
example), or it can provide initial and rapid
risk. Disaster risk analysis was born out of
estimates of human, physical, and economic loss
the financial and insurance sector’s need to
quantify the risk of comparatively rare high- in an event’s immediate aftermath. Moreover, risk
impact natural hazard events. As governments information for resilient reconstruction needs
increasingly seek to manage their sovereign to be available before an event occurs, since
financial risk or support programs that manage after the event there is rarely time to collect the
individual financial risks (e.g., micro-insurance information needed to inform resilient design
or household earthquake insurance), developing and land-use plans.
21
22
It is important to emphasize that exposure and
A Framework for
vulnerability, not just hazard level, drive the scale
Quantifying and and impacts of any disaster (figure O-2). Rapid
and/or unplanned urbanization—characterized by
Understanding Risk dense populations living in poorly constructed
In its most simple form, disaster risk is a function housing—sets the stage for significant losses in
of three components—hazard, exposure, and lives and property when it occurs in areas at risk
vulnerability (figure O-1). of flooding, earthquake, or other hazards. Indeed,
evidence now points to urbanization—the unplanned
• Hazard refers to the likelihood and intensity of and unchecked swelling of cities and megacities—as
a potentially destructive natural phenomenon, among the most important drivers of disaster risk
such as ground shaking induced by an earthquake (GFDRR 2012). Fortunately, a catastrophic disaster
or wind speed associated with a tropical cyclone. is not the inevitable consequence of a hazard event,
and much can be done to reduce the exposure
• Exposure refers to the location, attributes, and
and vulnerability of populations living in areas
value of assets that are important to the various
where natural hazards occur, whether frequently
communities, such as people, buildings, factories,
or infrequently.
farmland, and infrastructure, and that are
exposed to the hazard. The two strongest tropical cyclones ever to strike
India constitute an instructive example of what can
• Vulnerability is the reaction of the assets when
be achieved through understanding and managing
exposed to the spatially variable forces produced
risk. In 1999, the Odisha cyclone made landfall and
by a hazard event. For example, a building’s
resulted in 10,000 fatalities.2 Fourteen years later,
vulnerability to earthquake increases with the
Cyclone Phailin struck nearby and resulted in 45
intensity of ground shaking and decreases with
fatalities.3 This dramatic reduction in loss of life
improved conformity to seismic design standards.
highlights the extensive efforts made by the state of
Similarly, socioeconomic conditions can make
Odisha in disaster management and preparedness.
responding to a hazard event easier or more
A similar example is offered by New Zealand
difficult.
and Japan, where efforts by governments over
Of course, within this simple framework a multitude decades massively reduced potential losses from
of possible approaches to risk assessment and risk the Christchurch and Great East Japan (Tohoku)
modelling is possible. earthquake events in 2011.
1 2 Increased Exposure
and Vulnerability
Exposure Vulnerability Exposure Vulnerability
RISK RISK
Comparing Risk
1 2
Hazard
Hazard
3 4
3 Increased Exposure,
Vulnerability, and Hazard
4 Reduced Exposure
and Vulnerability
Exposure Vulnerability Exposure Vulnerability
RISK RISK
Hazard
Hazard
components—as well as operational and institutional
Advances in Disaster Figure O—2
progress and challenges associated with new Risk as a function of
Risk Assessment and Key modes of addressing risk such as multi-stakeholder hazard, exposure,
collaboration, communication, and open data and vulnerability.
Remaining Challenges
and models.
Note: Triangle 1 shows equal
Though important challenges remain in assessing contributions to the risk
Hazard. A wide range of data is required for
risk, since 2005 significant progress has been made equation. Triangle 2 shows
/// ///
understanding the potential extent and intensity a rapid increase in exposure
on each critical element of the risk assessment
of one or more natural hazards. In the last decade, and vulnerability, leading to
process. More hazard data and models are available;
there has been substantial progress toward creating increased risk (as in rapidly
tools and models for identifying, analyzing, and urbanizing cities). Triangle
and providing open access to many global and
managing risk have grown in number and utility; and 3 shows increased hazard,
national data sets critical to understanding hazard.
risk data and tools are increasingly being made freely exposure, and vulnerability,
Moreover, significant advances have been made leading to increased risk (as
available to users as part of a larger global trend
in generation of so-called synthetic catalogs of in a rapidly growing coastal
toward open data. More generally, and in contrast
hazard events, which are used to ensure that the city where the effects
to 2005, today there is a deeper understanding—on of climate change are
full range of hazard events is captured and the
the part of governments as well as development increasingly felt). Triangle 4
likelihood of different events assigned. Significant shows controlled exposure
institutions such as the World Bank—that risk must
challenges in acquiring and using hazard data and vulnerability (such as
be managed on an ongoing basis,4 and that DRM
remain, however. Consensus is emerging on the through proactive DRM),
requires many partners working cooperatively and leading to lower overall risk.
urgent need, particularly in developing countries
sharing information.
and high-risk coastal areas, for digital elevation data
This section summarizes technical advances at the appropriate level (that is, better than the 90m
and challenges associated with the fundamental resolution that is currently available). Similarly, lack
elements of risk—hazard, exposure, vulnerability, of historical hydrometeorological data in digital
and the modelling that integrates these format poses significant challenges in quantifying
23
24
current and future hydrometeorological risk in and to catalyze community mapping of buildings
low- to middle-income countries. There is also and infrastructure. (For more on the development
evidence of emerging attempts to integrate climate and use of GeoNodes, see box 3-1 and section 3-1.)
change scenarios into risk modelling; however, this Moreover, satellite imagery is increasingly becoming
adds significant additional uncertainty into the available for use in assessing and understanding
modelled results. risk. Meteorological data collected using satellite
imagery, for example, are increasingly being used
Exposure. The growing momentum in efforts
to determine flood and drought risks at global and
/// ///
to develop exposure data has given rise to
national scales. In addition, release of satellite
new approaches to data collection at various
imagery to the crowd is increasingly being used
scales, from global to individual-building level.
to map building footprints, roads, and other
The greater availability of global data sets on
characteristics of the built environment or disaster-
population, building types, satellite imagery,
impacted area—often by mappers thousands of
and so on is providing significant opportunity
kilometers away. However, all these efforts need
to model global exposure at higher and higher
to achieve scale and sustainability to ensure that
resolutions. At national and subnational levels,
exposure data are available to explain the impacts of
data and information from government ministries
disasters and climate change at different scales.
(such as statistics authorities, transportation and
infrastructure departments, and education and Vulnerability. Both structural (i.e., physical)
/// ///
health departments) are increasingly being liberated vulnerability and socioeconomic vulnerability are
and merged in order to understand community, relevant to risk assessment. Concerning structural
city, and national exposure. At city and community vulnerability, local engineers are increasingly
levels, the growing popularity of volunteer dedicating themselves to understanding the
geospatial initiatives (e.g., OpenStreetMap, or vulnerability of their local building stock (which
OSM—see box 1-2) is seen by authorities as a way varies significantly from country to country and
to engage communities, particularly youth, in the within countries) to different natural hazards.
collection of data that will help everyone to plan and Engineers in the Philippines and Indonesia, for
manage disaster risk. The Community Mapping for instance, are now developing vulnerability functions
Resilience program in Indonesia6 is a prime example relevant to their respective national building stocks.
of a government-led volunteer geospatial initiative: However, opportunities continue to be lost in
in a little over a year, more than 160,000 individual the collection of damage and loss data following
buildings were mapped into OSM. disaster events—data and information critical to
understanding future risks. In addition, efforts to
Underpinning these efforts has been the rapid rise of
quantify socioeconomic vulnerability and poverty
the open data movement, which aims to make data
remain limited, and information of this kind is rarely
technically open.7 The Global Facility for Disaster
integrated into risk assessments.
Reduction and Recovery and World Bank launched
the Open Data for Resilience Initiative in 2011 to Risk modelling. The last decade has seen a
/// ///
foster and catalyze the open data movement for revolution in open access hazard and risk modelling
climate and disaster resilience. Under this initiative, software packages. Users from beginner to expert
web-based geospatial platforms (GeoNodes) in can now choose from a range of tools to address a
more than 20 countries have been used to open range of problems. The packages vary in complexity
more than 1,000 geospatial data sets to the public from OpenQuake, which is designed for highly
advanced users, to multi-hazard risk platforms such are evident in recent successful efforts in countries
as CAPRA, to tools that enable nonspecialists to such as Jordan (see section 3-10), the Philippines
interact with data sets produced by both experts and (sections 3-1 and 3-4), Indonesia (section 3-4),
volunteers, such as InaSAFE (described in detail in and Bangladesh (section 3-2 and box 2-13), where
section 3-22). All these advances and innovations
8
agencies responsible for each element of risk
create a need for better standards and transparency, assessment worked together with decision makers
which would enable replicating risk results by other in finance, planning, and emergency management.
actors, reporting on modelling assumptions and Moreover, a number of global collaborative
uncertainty, and so forth. efforts have been formed to bring together
practitioners from public, private, academic, and
Another area of increased research and innovation
has been global and regional risk modelling nongovernmental organizations; an example is
activities, designed to provide insight into global and the Understanding Risk global community of
regional trends in disaster risk. For example, global practice (see box 2-11). What the case studies
flood risk models developed in recent years can make clear in aggregate is that there is no singular
quickly provide estimations of potential losses—in "correct" formula for building multi-institutional
monetary or human terms—from flood events with collaborations around risk assessment; effective
different return periods. With these advances comes approaches are context specific, build on existing
a need for clear communication of the limitations institutional mandates, and center on the specific
of global analysis, in terms of scale, data, and DRM problem being addressed.
assumptions (e.g., global and regional flood models
Risk communication. The delivery of a risk
rarely integrate information on flood protection).
/// ///
assessment is now widely recognized as a first
While the experts developing these models
step. The completion of the risk assessment
clearly understand their limitations, especially at
marks the beginning of a longer process of broadly
subnational levels, those using the information
produced by these models may understand their communicating risk information to all relevant
limitations less well. stakeholders—in a way that is meaningful to them
and fit for their purposes. There is no one right way
It is well recognized that risk is not static and that to communicate risk; instead practitioners need to
it can change very rapidly as a result of evolving draw on a toolbox of approaches, ranging from Excel
hazard, exposure, and vulnerability (recall figure spreadsheets, maps, and simple interactive tools,
O-2). Decision makers therefore need to engage to graphical representation of hazard and risk, to
today on the risk they face tomorrow. Fortunately,
clear action-orientated messages from authoritative
significant new methodologies and data sets
and respected voices explaining what citizens,
are being developed that will increasingly make
communities, and countries can do to reduce risk.
modelling future risks possible.
Much progress has been made in communicating
Multi-institutional collaboration. Risk
/// ///
risk—the Padang Build Back Better campaign
assessment is inherently multi-institutional, described in section 3-21 demonstrates this fact, as
and no single agency can be solely responsible does the growing use of new interactive geospatial
for generating, communicating, and using risk tools such as GeoNode and InaSAFE—but this is an
information. The opportunities for collaboration area that needs substantial additional investment in
and dialogue among multi-institutional stakeholders practical and considered research.
25
CHAPTER
5
limited uptake of this information. Experience
Recommendations for
shows that successful projects often partner risk
Future Risk Assessments. specialists with country counterparts to design,
implement, and communicate the results of
The recommendations we offer here draw on
the risk assessment. Now that citizens have the
submissions to this publication and on discussions
ability to map entire cities, it is also important to
with both users and developers of risk information.
recognize that the data they generate are more
For users of risk information—DRM practitioners,
likely to be used when the authorities are also
government officials, donors, and nongovernmental
engaged in this process.
organizations considering investing in risk
assessment—our key recommendations are meant 3. Cultivate and promote the generation
///
to ensure that such investment promotes more and use of open data. Experience gained in
///
resilient development and communities. For those the last decade strongly speaks to the need to
undertaking risk analyses, we see an opportunity to encourage the creation and use of open data.
promote greater transparency and accountability. The analysis of natural hazards and their risks
We stress, however, that the best outcomes are is a highly resource- and data-intensive process,
likely to be achieved when those investing in risk whereby the return on expended resources
information and those carrying out the risk analysis (time and money) can be maximized if the data
work in concert and share a common understanding are created once and used often, and if they are
of the undertaking. iteratively improved. Current approaches to
developing open exposure data on the location,
1. Clearly define the purpose of the risk
///
type, and value of assets continue to be improved,
assessment before analysis starts. Risk ///
and volunteered geospatial efforts and remote
assessments initiated without first defining a sensing products offer new opportunities to
question and an end-user often become scientific collect and update fundamental data. That said,
and engineering exercises that upon completion despite the progress made, some fundamental
must find a use case. Moreover, a risk assessment data gaps prohibit meaningful and accurate
that is not properly targeted may not be fit for its assessments of disaster and climate risks—for
intended purpose or may be over-engineered and/ example, we lack global digital elevation data sets
or over-resourced. Where risk assessments have available at resolutions appropriate for analyzing
been commissioned in response to a clear and the potential inundation from flood, storm surge,
specific request for information, they have tended sea-level rise, tsunami, and so on.
to be effective in reducing fiscal or physical risk.
4. Make better communication of risk
///
2. Promote and enable ownership of the
///
information an urgent priority. Clear ///
risk assessment process and efforts to communication throughout the risk assessment
mitigate risk. Ownership is critical for ensuring
///
process—from initiation of the assessment to
that knowledge created through a risk assessment delivery of results and the development of plans
is authoritative and therefore acted upon. It is in response—is critical for successfully mitigating
certainly possible for risk specialists to generate disaster risk.
risk analysis without ever engaging with local
authorities; but regardless of the sophistication or A case study featured in section 3-21—"Build Back
accuracy of their analysis, there will likely be very Better: Where Knowledge Is Not Enough"—is a
27
28
must-read for all risk assessment practitioners information has to do with the type of
and disaster risk managers. An exceptionally information communicated, and to whom.
planned and implemented "Build Back Better" Metrics like average annual loss and probable
campaign led by the government of Indonesia maximum loss, for example, are of interest and
in the aftermath of the 2009 Padang earthquake relevant to the financial sector, but they are poor
demonstrated conclusively that well-targeted metrics for communicating with almost all other
education and communication of risk decision makers involved in DRM. Far preferable
information can increase awareness of natural are interactive tools that enable people to answer
hazards and their potential impacts. Analysis also "what if?" questions robustly and simply ("What
showed, however, that progress from increased if an earthquake/cyclone/other natural hazard
awareness to action can be very difficult to hit my community—How many buildings would
achieve, even in a community that has witnessed collapse or be damaged?"). InaSAFE, a recently
at first hand the devastation of an earthquake. developed tool, meets this need and is now being
To put risk knowledge into practice and build used extensively at national and subnational
more resilient homes, people must be offered levels in Indonesia. That said, there is still
the correct combination of timely information, immense opportunity to develop a bigger toolbox
technical training, community supervision, of interactive, highly graphical visualization
and financial and nonfinancial incentives and tools, which would enable all decision makers,
disincentives. from individuals to national governments, to
A second point about communicating risk meaningfully interact with risk information.
5. Foster multidisciplinary, multi-institutional,
///
leads to better land-use planning, better response
and multi-sectoral collaboration at all capacity, greater risk awareness, and increased
levels, from international to community. ///
ability to set priorities for mitigation actions.
To generate a usable risk assessment product, Particular caution should be taken with risks in
technical experts and decision makers must food security and the agricultural sector, and
consult with one another and reach agreement we recommend that these risks be considered
on the risk information that is required by alongside flood and drought analysis.
the relevant development program, and more
7. Keep abreast of evolving risk. Risk
broadly on the purpose and process of the risk
/// ///
assessments need to account for temporal
assessment. The actual development of risk
and spatial changes in hazard, exposure, and
information is clearly a multidisciplinary effort
vulnerability, particularly in rapidly urbanizing
that takes place through collaborations ranging
areas or where climate change impacts will be
from international efforts to multi-institutional
felt the most. A risk assessment that provides
arrangements at national and subnational levels.
an estimation of evolving or future risk is a way
There are many efforts currently under way that
to engage stakeholders in carrying out actions
speak to the success of this approach. However,
now in order to avoid or mitigate the risk that
success has been comparatively limited in
is accumulating in their city or country. For
merging community-level understanding of risk
example, risk analysis offers an opportunity to
with a national or subnational understanding
quantify the decrease in future risk that arises
of risk. This is a missed opportunity wherein
from better enforcement of building codes, and
a common understanding of the risks and
hence to demonstrate the benefit of spending
necessary steps to reduce these risks could
additional funds on building code enforcement.
trigger greater action.
6. Consider the broader risk context. Rarely do
/// ///
Because risk is likely to evolve under climate
countries, communities, or citizens face potential change—according to the Intergovernmental
risks from only one hazard, or even from natural Panel on Climate Change, "a changing climate
hazards alone. Our complex environments leads to changes in the frequency, intensity,
and social structures are such that multiple or spatial extent, duration, and timing of extreme
connected risks—from financial hazards, multiple weather and climate events" (IPCC 2012, 7)—
or cascading natural hazards, and anthropogenic there is increasing interest in understanding
hazards—are the norm. Just as multi-peril risk climate change’s impacts and calculating losses
calculations are required for many financial under future adverse climate events. Using the
applications, territorial planning should draw modelling techniques and approaches developed
on information from assessments of multiple to model disaster risk, experts have demonstrated
hazards (flood, landslide, and earthquake, for the potential to determine future loss under
example) in order to reduce risk. We know that climate change. However, since the fundamental
failure to consider the full hazard environment data sets that enable the risks of today to be
can result in maladaptation (heavy concrete quantified are the same as those required to
structures with a ground-level soft story for determine the impacts of adverse events in
parking can protect against cyclone wind, for the future, it is critical for both the disaster
example, but can be deadly in an earthquake), and climate change communities to continue
whereas adopting a multi-hazard risk approach investing in fundamental data and innovation.
29
30
8. Understand, quantify, and communicate
/// 10. Encourage innovations in open source
///
the uncertainties and limitations of risk software. In the last 5 to 10 years, immense
///
information. Once risk information is produced,
///
progress has been made in creating new open
all users must be aware of and knowledgeable source hazard and risk modelling software. More
about its limitations and uncertainties, which can than 80 freely available software packages, many
arise from uncertainties in the exposure data, of which are open source, are now available
in knowledge of the hazard, and in knowledge for flood, tsunami, cyclone (wind and surge),
of fragility and vulnerability functions. Failure and earthquake, with at least 30 of these in
to consider these can lead to flawed decision widespread use. Significant progress has also
making and inadvertently increase risk. A risk been made in improving open source geospatial
model can produce a very precise result—it may tools, such as QGIS and GeoNode, which are
lowering the financial barriers to understanding
show, for example, that a 1-in-100-year flood will
risks at national and subnational levels. Yet all
affect 388,123 people—but in reality the accuracy
this innovation has created challenges around
of the model and input data may provide only an
assessing "fitness-for-purpose," interoperability,
order of magnitude estimate. Similarly, sharply
transparency, and standards. These need to be
delineated flood zones on a hazard map do not
addressed in a way that continues to catalyze
adequately reflect the uncertainty associated
innovation and yet also better supports risk
with the estimate and could lead to decisions
model users.
such as locating critical facilities just outside the
"flood line," where the actual risk is the same as
if the facility was located inside the flood zone.
Recommendations
It is incumbent upon specialists producing risk
information to clearly and simply communicate toward the Next Hyogo
uncertainties and limitations.
Framework for Action
9. Ensure that risk information is credible
Looking ahead to the next phase of the HFA, we
///
and transparent. Risk information must be
would encourage international policy makers to
///
scientifically and technically rigorous, open for
consider the above recommendations, which are
review, and honest regarding its limitations and
based on the case studies and analytical work this
uncertainties, which may arise from uncertainties
publication reports on. Future HFA indicators
in the exposure data, in knowledge of the hazard,
centered on risk information should articulate the
and in knowledge of fragility and vulnerability need for targeted, robust, authoritative, trusted,
functions. The best way to demonstrate open, understandable, and usable risk information—
credibility is to have transparent data, models, descriptors which were universally mentioned
and results open for review by independent, by contributors to this publication. Future HFA
technically competent individuals. Risk modelling indicators should also stress the importance
has become very advanced, yet also more of producing risk information that is driven by
accessible, and therefore anyone can feasibly the needs of end-users and the information and
run a risk model—but without the appropriate evidence gaps—whether at national, subnational
scientific and engineering training and judgment, or community levels—as well as the need for
the results may be fundamentally incorrect and appropriate communication of risk information for
may mislead decision makers. different stakeholders.
Endnotes
1 The Global Assessment Report, whose preparation is
more to be done to systematically integrate an assessment
overseen by the United Nations Office for Disaster Risk of risks into the design and implementation of World Bank-
Reduction, is released every two years. Like previous financed projects (9).”
reports, the 2015 edition addresses progress and
5 Liberated data are those that were at one time
challenges in achieving each of the Hyogo Framework
inaccessible due to format, policies, systems, etc., but are
for Action objectives. The Global Facility for Disaster
now being made available for use, either as discoverable
Reduction and Recovery led the development of the
and usable data sets or (in many cases) as technically open
analysis on “Priority Action 2: Identify, assess and monitor
data sets.
disaster risks.”
6 This program began in 2011 through a partnership led
2 The 1999 Odisha cyclone, Cyclone 05B, was the first
by the Australia-Indonesia Facility for Disaster Reduction,
storm to be categorized by the India Meteorological
Indonesia’s National Disaster Management Agency (Badan
Department (IMD) as a super cyclonic storm. The 10-minute
Nasional Penanggulangan Bencana), and the Humanitarian
sustained wind was derived using a factor of ~0.85 to
OpenStreetMap Team, with support from the GFDRR and
convert from 1-minute to 10-minute sustained winds. the World Bank.
3 According to IMD (2013), Cyclone Phailin’s winds at
7 Technically open generally means that data can be found
landfall were ~215km/hr. IMD uses 3-minute sustained on the Internet at a permanent address and are available
winds as an average. A factor of ~0.9 was used to convert in structured, nonproprietary formats via download or an
from 3-minute to 10-minute sustained winds. application programming interface (API).
4 According to GFDRR (2012), a recent report by the World
8 OpenQuake was developed under the Global Earthquake
Bank’s Independent Evaluation Group finds “a clear shift Model Foundation; for more information see http://www.
toward risk reduction in Bank-supported investment globalquakemodel.org/. For more information about
projects since 2006,” though it also notes that “there is CAPRA, see the program’s website at www.ecapra.org.
References
GFDRR (Global Facility for Disaster Reduction and UNDP (United Nations Development Programme). 2013.
Recovery). 2012. “Managing Disaster Risks for a Resilient
“A Comparative Review of Country-level and Regional
Future: A Strategy for the Global Facility for Disaster
Disaster Loss and Damage Databases.Bureau for Crisis
Reduction and Recovery 2013—2015.” https://www.gfdrr.
Prevention and Recovery, UNDP. http://www.undp.org/
org/sites/gfdrr.org/files/publication/GFDRR_Strategy_
content/undp/en/home/librarypage/crisis-prevention-
Endorsed_2012.pdf.
and-recovery/loss-and-damage-database/.
———. 2014. Open Data for Resilience Initiative Field Guide.
Washington, DC: World Bank. World Bank. 2012. The Sendai Report: Managing Disaster
Risks for a Resilient Future. Washington, DC: World Bank.
IMD (India Meteorological Department). 2013. Very Severe https://www.gfdrr.org/sites/gfdrr.org/files/publication/
Cyclonic Storm, PHAILIN over the Bay of Bengal (08-14 Sendai_Report_051012_0.pdf.
October 2013): A Report. New Dehli: Cyclone Warning
Division, India Meteorological Department. http://www. World Bank and GFDRR (Global Facility for Disaster
imd.gov.in/section/nhac/dynamic/phailin.pdf. Reduction and Recovery). 2013. Building Resilience:
Integrating Climate and Disaster Risk into Development—
IPCC (Intergovernmental Panel on Climate Change). 2012.
Managing the Risks of Extreme Events and Disasters to The World Bank Group Experience. Washington, DC:
Advance Climate Change Adaptation. A special report World Bank. http://www.worldbank.org/content/dam/
of Working Groups I and II of the Intergovernmental Worldbank/document/SDN/Full_Report_Building_
Panel on Climate Change. Cambridge and New York: Resilience_Integrating_Climate_Disaster_Risk_
Cambridge University Press. Development.pdf.
31
32
CHAPTER
01
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
INTRODUCTION
Earthquakes, droughts, floods and storms Experience has shown that a disaster risk
are natural hazards, but unnatural disasters assessment does not represent the conclusion
are deaths and damages that result from of a process, but instead provides a foundation
human acts of omission and commission. for a long-term engagement focused on the
Every disaster is unique, but each exposes communication and use of the risk information.
actions—by individuals and governments at Proactive responses to new risk information
different levels—that, had they been different, include retrofitting buildings to withstand the
would have resulted in fewer deaths and less assessed seismic risk, developing new land-use
damage. plans, designing financial protection measures, and
equipping and training emergency responders.
—World Bank and United Nations, Natural
Hazards, UnNatural Disasters In the context of rapidly growing disaster losses
and high-profile catastrophic disasters, it is often
A disaster-related risk assessment provides an difficult to imagine reducing the impact from
opportunity before a disaster event to determine hazard events. However, societies have successfully
the likely deaths, damages, and losses (direct overcome similar challenges in the past. For
and indirect) that will result, and to highlight centuries, urban fires were a global concern for
which actions will be most effective in reducing the public, private, and finance sectors, as well as
the impacts on individuals, communities, and for the communities directly affected. Urban fires
governments. This ability to model disaster loss and devastated Rome in 64 CE, London in 1666, Moscow
to provide robust analysis on the costs and benefits in 1812, Chicago in 1871, and Boston in 1872; the
of risk preparedness, reduction, and avoidance has 1906 San Francisco fire destroyed nearly 95 percent
made disaster risk assessments a powerful tool in of the city, and the Tokyo fire of 1923 killed over
disaster risk management (DRM). As a result, the 40,000 people. Yet we do not see urban fires any
number of risk assessments being undertaken is more, and this hazard has largely been consigned to
growing, innovation has flourished, and a vast array history. The reasons— implementation of modern
of approaches, experiences, and lessons learned building codes, land-use planning, establishment
now exists. and expansion of emergency services, greater citizen
33
34
INTRODUCTION
Box 01—1 How Risk Information Contributes to Mainstreaming
responsibility, and insurance regulations—are
of DRM in World Bank Group Operations
essentially the same levers that we can apply to
Recognizing that the risks from adverse natural events challenge its efforts consigning natural disaster events to history.
to end extreme poverty and promote shared prosperity, the World Bank
Group now has disaster and climate risk management at the core of its We have already seen construction practices
strategy. Moreover, under the IDA17 program of the International Development evolving in response to cyclones and earthquakes,
Association (the World Bank’s fund for the poorest countries), the World Bank and some areas have strict urban and land-use plans
Group has committed to incorporating climate and disaster risk considerations
designed to reduce loss from flood. California,
in all new country partnership frameworks and will screen all International
for example, has implemented a series of building
Development Association operations for climate and disaster risks. To carry
out this strategy of mainstreaming disaster and climate risk management into code changes in response to earthquake9—changes
World Bank Group operations, an even greater investment and focus on risk that today represent a reduction in risk. Recent
identification will be required. earthquakes in Chile, New Zealand, and Japan
have dramatically demonstrated the influence of
The World Bank’s investment in DRM is steadily rising. It grew from US$2
billion in fiscal year 2010 to US$3.8 billion in fiscal year 2013, with the most enforced building codes in reducing death, damage,
substantial growth in Africa. The large share of this investment—83 percent— and loss. These examples show that a society can
supports ex ante DRM activities. The role of advisory and analytical services reduce vulnerability and risk. But for these efforts
to support better information on natural hazard risk is also growing; in the to succeed, there must be robust and accessible
last three years, 43 countries have been supported in efforts to improve their
information on hazard, exposure, and vulnerability,
information about hazard exposure. To cite one example: the Global Facility
models that integrate this information and quantify
for Disaster Reduction and Recovery (GFDRR) supported analytical work on
seismic and flood risk in Manila, which has led the Philippine government to risk, and the commitment and resources to prioritize
endorse a US$9 billion flood reduction plan. This report offers further examples actions needed to implement risk reduction.
of efforts by the World Bank and GFDRR to implement risk assessment as the
first step toward reducing risk through DRM. The World Bank's approach to investment in DRM
through better risk information is summarized in
Source: Development Committee 2014.
box 1-1.
About This Publication
This publication was developed to help identify
the progress made in risk assessment under the
10-year Hyogo Framework for Action and to capture
through use-case analysis the diverse efforts made
to improve our awareness and understanding of
risk. It is not a technical guide on how to undertake
a risk assessment and instead offers a narrative
to a nontechnical audience interested in how risk
information can lead to more resilient communities,
cities, and countries. The authors are aware that this
publication does not capture all the engagements
and projects on risk assessment across the globe or
CHAPTER
01
all the innovations and advances that have taken
A Brief History of
place. However, it does provide both a snapshot
of use cases for those interested in application of Risk Assessment
risk assessment and some recommendations for
Societies have been dealing with risk for thousands
the future.
of years. The earliest records related to practices
The report begins with an overview and is then intended to minimize financial risk come from
divided in four parts. shipping. For example, in the second millennium
BCE the Babylonians invented maritime loans
Overview. This section summarizes the report's
that did not require repayment if the ship was lost
/// ///
key themes, observations, and recommendations
(Carter 1979). The origins of modern property
in order to prompt policy dialogue and discussions
insurance practices that are not associated with
among funders of risk assessment projects.
maritime ventures can be traced back nearly
1. Introduction. This section describes the history
/// ///
350 years, to the creation of the first fire mutual
of risk assessment, the recent rise of open data companies following the London fire of 1666.
and open risk modelling, and the alignment of Benjamin Franklin started the first U.S. mutual fire
risk assessments to different DRM applications. insurance company in 1792. The devastating fires
in U.S. cities during the 19th century bankrupted
2. Progress, Achievements, and Remaining
///
many insurance companies and fostered the use of
Challenges in Risk Assessment. Based ///
objective assessments of risk using fire insurance
on research and on submissions from and maps, which displayed building footprints,
discussions with experts, this section captures construction materials, and location information.
key achievements and progress in different
The modern approach to risk assessment—using
aspects of risk assessment in the last decade—
complex models as well as extensive exposure and
from availability of fundamental data sets, to
hazard data—came into being when computational
modelling tools, to new platforms that facilitate
resources became more powerful and more
collaboration. This section also articulates
common. But even before the advent of computers,
remaining challenges that need focus over
insurers seeking to track exposure and avoid
coming years.
unwanted concentrations of risk used pins on a map
3. Case Studies Highlighting Emerging to mark the location of underwritten properties.
Best Practices. This section showcases risk Thus tracking risk using data on exposure and
assessment initiatives from around the world, vulnerability is not a new practice.
grouped according to their focus on one of the
The invention of computers and their adoption by
following: data; modelling; risk assessment in
government and industry set the stage for coupling
practice; institutionalization and communication
exposure and vulnerability data with hazard models
of risk information; assessment of future risk.
to generate risk estimates. Perhaps the first modern
4. Recommendations. Based on
/// ///
risk models were developed for managing flood
recommendations received from developers and risk and designing dams. The U.S. Army Corps of
users of risk information and on emerging best Engineers Hydrologic Engineering Center (HEC)
practices, this section offers 10 recommendations was created in 1964 and released components of the
for future investment in risk assessment. first watershed models in 1966. The components
INTRODUCTION
35
36
INTRODUCTION
needed to be run separately because of memory closer to those actually experienced by the insurance
limitations in computers. The integrated version industry. The difference between experience-based
of the model, HEC-1 Flood Hydrograph Package, and model-derived loss estimates was driven in
was released in 1968. At that time, releasing the part by dramatic increases in exposure along the
integrated model components as a package was coast and by the limited sample of hurricane events
considered a major innovation that allowed linked, in the historical record. Today, many insurers and
related programs to be run without direct handling reinsurers have in-house capacity to undertake their
of intermediate results (HEC 1989). own probabilistic catastrophic modelling.
Other risk assessment-related efforts were also Emergency management agencies also began to
taking place during the late 1960s and early 1970s. adopt risk models for risk assessment in the 1990s.
During this period, for example, C. Allin Cornell In 1997 the Federal Emergency Management
(1968) published the seminal methodology for Agency (FEMA) released Hazus97, the first version
seismic risk assessment; efforts at assessing of Hazards US (Hazus), a geographic information
hurricane risk for NASA's Apollo project were under system (GIS)-based natural hazard loss estimation
way (Jarvinen, Neumann, and Davis 1984); and software package. The output from Hazus includes
the catastrophe risk models for a range of natural factors such as shelter needs related to emergency
hazards were under development for use by insurers management. The Hazus model has been adopted
(Friedman 1972). for use by emergency management organizations
outside the United States, in countries such as
Risk modelling became more common as Singapore, Canada, Australia, and Pakistan.
computational resources expanded. In 1981 the
first catastrophe risk modelling company, EQE During the first decade of the 21st century, there
was growing awareness that risk assessments could
International, was founded. The company provided
help countries develop tools and strategies to reduce
catastrophic risk management consulting, design,
disaster losses, and thus several efforts to develop
and research services to commercial, utility, nuclear,
risk models were initiated. Governments have
and other high-tech industries. The two other
increasingly started to use risk modelling to assess
major catastrophe risk modelling firms, Applied
their exposure to natural events, and in particular to
Insurance Research (AIR) and Risk Management
use probabilistic risk modelling techniques, which
Solutions (RMS), were formed in 1987 and 1989,
manage uncertainty by providing a robust measure
respectively. While catastrophe risk models
of risk and which allow for comparisons of risk.
provided objective assessment of risk, until the early
1990s much of the insurance industry still based In 2004 New Zealand began to develop RiskScape,
many business decisions on actuarial approaches a regional multi-hazard risk model; Australia
using historical data. The use of catastrophe risk similarly began development of seismic, cyclone,
models in the insurance industry grew dramatically and tsunami risk models; and in 2007 a partnership
after Hurricane Andrew struck Florida in 1992 and of Central American governments and development
insured losses turned out to be much greater than institutions began work on CAPRA (Central
those expected based on historical experience. Using American Probabilistic Risk Assessment). Many of
its hurricane risk model rather than an actuarial these models were developed to be open source
approach, AIR estimated insured losses that were and have led to large developer communities. In
much larger than any experienced in the past and addition to these initially regional efforts, the
decade also saw efforts to develop global models. a new global probabilistic model in 2011 (described
The Global Earthquake Model (GEM), for example, in more detail in section 3-6).
was conceived in 2006; the GEM Foundation was
Today, there are more than 100 freely available risk
officially formed in March of 2009; and the first models across the range of hazards. While many
official release of the GEM OpenQuake platform of these remain the domain of the experienced
is slated for 2014 (for more on GEM, see section scientist or engineer, and are poorly suited to city
3-6). The international development community or government officials responsible for managing
also joined this effort, beginning in 2005 with disaster risk, a growing number of more user-
collaboration under the ProVention Consortium friendly models are becoming available, such as the
by the World Bank and Columbia University, along InaSAFE tool developed through a collaboration
between the Indonesian and Australian governments
with a number of additional contributors, including
and GFDRR—World Bank (see section 3-22 for more
the Norwegian Geotechnical Institute (Dilley et al.
detail). Researchers are also beginning to couple
2005). This collaboration in turn spurred related
probabilistic risk models with predictions of climate
efforts, such as the Global Risk Identification change to account for future changes in hazard and
Programme (GRIP) of the United Nations risk (see for example sections 3-23 and 3-24). This
Development Programme, followed by the United approach is likely to become the norm in future
Nations Office for Disaster Risk Reduction's work on assessments.
INTRODUCTION
37
38
INTRODUCTION
to be carried out by resource-poor governments
The Rise of Open
and communities.
Models and Data:
• There is a global movement toward open data,
The Changing Risk which seeks to increase government transparency
and accountability and to broaden participation
Assessment Paradigm in governance. This effort can be seen in the
Over the last five years, the field of risk assessment establishment of initiatives such as the Open
has been increasingly driven by open data and open Government Partnership, whose 63 member
source modelling. The reasons for this evolution governments have pledged accountability to their
are multifold: citizens. In addition, development institutions
such as the World Bank, the U.S. Agency for
• Producing risk information requires a substantial
International Development (USAID), and the
investment in time, money, and effort, and
African Development Bank view openness as a
those commissioning it are no longer satisfied
means to make the development process more
with a published report as the sole end result.
inclusive and transparent.
The real value is increasingly seen in the data
that make the risk analysis possible, and in the • Open data and open models promote a level of
various hazard and risk maps and analysis that transparency in risk assessment that represents
can be further manipulated and used in a variety an appealing change from the past, when
of contexts. assumptions, data sets, and methodologies, along
with the associated uncertainties, were invisible
• The rapid changes in urban environments, in
to the end-user.
populations, and in extreme weather events
require that risk information be dynamic and • Driven originally by citizens frustrated by
updated frequently. Access to open data and lack of access to fundamental maps in the
modelling tools allows dynamic risk assessment United Kingdom, there is a surge in interest
Box 01—2 OpenStreetMap
OpenStreetMap, often called “the Wikipedia of maps,” is an online geospatial The database hosts data on transport networks, buildings, amenities,
database and a global community of over 1.5 million contributors, who are and natural landscapes across the globe. Data collection ranges from
engaged in building a free and open map of the world that anyone can
local-level surveys with handheld GPS units and paper maps to tracing
contribute to and that can be used in any tool or analysis.(A) OSM was
satellite imagery.
established in 2004 in the United Kingdom in reaction to restrictions around
the use and/or availability of geospatial data across the world.
The repeated discussion of OSM throughout the case studies in this
OSM is a confederation of organizations and technologies. OpenSteetMap. publication attests to the value of this innovative approach and its ability to
org is a database with over 2.2 billion map “nodes” hosted by University improve our understanding of risk from natural hazards and climate change.
College London, Imperial College London, Bytemark Hosting, and other
partners. The OpenStreetMap Foundation is a UK charitable organization (A) OSM is open data, licensed under the Open Data Commons Open
that oversees the state of the map. The Humanitarian OpenStreetMap Team Database License (ODbL); see http://www.openstreetmap.org/copyright for
(HOT) is a U.S. nonprofit corporation that applies the “principles of open more information on copyright and license.
source and open data sharing for humanitarian response and economic
development.”(B) HOT provides support to emergency operations and (B) See the HOT website at http://hot.openstreetmap.org/.
training for the collection of mapping data in communities at risk.
Box 01—3 Community Mapping in Indonesia
Open data initiatives, combined with bottom-up approaches such The initiative’s main goal is to use OpenStreetMap to collect building-
as citizen mapping initiatives, can be an effective way to build large level exposure data for risk assessment applications. OpenStreetMap
exposure databases. offers several important features: open source tools for online or offline
mapping, a platform for uploading and hosting data with free and open
The Community Mapping for Resilience program in Indonesia is access, and an active global community of users.
an example of a large-scale exposure data collection system. The
In a little over a year, more than 160,000 individual buildings were
program began in 2011 through a partnership led by the Australia-
mapped and new partners—including five of Indonesia’s largest
Indonesia Facility for Disaster Reduction, Indonesia’s National Disaster
universities, local government agencies, international development
Management Agency (Badan Nasional Penanggulangan Bencana), and partners such as Deutsche Gesellschaft für Internationale
the Humanitarian OpenStreetMap Team (HOT), with support from the Zusammenarbeit GmbH (GIZ), and civil society organizations—were
Global Facility for Disaster Reduction and Recovery and the World Bank. trained and are using the platform.
in community or participatory mapping that cost-effective solution to an otherwise expensive
has now become a global revolution led by the challenge of data collection. An example of this
OpenStreetMap community (box 1-2). approach is highlighted in box 1-3.
In addition, as demand grows for risk information To be considered open, models and data should be
at resolutions appropriate for community and city both legally and technically open (see figure 1-1). As
decision making, the need to collect exposure data development and use of open tools grows, the need
at these resolutions has also grown. Crowdsourcing to clarify and standardize the meaning of "open" will
is increasingly being viewed by governments and become more pressing. Box 1-4 describes how one
communities as a solution that enables bottom-up initiative, the Global Earthquake Model, resolved
participation in the understanding of risk and a differences of opinion about "open."
Data is Open If Legally Open Technically Open Figure 01—1
What makes data
“anyone is free to use, It is important to place The data needs to be made “open.”
resuse, and redistribute it a license on open data. available, in bulk, in
Source: GFDRR 2014.
subject only, at most, a machine-readable format.
The World Bank’s own data
to requirement to attribute Note: The quoted material in
policy is licensed under:
and/or share-alike.” category category category the first box is from http://
value value value opendefinition.org/.
ODC-BY value value value
Open Data Commons value value value
Attribution License
value value value
INTRODUCTION
39
Aligning and Targeting
Box 01—4 Defining “Open”
Risk Assessments The members of the Global Earthquake Model, a public-private partnership,
share an interest in credible, accessible risk information that is widely used
and understood. Although the principle of “open” data was central to GEM’s
Risk assessment as applied to DRM can easily
mission and self-understanding, over the course of GEM’s first six years
be framed around the formula risk = hazard X
members differed widely on what “open” meant and implied.
exposure X vulnerability.10 Under this single
formula, however, there is considerable variation These differences became obvious and somewhat contentious when concrete
licensing policies were proposed for the data and software developed under
in the types of and purposes for risk assessment. In
GEM: public sector participants typically viewed “open” to imply “free of
the DRM community, risk assessments are generally
charge,” while private sector participants, who sought an ongoing business
undertaken for one (or more) of five reasons: advantage from their sponsorship of GEM, did not want GEM data and
software to be made available free of charge to their competitors. In their view,
1. Risk identification. Understanding,
“open” did not necessarily entail “free.”
communicating, and raising awareness of
disaster risk GEM’s governing board convened a task group to study this issue further
and make a recommendation to the board. The task group, made up of
2. Risk reduction. Informing policies, investments, seven members representing both the public and private sector, proposed a
and structural and nonstructural measures compromise: data and model licenses would be embargoed for 18 months.
Under this arrangement, GEM initially releases any given version of a GEM data
intended to reduce risk
set or model with a license restricting commercial use for 18 months A;
3. Preparedness. Informing early warning after this period the same product is rereleased under a license without
commercial restriction.
systems and emergency measures and supporting
preparedness and contingency planning at (A) The license type is CC BY-NC-SA 3.0 (Creative Commons Attribution–
various levels Noncommercial-ShareAlike 3.0 Unported). See http://creativecommons.org/
licenses/by-nc-sa/3.0/.
4. Financial protection. Developing financial
Source: Helen Crowley, Nicole Keller, Sahar Safaie, and Kate Stillwell
applications to manage and/or transfer risk.
(GEM Foundation).
5. Resilient reconstruction. Informing early and
rapid estimates of damage and providing critical
information for reconstruction
Determining what constitutes a suitable risk
assessment product depends not only on the
purpose of the assessment, but on a number
of other factors as well: which decision makers
and stakeholders are involved, how the results
will be used, the scale and resolution at which
the assessment will be carried out, the data
requirements for the assessment, the complexity
of the analysis, and the resources available. Table
1-1 lists a range of assessment products for various
purposes, each with different attributes.
INTRODUCTION
41
42
INTRODUCTION
Experience has shown that when a risk assessment based assessment involves local stakeholders—
is well targeted to a purpose and end-user, it has a communities and local government—and can
greater chance of success—that is, the information be used in building community preparedness,
it generates is more likely to be used for decision supporting contingency planning, and identifying
making. It is therefore critical that there be vulnerable assets. On the other hand, it cannot be
consensus on a risk assessment's objective, that it be used in developing financial applications and will
designed to meet the project's basic requirements seldom be used in planning significant investments
and standards, and that it not exceed available in risk reduction, or in carrying out land-use
resources (money, personnel, time).
planning. In contrast, a catastrophic risk assessment
To understand how various factors influence risk for financial planning involves a different set of
assessment design, consider two different risk stakeholders—ministries of finance, international
assessment products, one a community-based and domestic financial markets, modelling
assessment that aims to engage communities in companies, and insurance and reinsurance
disaster risk reduction, to communicate risk, and companies—and is carried out on a larger (national
to promote local action (second row of table 1-1), to multi-country) scale using high-quality, high-
and the other a catastrophic risk assessment for resolution data. This type of analysis is rarely used
financial planning (bottom row). The community- for local DRM or community preparedness.11
Table 01—1
Product Purpose Scale Data Requirements Cost
Sample Risk
Assessment Products Qualitative national risk For advocacy and initiation of
Low: Requires global,
National regional, and/or national $
and Their Attributes profile DRM dialogue
data sets
Source: World Bank and
To engage communities,
GFDRR 2013. Community-based disaster
communicate risk, and Community level
Low: Typically based on
$
risk assessment historical disaster events
promote local action
Note: $ = <$100,000;
$$ = 100,000 to $500,000;
For advocacy and initiation of Low-moderate: Requires
Quantitative national risk
$$$ = >$500,000 DRM dialogue based on National global, regional, and/or $$
profile
quantitative assessment national data sets
To inform design of building-
Moderate-high: Requires
Asset-level risk assessments, level/asset-level risk
high-resolution local data for
including cost-benefit and reduction activities and Building / infrastructure level $$
large spatial areas with clear
engineering analysis promote avoidance of new
articulation
risk
Macro-level risk assessment Moderate-high: Requires
To inform urban/regional risk
for risk reduction, including Urban, regional, national moderate to high resolution $$$
reduction measures
cost-benefit analysis across large spatial areas
To inform preparedness and
Risk identification to identify
risk reduction, based on Moderate-high: Requires $$-$$$ (broad range
critical infrastructure and
understanding of potential Urban, regional, national asset-level information depending on geographic
establish early warning
damage at the regional/local across large spatial areas scope)
systems
level
For financial and fiscal
High: Requires high-
Catastrophic risk assessment assessment of disasters and
National to multi-country resolution, high-quality data $$$
for financial planning to catalyze catastrophe risk
of uncertainty
insurance market growth
CHAPTER
01
Endnotes
9 See State of California Seismic Safety Commission
11 However, data in this type of assessment can sometimes
(2000). serve as the foundation for local applications, as was
the experience with the Pacific Catastrophic Risk and
10 Alternately, risk can be expressed as a function: risk =
Financing Initiative (see section 3-9).
f(hazard, exposure, vulnerability).
References
Carter, R. L. 1979. Reinsurance. Brentford, Middlesex, HEC (Hydrologic Engineering Center). 1989.
UK: Kluwer Publishing Ltd. and Mercantile & General “Hydrologic Engineering Center—A Quarter Century,
Reinsurance Co. Ltd. 1964—1989.” http://www.hec.usace.army.mil/
publications/AdministrativeDocument/HEC_A_
Cornell, C. A. 1968. “Engineering Seismic Risk Analysis.” QuarterCentury_1964-1989.pdf.
Bulletin of the Seismological Society of America 58 (5):
1583—1606. Jarvinen, B. R., C. J. Neumann, and M. A. S. Davis. 1984. “A
Tropical Cyclone Data Tape for the North Atlantic Basin,
Development Committee (Joint Ministerial Committee of 1886—1983: Contents, Limitations, and Uses.” NOAA
the Boards of Governors of the Bank and the Fund on Technical Memorandum. NWS NHC 22. http://www.nhc.
the Transfer of Real Resources to Developing Countries). noaa.gov/pdf/NWS-NHC-1988-22.pdf.
2014. “Progress Report on Mainstreaming Disaster
Risk Management in World Bank Group Operations.” State of California Seismic Safety Commision. 2000. “The
March 25. History of the California Seismic Safety Commission:
Living Where the Earth Shakes, 1975—2000.” http://
Dilley, M., R. S. Chen, U. Deichmann, A. Lerner-Lam, M. www.seismic.ca.gov/pub/CSSC_HISTORY.pdf.
Arnold, J. Agwe, P. Buys, O. Kjekstad, B. Lyon, and G.
Yetman. 2005. Natural Disaster Hotspots: A Global World Bank and GFDRR (Global Facility for Disaster
Risk Analysis. Washington, DC: International Bank for Reduction and Recovery). 2013. Building Resilience:
Reconstruction and Development/World Bank and Integrating Climate and Disaster Risk into Development—
Columbia University. The World Bank Group Experience. Washington, DC:
World Bank. http://www.worldbank.org/content/dam/
Friedman, D. G. 1972. “Insurance and the Natural Hazards.” Worldbank/document/SDN/Full_Report_Building_
ASTIN Bulletin International Actuarial Association 7 (1): Resilience_Integrating_Climate_Disaster_Risk_
4—58. Development.pdf.
GFDRR (Global Facility for Disaster Reduction and World Bank and United Nations. 2010. Natural Hazards,
Recovery). 2014. Open Data for Resilience Initiative Field UnNatural Disasters: The Economics of Effective
Guide. Washington, DC: World Bank. Prevention. Washington, DC: World Bank.
INTRODUCTION
43
44
IMAGE
CHAPTER
02
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
Progress, Achievements,
& Remaining Challenges in
RISK ASSESSMENT
R isk assessments require hazard, exposure, and
vulnerability data at the appropriate scale as
well as models with the appropriate resolution to
triggered by a primary hazard event—for example,
fire or tsunami after earthquake.
These are not simple decisions. Since it is a
address the problem of interest. They also require a
rare country or community that is affected by
considered approach to building multidisciplinary,
only a single hazard, assessments that consider
multi-institutional platforms and nontraditional the full range of hazard events often achieve
partnerships around the technical analysis. In this greater traction; on the other hand, the level of
section, we discuss these aspects by reviewing investment for considering all hazards may be too
promising innovations in risk assessment over the great, or momentum following a disaster event
last decade and highlighting some of the greatest may be driving interest in single hazard. Adding
remaining challenges. the complexity of secondary hazards will further
increase the resource and data requirements and
may significantly broaden the institutions involved
Hazard Assessment in a risk assessment. For example, considerations
of fire after an earthquake require additional data
Essential steps required to quantify risk are the sets, as well as engagement with fire authorities,
identification of the relevant hazard(s) and the energy, and water companies. These challenges are
collection of hazard-related data. Although these discussed further in box 2-1.
steps usually occur at the start of a risk assessment,
Once the hazards of interest are defined, the
they are often not easy or straightforward. They next step often involves acquiring a variety of
often involve deciding whether to undertake a single hazard-related data. The most fundamental data
hazard or multi-hazard assessment of the primary define historical events, in particular their date,
hazards and then deciding whether to consider geographical location and extent, and maximum
secondary (or cascading) hazards that may be intensity. Historical events are often used in
45
46
Box 02—1 Multi-Peril Risk Assessment: An Overview deterministic analyses that assess the impact of
past events with current exposure. Historical event
In spite of growing interest in and use of multi-risk assessment approaches,
information is also used to estimate the probability
devising an integrated multi-risk assessment scheme remains a major
challenge. It implies adopting a quite different perspective from that of a
of a hazard occurring at a location with a specific
classical single-risk analysis. A multi-risk analysis does not merely consider intensity.
more than one type of risk. It deals with the various spatial and temporal
interactions that may arise between risks (European Commission 2010). For An event set comprises a suite of stochastic, or
example, cascading or domino effects may include cases in which one event computationally generated, synthetic hazard events
directly triggers another (such as the 2011 Great East Japan earthquake, where with statistical characteristics consistent with the
the earthquake triggered a tsunami, and the ensuing tsunami resulted in historical record. Such event sets can typically
catastrophic failures at the Fukushima nuclear facility). Cascading or domino
include thousands or tens of thousands of potential
effects may also include cases in which the occurrence of one event modifies
events and are intended to define the full range of
the likelihood of another (such as drought and wildfires) and/or increases
the vulnerability of an area to later events. There are also situations where potential events for a hazard. Event sets are used
more than one event may occur at around the same time, without any actual with other information to quantify probabilities of
physical link (e.g., an earthquake just after a windstorm). loss and risk from a hazard.
Another example of cascading effects from a hazard is combustion of a Additional information is used to define the spatial
building by fire caused by an explosion of gas released from a pipeline
distribution of the forces (e.g., the wind field from
ruptured by an earthquake. This scenario occurred following the 1994
a tropical cyclone or the ground motion from an
Northridge earthquake, when approximately 110 earthquake-related fires
were reported within 24 hours of the earthquake (Scawthorn 1997). A slightly
earthquake) associated with a hazard event. Such
different scenario occurred following the 1995 Kobe earthquake, when a information is often incomplete or unavailable and
similar number of fires ignited. Damage to structures from fire caused by the in most cases must be derived from a very limited
Northridge earthquake was well contained; however, nearly 5,500 buildings set of observations. Typically, a combination of
were lost to fire caused by the Kobe earthquake.
observational data and theory is used to define the
The results provided by a full multi-risk approach would need to include a spatial and temporal characteristics of an event.
harmonized quantitative assessment of the different risks and the effects of A collection of the spatial, intensity, and temporal
the possible interactions. Thus, while a multi-risk assessment may make it characteristics for events in an event set is termed a
possible to establish a hierarchy of risks, it can also be used to identify areas hazard catalog.
where efforts to mitigate one hazard may conflict with, or create synergies
with, the response of the system to a second type of hazard, or where planned Hazard catalogs and event sets can be used with
adaptation and mitigation activities may potentially increase or decrease the
risk models in a deterministic or probabilistic
risk from other hazards. An example of this potential risk is the challenge of
manner. Deterministic risk models are used to
building for cyclone wind and earthquake—wherein the strongest concrete
building may decrease vulnerability in a cyclone, but create additional
assess the impact of specific events on exposure.
vulnerability in an earthquake (as happened in Haiti in 2010). Typical scenarios for a deterministic analysis include
renditions of past historical events, worst-case
Source: Anna Scolobig, Alexander Garcia-Aristizabal, Nadejda Komendantova,
scenarios, or possible events at different return
Anthony Patt, Angela Di Ruocco, Paolo Gasparini, Daniel Monfort, Charlotte
Vinchon, Mendy Bengoubou-Valerius, Roger Mrzyglocki, and Kevin Fleming, periods.12 For example, a deterministic risk (or
“From Multi-Risk Assessment to Multi-Risk Governance: Recommendations impact) analysis will provide a robust estimation of
for Future Directions,” input paper prepared for the 2015 Global Assessment
the potential building damage, mortality/morbidity,
Report on Disaster Risk Reduction, available at www.preventionweb.net/gar .
and economic loss from a single hazard scenario.
Risk models are used in a probabilistic sense when
an event set contains a sufficient number of events
for the estimate of the risk to converge at the longest
return period, or the smallest probability, of interest.
Box 02—2 Assessing Damage and Loss Caused by Drought: Example of a Deterministic Assessment
Most studies that evaluate drought damage look at past drought events on an other large-scale natural disasters, such as floods, yet little is known about how
ex post basis. They use self-reports or media accounts, or compare production best to reduce its impact.
for drought and non-drought years (Martin-Ortega and Markandya 2009).
Deficiencies in current approaches to assessing damage and loss caused by
These ex post approaches may fail to determine susceptibility to drought,
drought could be ameliorated using the following:
due to predefined relations between certain drought hazard and resistance
parameters and expected damage. Moreover, they also fail to deal with the • Ex ante evaluation methods. Properly designed, these will help to address
dynamics of drought risk and damage over time. Specific problems with these the projected increase in frequency and intensity of droughts, make it
ex post approaches include potential bias from self-reports and media accounts possible to learn about changes in drought damage over time, and facilitate
of damage, and significant uncertainty in comparisons between drought and evaluating and prioritizing mitigation strategies for drought damage.
non-drought agricultural production. Additionally, these comparisons fail to
• More sophisticated drought damage models that are based on assessments
account for factors other than drought that influence production. They do not
of losses to economic flows. These models account for indirect losses of
distinguish between direct drought effects that damage crops and indirect sector-specific added value, wage losses, or relocation expenses and could
effects spreading through the economy. significantly improve current cost assessments.
A further problem with current drought damage models is that they are not • Models that capture the effect of drought mitigation measures. Existing
designed to account for drought mitigation measures. This means that the databases on drought-induced soil subsidence and its effect on different
damage-reducing effects of drought mitigation measures are largely unknown, a building types could provide a basis for this future work.
situation that makes choosing among the different mitigation measures difficult.
Source: Heidi Kreibich and Philip Bubeck, “Natural Hazards: Direct Costs and
This lack of information about mitigation strategies is especially problematic in
Losses Due to the Disruption of Production Processes,” input paper prepared
the case of drought-related soil subsidence. Existing studies suggest that soil for the 2015 Global Assessment Report on Disaster Risk Reduction, available at
subsidence (which can severely damage buildings) can be as destructive as www.preventionweb.net/gar.
The Small Sample Size of Hazard Records
Figure 02—1
SAMPLE SIZE
Hypothetical drought
index showing
periods of extreme
DROUGHT INDEX
dryness (above the
dotted red line) and
periods of extreme
wetness (below the
dotted blue line); the
historical record does
not capture extreme
dry and wet periods
experienced prior to
1300 1400 1500 1600 1700 1800 1900 2000
its start in 1900.
YEAR
Extreme Dryness Extreme Wetness
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
47
48
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
In other words, a probabilistic risk model contains Figure 2-1 illustrates this challenge. If the sample
a compilation of all possible “impact scenarios” for size (1900 and after) is the historical record, then
a specific hazard and geographical area. Note that it would appear that extreme flood and drought are
hazard catalogs are generally associated with rapid not a concern. Similarly, if we consider the period
onset hazards. Risk assessments for slow onset 1800–1900, flood would be seen as a risk, but not
hazards, such as drought, are typically undertaken drought. Herein lies the challenge of determining
using deterministic approaches. (Additional issues the return period for rare and extreme hazard
associated with modelling drought risk and impacts events. In the case of hydrometeorological cycles,
are discussed in box 2-2. For a cost-benefit approach determining the return period is difficult; for
to risk that deals with the effects of drought on geophysical hazards such as volcanic eruptions and
livestock, see box 2-3). large earthquakes, which may occur every 1,000,
10,000, or 100,000 years, it is incredibly complex.
Convergence of results is a concern when using a
risk model probabilistically. As a simple example, A variety of hazard-dependent data are required
consider a simulation of 100 years of hazard events. to generate a hazard catalog. Knowledge of the
This simulation is too short to determine the 100- distribution of soil types, for example, is required to
year return period. A random sample of 100 years of model the spatial variation of ground acceleration
events could easily omit events, or include multiple (shaking) from an earthquake; values for surface
instances of the same event, that on average roughness are needed to define the distribution of
would occur once every 100 years and therefore wind speed from a tropical cyclone; and a digital
dramatically affect determination of return period. elevation model (DEM) is needed to determine
Box 02—3 A Cost-Benefit Analysis of Livestock Protection in Disaster Risk Management
Animal-related income streams are critical to underlying causes of To assess the number of animals reached and the total cost of WSPA's
risk and provide economic and social well-being in the world’s poorest intervention, WSPA post-intervention response reports were used. The
and most vulnerable regions. Protecting livestock is crucial because potential income derived from animals treated was considered the
it protects the livelihoods of livestock producers and guarantees food benefit of the intervention. For the sake of this preliminary analysis, it
was assumed that half of the animals treated would have died had they
security for millions of people.
not received treatment.
To learn more about the role of livestock protection in disaster risk
The intervention is estimated to have generated $2.74 of benefits in
management (DRM), the World Society for the Protection of Animals
the form of avoided losses for every $1.00 spent. If the time period for
(WSPA) commissioned Economists at Large Pty Ltd to conduct a cost-
potential income generated by the livestock is extended to three years
benefit analysis of a WSPA intervention in the Mwingi District in Kenya.
and the cumulative effect of secured livelihoods is taken into account,
The intervention began in 2011, in response to long-lasting drought the benefit-cost ratio increases to $6.69 in benefits for every $1.00 spent.
conditions, and involved treating livestock brought to WSPA’s Mwingi Based on the research described here, WSPA is developing a framework
operation to increase the likelihood that the animals would survive until for estimating the impacts on communities and households of losing
the next rainy season. livestock in a disaster.
The analysis focused on the household income impacts to owners of Source: Nicole Fassina, World Society for the Protection of Animals,
livestock who brought their animals for treatment. Beyond this, the "Cost-Benefit Analysis of Livestock Protection in Disaster Risk
Management," input paper prepared for the 2015 Global Assessment
analysis sought both to understand the economic impact of livestock
Report on Disaster Risk Reduction, available at www.preventionweb.
operations on local and regional economies and to create an applicable
net/gar; based on Economists at Large, Cost-benefit Analysis of WSPA’s
and scalable risk reduction model that would assess vulnerabilities and Mwingi Intervention in Kenya (Melbourne: World Society of the Protection
return on investment strategies within livestock-dependent communities. of Animals, 2013).
CHAPTER
02
flood depth. Fortunately, some data can be common data may be usable to different degrees—for
to multiple perils. For example, topography as example, data may not be digitized, may lack
defined by a DEM is required for modelling floods, necessary metadata, or may require substantial
tsunamis, sea-level-rise inundation, landslide improvement before use. A compilation of publicly
susceptibility, storm surges, and detection of available hazard-related data with global coverage
earthquake fault lines. is given in table 2-1. Some of these data sets, such
as the records for the location and intensity of
Hazard data can be open, proprietary, or (if they earthquakes and tropical cyclones, provide global
have yet to be collected) unavailable. Even available coverage and are considered authoritative records
Data Use Source Table 02—1
Examples of Globally
De ine date, intensity, and location o
Earthquake events http://www.globalcmt.org
earthquakes Available Hazard-
related Data
Earthquake events Earthquake date, location, and intensity http://www.ncedc.org/anss/
(A) Best-track data are
Assess distance rom known aults and de ine defined as “a subjectively-
Quaternary ault maps http://earthquake.usgs.gov/hazards/q aults/download.php
ault motion smoothed representation of
a tropical cyclone's location
Attenuation relationships Calculate propagation o seismic waves http://www.opensha.org/glossary-attenuationRelation and intensity over its
lifetime.” National Hurricane
30m shear velocity (Vs30) Determine seismic wave attenuation http://earthquake.usgs.gov/hazards/apps/vs30/ Center, “Glossary of NHC
Terms,” http://www.nhc.
Topography—digital elevation De ine elevation and slope or loods,
noaa.gov/aboutgloss.shtml.
http://eros.usgs.gov/elevation-products
data (~90m resolution) tsunamis, landslides, etc.
Tropical cyclone best-track Determine location and intensity o tropical
http://www.ncdc.noaa.gov/ibtracs/
data(A) cyclones
Assign roughness or calculating winds rom
Land cover http://due.esrin.esa.int/globcover/ US: http://www.mrlc.gov/
gradient-level winds
De ine behavior o waves rom storm surge
Bathymetry http://www.ngdc.noaa.gov/mgg/inundation/tsunami/
and tsunamis
Develop event sets or tornadoes and hail
Tornado and hail paths http://www.spc.noaa.gov/wcm/#data
rom severe convective storms
Catalog o all known historical (and in some
Volcanic eruptions cases geological) eruptions with indicative http://www.volcano.si.edu/search_eruption.c m#
impacts (where known)
Tsunami events and run-ups Tsunami hazard http://www.ngdc.noaa.gov/hazard/tsu_db.shtml
Flood events since 1985 Flood hazard http:// loodobservatory.colorado.edu/Archives/index.html
Fire events 1997–2011 Wild ire hazard http://due.esrin.esa.int/w a/
Reconstruct atmospheric winds, precipitation, http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanal
Atmospheric reanalysis data
temperature, etc. ysis.html
49
Hurricane satellite data
Homogeneous estimates o hurricane intensity http://www.ncdc.noaa.gov/hursat/
(HURSAT)
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
50
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
that compile the best available data.13 Other develop global flood models, which will use a global
global data sets may not be of optimal quality for flood catalog; one model, GLOFRIS (GLObal Flood
risk assessment. For example, openly available Risk with IMAGE Scenarios), is already in use (see
topographic data are not optimal for modelling section 3-23 for a more detailed discussion).
hydrometeorological hazards because of their
A critical requirement acknowledged by all experts
relatively coarse resolution. Poor resolution of
working in hazard modelling is the need for a high-
elevation data has a significant impact on flood risk,
resolution, open DEM. Currently, the 90m Shuttle
since small changes in elevation can involve huge
Radar Topography Mission (SRTM) is the only
changes in the predicted inundation area in many
global open DEM, with 30m resolution available
relatively flat floodplains and coastlines.
in some countries. Satellite-based Interferometric
The spatial characteristics of an event are usually Synthetic Aperture Radar (InSAR) appears to be one
defined by combining theoretical and empirical promising approach for generating these data on a
knowledge with other observational hazard-related global scale; one satellite currently using InSAR is
data because of the sparseness of the relevant the TerraSAR/Tandem-X of DLR (German Aerospace
observations. For example, quantifying the wind Center) and Astrium Geo-Information Services. A
field for a tropical cyclone as it travels inland growing alternative to a satellite-based collection
highlights the difficulty of estimating the spatial of elevation data is the use of airplanes and/or
distribution of a hazard. Wind speed and pressure helicopters to derive high-resolution surface data
measurements from observing stations can be used on a smaller scale via LiDAR14 or airborne InSAR.
to estimate two parameters, a cyclone’s maximum Both of these “active” methods, while expensive,
wind and the radius of maximum wind. It is often are capable of generating very accurate and high-
resolution surface and terrain elevations. Collection
impossible to obtain high-quality measurements,
of LiDAR data is growing across the globe; however,
however: the number of observational platforms is
the cost, time, and technical processing aspects of
limited, existing observation stations are not sited
this approach prohibit its widespread accessibility.
optimally, power may fail during the cyclone, and
anemometers may be damaged by flying debris. There are two types of DEMs: a digital surface
Surface pressure measurements of the cyclone are elevation model and a digital terrain model. A digital
easier to collect, and the minimum central pressure surface elevation model provides surface elevations
has a large influence on maximum wind speeds, but that describe the elevations of features such as
these surface pressures must be converted to surface buildings and treetops. A digital terrain model
wind speeds for risk modelling purposes, and this provides elevations of the bare ground surface and
is where the theoretical and empirical knowledge is neglects objects such as buildings and trees. The
critical. impact of the different models on hazard and risk
assessments can be significant—see box 2-4—but
Most hazard event sets and catalogs are developed
the combination of these different DEMs offers
region by region. Exceptions include the global
opportunities for better characterizing the built
earthquake event set generated by the Global
environment.15
Earthquake Model (GEM), and the tsunami, volcanic
eruption, cyclone, and drought hazard event sets To assess risk from multiple meteorological hazards
developed as part of the global risk model under on a global scale, one should consider the hazards’
the leadership of the UN Office for Disaster Risk spatial and temporal correlations and how they vary
Reduction. There are also a number of efforts to as a function of climate. For example, the probability
CHAPTER
Box 02—4 The Importance of Accurate Elevation Data for Understanding Tsunami Hazard
Tsunami inundation models provide fundamental information about coastal
02
2. SRTM and ASTER (D) data sets, although freely available with near global
areas that may be inundated in the event of a tsunami. This information coverage, should not be used for modelling onshore tsunami hazard, since
has relevance for disaster management activities, including evacuation the results can be dangerously misleading.
planning, impact and risk assessment, and coastal engineering. A basic
input to tsunami inundation models is a digital elevation model—that is, a This study makes clear that accurate elevation models are crucial for
model of the shape of the onshore environment. Onshore DEMs vary widely understanding tsunami hazard. Investing in high-quality, accessible elevation
in resolution, accuracy, availability, and cost. Griffin et al. (2012) assessed data in tsunami-prone areas will underpin better risk reduction planning at
how the accuracy and resolution of DEMs translate into uncertainties in the local level.
estimates of tsunami inundation zones. The results showed that simply using
the “best available” elevation data, such as the freely available global SRTM (A) The observation data are from Tsuji et al. (1995).
elevation model, without considering data accuracy can lead to dangerously
(B) See E. Rodriguez, C. S. Morris, J. E. Belz, E. C. Chapin, J. M. Martin, W.
misleading results.
Daffer, and S. Hensley, “An Assessment of the SRTM Topographic Products,”
The top part of the figure shows tsunami inundation models for the Jet-Propulsion Laboratory D-31639. http://www2.jpl.nasa.gov/srtm/SRTM_
1992 tsunami in Flores, Indonesia (Griffin et al. 2012). For each model all D31639.pdf.
Depth (m)
INUNDATION LIMIT
Elevation (m)
Modelled inundation for the 1992 tsunami in Flores, Indonesia (top) and underlying elevation data used in the model (bottom).
Source: Griffin et al. 2012.
Note: Top images show inundation estimates from the 1992 tsunami in Flores, Indonesia, with arrow pointing to black line showing the observed inundation limit.
Bottom images show elevation data for LiDAR (left), airborne InSAR (middle), and SRTM (right).
parameters are the same except for the elevation data, shown in the bottom (C) However, further testing of tsunami inundation sensitivity to
of the figure. Inundation model results are overlain with field observations underlying DEM may be required in other coastal environments
of the actual inundation.(A) LiDAR and airborne InSAR give inundation area with different geomorphology before this inference becomes a
extents that are comparable with historical data. However, results obtained widespread recommendation.
using the SRTM data set, with lower vertical accuracy, (B) show negligible
tsunami inundation. (D) ASTER elevation data also significantly underestimate the wet area. See
Griffin et al. (2012) for the full analysis.
Two main inferences can be drawn from the results:
Source: Jonathan Griffin (Australia-Indonesia Facility for Disaster Reduction,
1. .The most accurate and expensive data are not always needed, depending Geoscience Australia); Hamzah Latief (Bandung Institute of Technology); Sven
on the purpose. Airborne InSAR, which is an order of magnitude cheaper to Harig (Alfred Wegener Institute); Widjo Kongko (Agency for Assessment and
acquire than LiDAR, may be suitable for tsunami evacuation planning. (C) Application of Technology, Indonesia); Nick Horspool (Geoscience Australia).
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
51
52
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
of tropical cyclone landfall varies as a function of the Pacific region. Regardless of the uncertainties
the El Niño-Southern Oscillation (ENSO) along the associated with quantifying future changes in
Queensland coast of Australia, the U.S. coastline, meteorological hazards, sea level is certain to
and in the northwest Pacific. Generally, warm El rise in response to melting of continental ice
Niño years are associated with a reduced rate of caps and thermal expansion of seawater. Higher
landfall, and cool La Niña years are associated with sea levels will exacerbate coastal flooding from
a higher rate of landfall (Flay and Nott 2007; Elsner storm surge, intense precipitation events, and
and Jagger 2006; Wu, Chang, and Leung 2004). tsunami inundation.
There are also possible cross-peril correlations. In
Climate change and sea-level rise are not the only
many areas, for example, flood risk and drought risk
future threats for coastal regions. Many coastal
are strongly correlated with ENSO.
regions suffer from severe subsidence. In some
The response of meteorological hazards to natural locations the increase in subsidence is much larger
climate variability highlights the possibility that than the sea-level rise. For example, in Jakarta
the risk from these hazards will respond to future the subsidence is currently over 10cm per year.
changes in climate. It is difficult to specify with According to Brinkman and Hartman (2008), Jakarta
certainty how hazard occurrence and intensity will is heading toward a disaster with the juxtaposition
change by region, and this is an area of significant of the high sea tides and the subsidence rate. Up to
research and modelling. A case study described in
16
4 million people and approximately 25 percent of
section 3-24 highlights the changing risk associated the city will be affected by inundation from the sea
with future changes in tropical cyclone activity in within the next 15 years if action is not taken.
CHAPTER
02
geospatially linked inventories that include public
Exposure 17
infrastructures are rare and not publicly available in
Exposure modelling has a critical role to play in most developing countries, where exposure model
risk assessment. Empirical studies suggest that the development is most needed for risk assessments.
greatest influence on output loss estimates from risk At the global scale, efforts to generate globally
models derives from exposure data, as opposed to consistent exposure data sets in terms of quality
either hazard or vulnerability data (see for example and resolution have increased. Experience has
Bal et al. 2010; Chen et al. 2004; Spence et al. 2003; shown that development of exposure data sets
Lavakare and Mawk 2008). requires innovative, efficient methodologies
for describing, collecting, validating, and
The process of exposure modelling identifies the
communicating data, while also accounting for the
elements at risk in areas that could potentially be
inherent spatiotemporal dynamics associated with
affected by natural hazard events (UNISDR 2009;
exposure—that is, the dynamics by which exposure
Ehrlich and Tenerelli 2013; van Westen 2012). In
evolves over time as a result of (unplanned)
other words, if a hazard occurs in an area with
urbanization, demographic changes, modifications
no exposure, there is no risk. This is the case, for
in building practices, and other factors.
example, with an earthquake in an unpopulated area
of Alaska. The information used to develop exposure data sets
can be derived from various sources and methods.
Exposure modelling techniques have been developed
At a local level, common data sources are council
at various scales, from global to local. Significantly,
and local government agencies, household surveys,
global-scale and local-scale modelling use different
aerial photos, and individual architectural/structural
methodologies: the former tends to take a top-
drawings. At a regional level and above, state-based
down approach, with work being carried out by
agencies, statistical offices, census data, investment
governments or large institutions, whereas the
and business listings, employment figures, and
latter works from the bottom up by methods such
existing geographic information system (GIS) data
as crowdsourcing and in situ surveys. At least four
are common sources of exposure information. At
homogeneous inventory regions—urban residential,
the coarsest level of resolution, national statistical
urban nonresidential, rural residential, and rural
agencies, census data, global databases, and remote
nonresidential—are usually defined to capture the
sensing are used for developing exposure data.
differences in occupancy and construction. Data
sources also vary by resolution. Commercial risk models have developed the so-
called industry exposure databases for regions
At the local scale, high-resolution exposure
where risk models are offered. These exposure data
data have been developed on an ad hoc basis, in
can include detailed information on construction as
areas where risk modelling has been carried out.
well as estimates of the value of the contents within
Crowdsourcing has become a common and valuable
a structure. The resolution of the exposure data is
tool for collecting detailed bottom-up data, but this
typically at the postal code level with varying levels
approach has limits, both in the type of data it can
of occupancy types. However, these data are almost
collect and in the quality of those data. In addition
always proprietary.
to being used to develop exposure data at a local
scale, crowdsourcing has also been used to validate The classification (taxonomy and ontology) used to
global-scale data. At the national scale, complete generate these exposure data varies from data set
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
53
54
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
to data set; this variation is problematic for efforts Wald, and Porter 2010a). In addition, three global
to merge independently developed data sets. Nor exposure databases are slated for publication in
is there a commonly agreed upon taxonomy that 2014, the global risk model by UNISDR (De Bono
accounts for features such as construction attributes 2013; see section 3-7), the GED4GEM by the Global
and asset valuation across different hazards. Earthquake Model (Dell’Acqua, Gamba, and Jaiswal
2012; see section 3-6), and the World Bank exposure
In recent years, several data sets with global database, which will be completely open and
coverage have made the first step in overcoming suitable for multi-hazard analyses (Gunasekera et al.
these obstacles. The first such global exposure data 2014).18 Many of these newer exposure models take
set was developed in 2010 for PAGER (Prompt advantage of aspects of building typology taxonomy
Assessment of Global Earthquakes for Response), originally compiled in the PAGER database. Several
a global near-real-time earthquake loss estimation examples of global exposure data sets are given in
system, by the U.S. Geological Survey (Jaiswal, box 2-5.
Box 02—5 Global Exposure Data Sets
Global human exposure. Global models of human exposure mostly
/// ///
suitable mainly for earthquake and cyclone probabilistic risk modelling.
describe population data either on a regular grid or in specific settlement It employs building type classifications for different size categories of
coordinates or geographical boundaries. A widely used product is settlements as developed by the World Agency of Planetary Monitoring
the Gridded Population of the World (GPWv3), a gridded data set that and Earthquake Risk Reduction (Wyss et al. 2013). The goal of the
provides a spatially disaggregated population layer constructed from GED4GEM (Dell’Acqua, Gamba, and Jaiswal 2012) is to create an open
national or subnational input units of varying resolutions.(A) The native homogenized database of the global building stock and population
grid cell resolution is 2.5 arc-minutes. Population estimates are provided distribution, with spatial, structural, and occupancy-related information
for the years 1990, 1995, and 2000, and are projected to 2005, 2010, at different scales, as input to the GEM risk platform OpenQuake.(C) Its
and 2015. Other global human exposure models include commercially building type classifications follow the GEM taxonomy, which is designed
available LandScan (Bhaduri et al. 2007) and the open WorldPop. These primarily for earthquake vulnerability assessments, and its multi-scale
models are based on the integration of several information sources, database structure contains information on buildings and populations
including census and remote sensing, and are affected by a significant from the country scale down to the per-building scale. The initial version
range of uncertainties (Potere et al. 2009; Mondal and Tatem 2012). of GED4GEM, planned for late 2014 release, will contain aggregate
information on population, built area, and reconstruction costs of
Characterization of global built-up area. The Global Human Settlement
residential and nonresidential buildings at 1km resolution. Detailed data
/// ///
Layer (GHSL) is developed and maintained by the Joint Research Centre
sets on single buildings will be integrated for a selected number of areas
of the European Commission. GHSL integrates several available sources
and will increase over time.
about human settlements with information extracted from multispectral
satellite images. The underlying automatic image information extraction (A) See the Gridded Population of the World website at http://sedac.
work flow makes use of multi-resolution (0.5m–10m), multi-platform, ciesin.columbia.edu/data/collection/gpw-v3.
multi-sensor (pan, multispectral), and multi-temporal satellite image data
(Pesaresi and Halkia 2012). The Global Urban Footprint is being developed (B) For PAGER, see Wald et al. (2008) and the website at http://
by the German Aerospace Center (DLR) and is based on the analysis of earthquake.usgs.gov/earthquakes/pager/; for GED13, see De Bono (2013);
Synthetic Aperture Radar (SAR) and optical satellite data. The project for GED4GEM, see http://www.nexus.globalquakemodel.org/ged4gem/
intends to cover the extent of the large urbanized areas of megacities for posts.
four time slices: 1975, 1990, 2000, and 2010 (Taubenböck et al. 2012).
(C) For OpenQuake, see http://www.globalquakemodel.org/openquake/
Global description of building stock. Several global exposure databases
/// ///
about/.
include physical exposure information; examples include PAGER, the
Source: Massimiliano Pittore, Marc Wieland, and Kevin Fleming, “From
Global Exposure Database for the 2013 Global Assessment Report on
Remote Sensing to Crowdsourcing: Perspectives of a Global, Dynamic
Disaster Risk Reduction (GED-13), and the Global Exposure Database for
Exposure Model for Georisk Assessment,” input paper prepared for the
GEM (GED4GEM).(B) Using the CAPRA platform (Cardona et al. 2012), GED- 2015 Global Assessment Report on Disaster Risk Reduction, available at
13 aims to create an open global building and population inventory www.preventionweb.net/gar.
CHAPTER
02
Categories of information included in
///
of the risk analysis, and consequently into the
exposure models. There are several categories of
///
loss estimates. The uncertainties and associated
assets that need to be included in a comprehensive limitations in the final risk assessment then
exposure model (table 2-2). The broad variety of need to be communicated to the end-users of
categories illustrates the necessity of combining this information.
efforts from different disciplines, such as
geographical science, statistics, engineering, Information required for the modelling of
///
mathematics, economics, remote sensing, and socio- physical damage. On a national scale, reliable
///
demographics. data on physical exposure are less available than
population data. Information is often missing or
It is clear that as more data are integrated, modelled,
incomplete, and few governments have developed
and jointly analyzed, uncertainties propagate in the
national exposure databases of buildings and
model and in the subsequent results. A choice needs
infrastructure that are open and can be used to
to be made about whether slightly more-detailed
understand the impacts of multiple hazards (Turkey,
data will improve a model or merely add to the noise
Australia, the United States, and New Zealand are
and confusion. The impossibility of eliminating
uncertainty in hazard and vulnerability modelling is exceptions). Thus it is not surprising that most
widely recognized. After all, every model constitutes exposure data sets at the national scale or above
a simplified approximation of reality. Depending on use the spatial distribution of population as a proxy
geospatial data characteristics (including resolution for developing exposure estimates. This is a rapidly
aspects) and integration factors, uncertainty may evolving area, however, and more governments
increase. It is therefore essential for uncertainties are seeing the widespread value of developing
to be conceptually integrated into the framework exposure information.
ASSET CATEGORIES DESCRIPTION Table 02—2
Categories of a
Population Demographic characteristics Comprehensive
Various occupancy types such as residential, commercial,
Exposure Model
public, administrative, industrial classes. Also includes
Property (buildings, etc.) Source: Adapted from
various different structural building types such as exterior
GFDRR (2011).
wall and roof types.
Agriculture Crop and land-use characteristics
Transportation Road, rail, air, and other transport-related networks
Sports stadiums, marketplaces, churches/temples/mosques,
Large loss facilities
schools and other high population density infrastructure
Hospital and health care facilities, public buildings,
Critical/high-risk loss facilities telecommunications, airports, energy systems, bridges and
other facilities critical to the recovery of a disaster
Oil, gas, and water supply pipelines/distribution systems,
Other lifelines—utilities, pipelines nuclear and chemical power plants, wastewater, and
electricity systems
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
55
56
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
The basic information needed to model the response exposure can be a key factor in determining the
of a structure to a hazard event includes its location, impact of rapid onset events such as earthquakes,
occupancy, construction type, length or density landslides, or tsunamis. Models of building
(for road and railway), and replacement value. occupancy that consider daily patterns have been
The response of a structure to a hazard event can proposed (Coburn and Spence 2002; Coburn,
be more realistically simulated using additional Spence, and Pomonis 1992), but collecting the
structural information such as its square footage, necessary data to update such models can be
shape, height, age, roof type, irregularities, and very time- and resource-intensive. A promising
material and mechanical properties, as well as alternative approach takes advantage of cellular
building codes applicable to it. For hydrological phone data provided by telephone companies
hazards, additional details useful for vulnerability (Wesolowski et al. 2013; Lu et al. 2013).
assessments include information on the height
above ground of the first occupied floor, distance Exposure data collection approaches—full
///
from water channels, and the presence of enumeration, sampling, or disaggregation
basements. Knowledge of the replacement value using proxy data. In general terms, top-down
///
makes it possible to estimate the direct loss and bottom-up approaches are used to collect
associated with an event. exposure data. Approaches that use bottom-up
methods commonly employ direct observation,
Modelling economic losses. Valuation data
/// ///
which relies on two principle strategies: full
are critical for quantitatively assessing economic enumeration or sampling. With the full enumeration
loss from disasters. The reinsurance industry uses approach, each exposed asset in the study area is
claims and other economic data sets to calibrate detected and defined. This approach can be very
its exposure models. However, this information
accurate and detailed but also requires a greater
is often proprietary and limited to insured risks.
expenditure of time and other resources. Census
Obtaining comprehensive loss data for uninsured
data are commonly used to fully enumerate human
property is much more difficult. Proxy data such as
populations, though this approach is best suited to
socioeconomic surveys, labor statistics by economic
developed countries, which are likely to have slow or
sector, floor area per employee by type of activity,
moderate population growth and up-to-date census
etc. are used to determine nonresidential building
data. Volunteered geographic information (VGI),
stock values. Accounting for a structure’s contents
another approach to full enumeration, derives data
becomes particularly significant when modelling
from the joint efforts of many individuals who
nonresidential occupancy classes.
voluntarily collect and submit data. VGI may be
Incorporating the temporal variation in
///
either structured or unstructured—the latter applies
human exposure. Other important factors related
///
to unsystematic, non-authoritative initiatives such
to exposure data are population and demography as OpenStreetMap (see box 1-2), which rely on
characteristics that highlight the movement of participants’ interest and motivation. The structured
population through the course of a day. Consider, approach also involves volunteers but has an
for example, the swelling of populations in major authoritative component that directs volunteers’
metropolitan areas during the work day, or the efforts toward certain tasks (Chapman 2012), such
varying population characteristics of areas of as a government-led participatory mapping program
cultural or religious value depending on the day and/ to collect exposure data for risk assessment (see
or time of the year. Temporal variability in human section 3-3).
Box 02—6 Indirect Characterization of Exposure
Population: A global distribution of population data, in terms of counts or
/// ///
mostly characterized by artificial structures, including roads and
density per unit area, is considered the primary source of information for buildings. Built-up areas are often described by binary masks that
exposure assessment. For instance, the GAR13 exposure database uses clearly outline the boundary of settlements. This can be considered an
the commercial global LandScan population database to obtain a spatial intermediate description of exposure, where the characterization of the
distribution of buildings’ structural types (de Bono 2013). Analogously, built-up environment is improved with respect to a simple population
the GED4GEM database exploits population data to disaggregate layer. Built-up masks can be reliably obtained by processing different
exposure estimation (Dell’Acqua, Gamba, and Jaiswal 2012). In both cases remote-sensing data, thus effectively addressing global-scale mapping.
the knowledge of the percentage of population living in each building Examples of global built-up area products include the Global Rural-Urban
type, or the estimated average dwelling occupancy, is used to link the Mapping Project (GRUMPv1),(A) the Global Human Settlement Layer
population to the physical exposure. Global population models also allow (Pesaresi and Halkia 2012), and the Global Urban Footprint (GUF) (Esch et
use of empirical vulnerability functions, where direct estimates of loss al. 2010).
are obtained directly in terms of population exposed, and the main loss
(A) See the GRUMP website at http://sedac.ciesin.columbia.edu/data/
metrics account for fatalities (Jaiswal and Wald 2010). Many global models
collection/grump-v1.
use human exposure as a basic ingredient to define a more refined
“hazard-specific exposure” (Dilley 2005; Peduzzi et al. 2009; Allen et al.
Source: Massimiliano Pittore, Marc Wieland, and Kevin Fleming, “From
2009). Remote Sensing to Crowdsourcing: Perspectives of a Global, Dynamic
Exposure Model for Georisk Assessment,” input paper prepared for the
Built-up areas: A further step with respect to population distribution is
/// ///
2015 Global Assessment Report on Disaster Risk Reduction, available at
the spatial delineation of built-up areas, that is, impervious surfaces www.preventionweb.net/gar.
With a sampling approach, summary statistics sampling approach where high-resolution imagery
for a large area are estimated based on smaller (manual or automatic extraction of features) or
subset areas. Increasingly, census methodologies direct observation is used to fully enumerate assets
are turning to sampling and statistical modelling (buildings, roads, bridges) and their geometric
rather than full enumeration because they provide characterization (footprint, shape, height) within
more up-to-date and more accurate information each of the sampling areas that represent the
with less effort than traditional methods. A rolling common density pattern classified during the
census approach—in which only small areas are first step. Alternatively, if time and resources
fully enumerated and other, highly populous areas permit, optical satellite or aerial images can be
are continuously sampled at the rate of around used to extract all of the footprints for buildings
10 percent a year—makes it possible to update in an exhaustive manner. To provide a complete
data annually instead of every 5 to 10 years (UN description of the exposure, however, the footprints
Statistical Division 2008). should be combined with in situ direct observations
or other data sets (such as national statistics
Remote sensing is on occasion used in
information) that provide additional data that
conjunction with these sampling methodologies
cannot be captured from above (e.g., construction
(Adams and Huyck 2006; Müller, Reiter, and
features or building use).
Weiland 2011; Geiß and Taubenböck 2013). For
instance, urban areas can be classified according to In recent years, digital in situ data capturing systems
their density using satellite images, followed by a have started to emerge, which allow the user to
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
57
58
Box 02—7 How Study Scale Drives Exposure Data collect and generate exposure information using
Collection Methods handheld direct observation tools in combination
Assessing how a community will be affected by natural hazards requires a
with other disaggregation or extrapolation
fundamental understanding of the elements at risk. The type of data needed methodologies (FEMA 2002). An example includes
for a hazard impact assessment depends on the nature of the problem that is
the open source suite of tools developed under GEM
being addressed and is independent of the location or scale of the problem. In
direct contrast to this, for projects with limited resources (e.g., time, funding),
called Inventory Data Capture Tools (IDCT). IDCT
the methods used for data collection depend on the scale of the study. takes information generated from the analysis of
satellite images to characterize built-up areas and
If the goal of a natural hazard impact assessment is to understand whether
a particular feature will be affected by a certain level of hazard, then it will be combines it with sampled direct field observations
enough to simply know the location of that feature, and whether the location on individual buildings using handheld devices
lies in a zone of potential hazard. For example, landowners who want to know
or paper survey forms. This information is then
whether their land is likely to be inundated by a flood need only locate their
land within published flood hazard information. This example demonstrates integrated through the use of mapping schemes to
scale independence: if the entire population sought this information, it would generate exposure information.
still be necessary to know only the location of land relative to zones of hazard.
Indirect, top-down disaggregation approaches
In contrast, if the aim of a study is to understand the potential economic
losses and casualties that could result from a natural hazard, then it is use exposure proxies to develop exposure data
necessary to understand more than just the location of a feature. For the sets when direct observation alone is not feasible.
quantitative estimates required by this more comprehensive risk assessment,
Information on the spatial distribution of
understanding the type of construction materials, the age of construction, and
the number of people within a building is necessary. Note that while additional population and built-up areas allows the exposure
information is required in this example, the information is still independent to be disaggregated into finer resolutions. Some
of the scale of the study: whether data are for a single household or every
examples of this approach are described in box 2-6.
household in a megacity, assessing the possible economic losses from
flooding requires information about the number of stories in a building and its
Multi-source integration. The growing variety
construction type and age.
/// ///
of possible exposure information sources requires
The same example that demonstrates scale independence for the type of
the flexible integration of existing information
data collected demonstrates scale dependence for data collection methods.
For the individual landowner/household, firsthand observation is the most from different acquisition techniques, scales, and
effective method for collecting relevant data, regardless of whether they are accuracies, so that no available information is
for a simplistic “wet/not wet” assessment or a quantified estimate of risk to
discarded. An example for a probabilistic integration
inform an insurance policy. However, undertaking either of these types of
assessments through firsthand individual data capture at a megacity, national, approach is given in Pittore and Wieland (2013).
or regional scale is impractical and likely impossible. This method is based on Bayesian networks and
Source: A. T. Jones, K. Van Putten, M. Jakab (Geoscience Australia); M. L. allows for the sound treatment of uncertainties
Bautista, B. C. Bautista, I. C. Narag (Philippine Institute of Volcanology and
and for the seamless merging of different data
Seismology); A. Wibowo (Badan Nasional Penanggulangan Bencana); K.
Chapman (Humanitarian OpenStreetMap Team). sources, including legacy data, expert judgment, and
inferences based on data mining.
There are clearly many approaches to collecting
exposure data; however, for best results the decision
on the approach must be aligned with the scale and
purpose of the risk assessment (see box 2-7).
Damages to
structures
60% Unreinforced rubble stone masonry construction
10% Reinforced concrete block construction
0
low high extreme
Hazard intensity
(magnitude of wind, flood, earthquake, etc.)
estimates for contents, business interruption, and
Vulnerability and Loss
Figure 02—2
outlying structures tend to be just a simple function The relationship
Vulnerability is typically described in terms of of loss to the main structure. Fatality estimates tend between hazard
damage and/or loss. Damage and loss to a structure to be based on knowledge of local population and intensity and damage
are assessed using functions that relate hazard empirical relationships based on structural damage to structures
intensity to damage; see figure 2-2 for an illustration. or hazard characteristics. For example, PAGER A hazard of the same
Various adjectives are used to describe the functions, estimates fatality rates based on ground-shaking intensity results in
including “fragility,” “damage,” and “vulnerability.” intensity and a region-specific fatality rate (Jaiswal significantly different
Engineers use fragility functions to quantify damage and Wald 2010). A somewhat similar approach is damage to a reinforced
and vulnerability functions to quantify loss caused concrete block construction
used for floods, where the fatality rate is a function
by a hazard. However, it is not uncommon to use building than to an
of flood depth (Boyd et al. 2010). unreinforced rubble stone
the term vulnerability function when discussing
masonry construction
damage. Damage is often quantified using a damage Generally, functions are defined using mean values building.
ratio where 0 is equivalent to no damage and 1 is and a coefficient of variation (CV) for a range of
equivalent to complete destruction. Multiplying hazard intensities (three-second gust wind speed
value by the damage ratio gives an estimate of direct at 5km/hr intervals, peak ground acceleration at
loss. intervals of 0.1g, flood depth at 50cm intervals, etc.)
The CV tends to decrease with more information.
The resolution of loss estimates will vary by model.
For a global- or regional-scale model, the losses may For example, a relatively precise (small CV)
resolve only total direct loss, whereas detailed site- estimate of damage would be expected if one had
specific models may estimate loss to a structure, its a vulnerability function that accounted for the
contents, and outlying buildings and include time- structural details of a building designed and built
dependent losses such as business interruption. to withstand the expected hazard intensities.
Site-specific fragility and vulnerability functions can The damage estimate would have considerable
account for differences in structural characteristics, uncertainty (large CV) if the structure were part
such as roof covering and how it is attached. Loss of aggregate occupancy data. An alternative to a
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
59
60
Box 02—8 The Uses of Loss Inventories function that provides a mean and a CV is to use a
damage probability matrix.
The terms “loss” and “damage” are often used interchangeably in reference to
the adverse impacts of disasters on society, economies, and the environment.
Methods of assessing damage vary greatly depending
In the context of disaster loss inventories, losses are quantifiable measures
upon the type of exposure under consideration
expressed in either monetary terms (e.g., market value, replacement value)
or counts such as number of fatalities and injuries. Damage is a generic term
(e.g., people, buildings, livestock), the resolution
without quantitative characteristics, which does not mean that damage cannot of the exposure information (e.g., site specific or
be measured and expressed as a loss. The damage to a roof, for instance, can aggregate data at postal code resolution or lower),
be translated into monetary terms (the cost of repairs), which in turn can be and the details available for a given resolution
included in loss inventories.
(e.g., whether just occupancy is known or detailed
Loss inventories are tools of accountability and transparency for DRM. structural information is available). In addition, the
Despite their shortcomings (such as quality issues), they provide a process choice of whether to use a mean value or a sampled
for documenting a country’s disaster losses. Loss inventories establish an
value for damage depends on the details of a risk
historical baseline for monitoring the level of impact on a community or
analysis. A sampled value is generated using the
country. They make it possible to quantify the impact of individual hazards
so that communities can focus disaster risk reduction efforts on frequently mean and CV from the vulnerability function at the
occurring hazards rather than the last disaster. Inventories allow governments requisite hazard intensity. Other factors that can be
to allocate resources by community or by hazard—that is, to prioritize areas of incorporated into damage and loss estimates include
heightened risk (hot spots) or to focus on a particular hazard. when the structure was built, given that building
Loss information can also be harnessed for, and integrated into, risk practices and codes have changed over time, and the
assessments as part of efforts to promote community resilience. Loss and timing of an event, given that the use of a structure
hazard profiles can inform land-use planning, zoning, and development varies over the course of a day.
decisions; local ordinances on building codes and housing density; taxation
and budget decisions; and policy setting at local to national levels. A sound Often losses are adjusted for a variety of additional
understanding of the drivers and causes of losses, as well as their societal, factors, such as having to replace a structure if
environmental, and economic implications, enables communities to manage
damage exceeds a certain threshold; accounting
hazards and disasters proactively rather than reactively.
for business interruption costs for commercial or
Where loss inventories are consistently updated, the expanded historical industrial properties or additional living expenses
record provides the basis for temporal studies and trend analysis of losses.
for residential properties; incorporating the effects
High-quality loss data of good temporal and spatial resolutions can be coupled
of demand surge on large or sequential disasters;
with ancillary data like DRM expenditures or demographic information.
Combining these data makes it possible to evaluate the effectiveness of
and including damage to a structure’s contents. A
policies and to determine whether DRM expenditures are making a difference good overview of loss calculations is provided in
in loss trends, whether DRM efforts are effective, whether the mere presence section 3-18.
of more people is driving the rise in losses, and whether climate change is
affecting losses. Losses can be estimated ex ante and ex post.
Modelled losses often differ from observed losses
Source: Text is from Melanie Gall, Christopher T. Emrich, and Susan L. Cutter,
“Who Needs Loss Data,” input paper prepared for the 2015 Global Assessment for a variety of reasons. One reason is that modelled
Report on Disaster Risk Reduction, available at www.preventionweb.net/gar. losses represent only losses that are captured by the
model, and these losses depend upon the quality (in
terms of resolution and detail) of the exposure data.
Another reason is that loss inventories are typically
collected in an ad hoc manner. Better records of
disaster losses would provide a range of benefits
(see box 2-8).
Box 02—9 Incorporating Disaster Resilience into Cultural Heritage Buildings in Bhutan
Cultural heritage sites in Bhutan are considered “living” heritage sites because several programs and trainings have been conducted to proactively address
they continue to play an active role in the daily lives of the society. In addition disaster resilience in cultural heritage sites, and good construction guidelines
to their architectural, aesthetic, historical, and archaeological significance, have been formulated by the national government to help prevent or minimize
most of the cultural heritage sites in Bhutan have deep spiritual and cultural damage to cultural heritage sites during disaster events. A study of indigenous
significance. In Bhutan, sites are deemed to be part of the country’s cultural
construction practices, begun after the 2009 earthquake, has been ongoing,
heritage based on their use as religious and communal centers as well as
and hundreds of carpenters and masons in the affected districts have been
their antiquity.
trained in safe construction practices to facilitate reconstruction of the damaged
Disasters have physically affected Bhutan’s cultural heritage sites and have also cultural heritage buildings and rural houses.
disrupted centuries-old communal and social traditions. The great vulnerability
of Bhutan’s unique cultural heritage sites can be seen in the effect of events One positive and surprising outcome of this training program was the discovery
over the last 20 years, starting in 1994, when the Punakha Dzong (a huge that most of the local carpenters and masons already had the knowledge and
structure built as a fortress in the 17th century) was severely damaged by a skills needed for traditional—and more disaster-resilient—construction, though
glacial lake outburst flood, and continuing to 2009 and 2011, when earthquakes this knowledge had deteriorated over time as the traditional construction
damaged over 200 cultural heritages sites and thousands of rural dwellings. practices grew less popular and as the rapid completion of buildings was
made a priority. It also appeared that in the interest of saving time and
It was estimated that the physical loss of the structures—mainly lhakhangs
money, compromises were being made in the quality of materials as well as
(temples) and dzongs (fortresses)—was US$13.5 million for the 2009 earthquake
construction techniques, leaving structures even more vulnerable to disasters.
and US$6.96 million for the 2011 earthquake. These are large losses for a small
The safe construction training program has highlighted the importance
developing country. The actual loss, however, is much larger, since it goes
beyond the loss of the physical structures and includes the loss of interior of safety for both homeowners and builders during the post-earthquake
assets known as nangtens (paintings, sculptures, carvings, etc.). In many cases, reconstruction phase.
these were one of a kind and irreplaceable. Moreover, the loss to spiritual values
The government of Bhutan faces some clear challenges as it seeks to improve
and traditions brought about by such disasters cannot be estimated in terms of
monetary value. the understanding of disaster management and the resilience of cultural
heritage sites, with access to appropriate technical skills and financial resources
Bhutan has a variety of programs and policies in place designed to protect its to monitor and sustain the program posing the greatest challenge.
cultural heritage, but these have tended to be reactive rather than proactive.
There are signs that this reactive approach is beginning to change, however Source: Dechen Tshering (World Bank).
The historical record of loss mainly represents direct intangible losses are associated with assets that are
tangible losses produced by an event. Examples of difficult to value. Examples include the loss of a
direct tangible loss include damage to public and life, damage to ecosystem services, and damage to
private infrastructure, commercial and industrial sites related to cultural heritage. (Box 2-9 describes
facilities, dwellings, and the contents of a structure. efforts to increase the resilience of heritage sites in
The cost of business interruption and the expense of Bhutan.) A full consideration of all direct, indirect,
housing a structure’s inhabitants while a dwelling is and intangible losses would produce much higher
repaired or replaced are considered indirect losses. loss estimates than the more easily quantified and
Indirect losses generally arise from disruptions commonly seen records of direct loss.
in the flow of goods and services, though such
disruptions can produce positive as well as negative It can be difficult to anticipate and quantify the
impacts. An example of a positive impact would be potential for indirect losses despite their size. The
the increased demand for construction material. In 2011 Tohoku earthquake and tsunami in Japan and
contrast to tangible losses that are relatively easy flooding in Thailand offer an example of the global
to value, such as damage to structures or contents, indirect impacts from local events. The Japanese
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
61
62
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
tsunami was much more spectacular and had 2012b). The reason Thailand’s flooding had such
dramatic news coverage; however, the Thailand a dramatic impact on the global economy is that
floods caused much more damage to industrial industrial parks outside of Bangkok were a critical
supply chains on a global basis.
node in the global supply chain for the production
The 2011 Tohoku earthquake and tsunami slowed of automobiles and electronics (Haraguchi and Lall
the Japanese and global economies. For the full 2013).
year of 2011 the gross domestic product (GDP) of
Japan was 0.7 percent lower than in 2010 (Trésor- As box 2-8 suggests, collecting and analyzing
Economics 2012). The largest quarterly decline damage and loss data from previous disasters
(1.8 percent) occurred in the first quarter when provides valuable insight into the understanding
the earthquake and tsunami struck. There was a of physical, social, and economic vulnerability.
rebound in the third quarter followed by a decline Collecting information post-disaster can build
in the fourth quarter that was associated with
damage scenarios to inform planning processes,
the Thailand floods. On a global basis there was
assess the physical and financial impact of disasters,
negligible impact on full-year GDP because of a
rebound in the second half of 2011. In addition, develop preparedness measures, and facilitate
spending on public sector reconstruction resulted in dialogue for risk management. A number of global
a positive impact in 2012. and national disaster loss systems, some open and
some proprietary, record the losses associated
In contrast to the Japanese disaster, the 2011
with disasters; these are listed in table 2-3. For
flooding in Thailand was estimated to have reduced
global production by 2.5 percent (UNISDR 2012) more detailed information, see the United Nations
and reduced Thailand’s GDP growth rate from 4.0 Development Programme survey of loss databases
percent to an expected 2.9 percent (World Bank (UNDP 2013).
CHAPTER
02
DATABASE NAME DESCRIPTION DIRECT LINK
Regional Table 02—3
http://www.gripweb.org/ Sources of Disaster
Andean Information System for Disaster gripweb/?q=countries-risk-information/ Loss Data
http://www.siapad.net/
Prevention and Relief (SIAPAD) databases-information-systems/
andean-information-system-disaster
See countries at http://www.
DesInventar http://www.desinventar.org/
desinventar.org/en/database
CMC Nikolay Grigoryan (nik@
Armenia Emergency Management Stand alone
emergency.am)
http://www.emknowledge.gov.au/
Australia Disasters Database
disaster-information/
http://www.gripweb.org/
gripweb/?q=countries-risk-information/
Disaster Incidence Database (DIDB) of
databases-information-systems/ http://www.dmic.org.bd/didb
Bangladesh
disaster-incidence-database-didb-
bangladesh
http://www.publicsafety.gc.ca/cnt/
Canadian Disaster Database http://cdd.publicsafety.gc.ca/
rsrcs/cndn-dsstr-dtbs/index-eng.aspx
http://www.cdema.org/index.
Caribbean Disaster Events Database php?option=com_content&view=arti-
cle&id=110&Itemid=88
http://calamidatph.ndrrmc.gov.ph/dm/
Calamidat
web/
http://webra.cas.sc.edu/hvri/products/ http://webra.cas.sc.edu/hvriapps/
Sheldus
sheldus.aspx sheldus_web/sheldus_login.aspx
US Billion Dollar Weather/Climate http://www.ncdc.noaa.gov/billions/ http://www.ncdc.noaa.gov/billions/
Disasters overview events
http://www.gripweb.org/
gripweb/?q=countries-risk-information/
Damage and Needs Assessment http://www.ccfsc.gov.vn/KW6F2B34/
databases-information-systems/
system (DANA) of Vietnam Disaster-Database.aspx
damage-and-needs-assessment-
system-dana
Global
http://www.glidenumber.net/glide/ http://www.glidenumber.net/glide/
GLIDE
public/about.jsp public/search/search.jsp
EM-DAT http://www.emdat.be/about http://www.emdat.be/database
https://www.munichre.com/touch/ https://www.munichre.com/
NatCatSERVICE naturalhazards/en/natcatservice/ touch/portal/en/service/login.
default.aspx aspx?cookiequery=firstcall
Sigma http://www.swissre.com/sigma/ http://www.swissre.com/sigma/
http://catastropheinsight.aonbenfield. http://thoughtleadership.aonbenfield.
Aon Benfield
com/Pages/Home.aspx com
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
63
64
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
user interfaces (GUIs) to ease of installation, to
Hazard and Risk
user support and frequent updates, to capacity
Assessment Tools 19 for customization. It also highlights some of the
challenges that a user of a modelling tool might
Since 2005, the number of nonproprietary hazard
face, from difficulty with installation to poor
and risk modelling tools has grown rapidly as part
documentation and many other factors. It is
of the global movement to understand and manage
risk. These tools allow users to calculate risk and important to note that many modelling tools are
better understand, prepare for, and mitigate the frequently updated, so the challenges presented in
likely impact of potential disasters. this analysis may have been overcome with a recent
software update.
Given the plethora of tools available and the variety
of reasons for seeking to assess risk, users may The evaluation of software packages included the
find it challenging to choose the appropriate tool following steps:
for addressing the hazard, exposure, and/or risk
1. Evaluation criteria were developed for open
question under consideration and is aligned with
access software packages based on Daniell (2009)
their modelling and computational experience.
Some attempts have been made to evaluate the and through consultations.
many modelling tools that are available to users
2. A preliminary review of available open source
at no cost, but these efforts did not include in-
packages worldwide in the four peril types was
depth review or testing. Thus the evidence base to
undertaken. More than 80 software packages
differentiate tools for different purposes and end
were downloaded and initial checks made
uses has been lacking.
concerning availability, source code, active or
To address this gap and meet the need for inactive status, and so on.
a systematic review of tools against a set of
3. An initial multi-criteria analysis was undertaken
established criteria, the Global Facility for Disaster
Reduction and Recovery (GFDRR) and World Bank in order to select the packages to review in depth
undertook testing and evaluation of free hazard and for each peril.
risk modelling software using a consistent approach.
4. The 31 selected packages were installed and
The review considered over 80 open access
tested using tutorials, data sets, and examples
noncommerical software packages. A preliminary
in order to create outputs. This step included
analysis based on whether the 80 models were
noting advantages and disadvantages of these
currently supported was used to select a subset
software packages, and then filling out a
of eight earthquake models, four cyclone models,
detailed final set of about 180 criteria under 11
eleven flood models, and eight storm surge/tsunami
models for more detailed analysis. The detailed key classification themes (open source, GUI,
analysis evaluated the models on the basis of over software documentation, technology, exposure
100 criteria, and the results provide a synopsis of component, vulnerability, hazard, risk, post-event
key open access natural hazard risk modelling tools analysis, scenario planning, and output).
available worldwide.
A sample page of the review (for MAEvis/mHARP) is
This analysis highlights the strengths of different shown in figure 2-3. The review of every package is
modelling tools, from sophisticated graphical available in Daniell et al. (2014).
CHAPTER
02
SOFTWARE NAME PERIL LICENSE CURRENT VERSION OPEN SOURCE OPERATING SYSTEM
MAEviz Earthquake Single User V3.1.1 Build12 Yes, svn Win, Mac, Linux
Preferred Specific Information
SOFTWARE MODULES
CODING LANGUAGE MANUAL GUI HELP
(see below for more info)
Java using Eclipse RCP Many risk modules – NCSA GIS, Eclipse RCP, MAEviz. Yes Yes Yes
Goal of the Software
Another Hazus-based application, MAEviz (Mid-America Earthquakes system uses a combination of Sakai (an open source web portal), NEESgrid (a
Visualization) was developed to perform seismic risk assessment in the middle framework of tools to allow researchers to collaborate), and SAM (Scientific
U.S. states. At first glance, it seems specialized; however, its huge potential Annotation Middleware) in order to allow users to add their own hazard data. It
can be seen in the flowchart of analysis procedures (48 and counting) and is easily extendable; the European Union (EU) project SYNER-G, for example, has
its complete Hazus system, including detailed algorithms. The visually driven added a large fragility function manager to it, in addition to other tools.
File Types Used
HAZARD VULNERABILITY EXPOSURE KEY HAZARD METRICS
Spectral ordinates are used in terms of PGA and Sa. This is calculated using GMPEs and
.txt, .csv .xml *.shp
source-site distance, source geometry, and seismicity.
Description of Software Risk Outputs
Damage estimates include options for multiple mitigation strategies, testing of scientific and engineering
principles, and estimating the earthquake hazard impact on lifelines and social or economic systems (based
on Hazus and extra analysis).
The outputs are economic losses (direct, indirect, downtime, business interruption), social losses (social
vulnerability, fatalities, injuries, homeless), and management options. A detailed list of the modules is
shown in the appendix. Simple reports and data views are given. The software creates all scenario output
(disaggregated or not).
An overview of the MAEviz options (McLaren
et al. 2008)
Advantages and Disadvantages Recommendations for Improvements
for Greater Utility
• It is completely open source and features inbuilt GIS; the software is well
formatted with the GIS user interfaces. mHARP will give this fantastic software an additional use. It should be
• Is easily the best software for scenario risk assessment and decision support integrated with Deltares or other risk software, given the common structure.
(mitigation, benefit-cost). It has already been integrated in HAZturk and SYNER-G. A combination with
EQRM for probabilistic modelling would be useful. An InaSAFE-style command
• It has an outstanding array of modules that provide end analysis such as
system could simplify the software even further for the most basic users, but it
shelter needs or business interruption.
is currently fairly user-friendly.
• There is a developer and community, and the function codes are easy to read
and improve.
• Basic users find it easy to use; the large array of infrastructure types can be
used for hazard and loss.
• Combining detailed hazard, detailed vulnerability, and management and risk
modelling, the software is easily extendable.
• It is currently tuned only for deterministic analysis.
Figure 02—3 Sample software package review.
Source: Daniell et al. 2014.
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
65
66
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
The information generated with this assessment can purposes. Users at all levels should understand the
aid users in selecting suitable software packages. sensitivity of models to changes in inputs and would
It is highly recommended that users test as many probably benefit from training, as box 2-10 suggests.
packages as possible in order to make an informed (For case studies that demonstrate the importance
decision about which software is right for their of training, see sections 3-9 and 3-12).
Box 02—10 Training in Use of Risk Models: The GEM Perspective
While specific risk modelling software packages may be more or less The Global Earthquake Model has developed a variety of approaches to
appropriate depending on the experience level of the end-user, users at any training users in its tools. It holds workshops targeted to users at the same
level may benefit from training. It is important for users of hazard and risk level of experience and education, and it hosts professionals at the GEM
secretariat for hands-on training that may last for weeks or months. For
models to understand the sensitivity of the models they are using and to be
local experts in developing countries, GEM has found that “learning by
aware of the large impact on assessment results that changes in the input
doing” has been the most effective way to gain necessary skills and to
parameters can have. The figure shows that the OpenQuake engine may
develop needed capacity. Offering training of this type requires a few years
produce two different hazard maps for Japan depending on the user-defined of ongoing engagement and is possible only through strong partnerships
modelling decisions (in this case related to the probability of a Tohoku-like at both the institutional and individual levels. Through its Earthquake Model
event occurring in the next 50 years). for the Middle East (EMME) project, for example, GEM offered local technical
experts their first exposure to probabilistic earthquake modelling. Although
Training could be most beneficial in governments of developing countries, hazard and risk assessments might initially develop more slowly under the
where capacity in conducting seismic hazard and risk assessment, using “learning by doing” approach, the newly built local capacity for maintaining,
probabilistic modelling, and understanding results tends to be especially low understanding, and advising governments is invaluable.
and sporadic. But even governments in developed countries need access to Source: Helen Crowley, Nicole Keller, Sahar Safaie, and Kate Stillwell
technical advice, including the expertise of their own specialists. (GEM Foundation).
CHAPTER
02
Box 02—11 The Understanding Risk Community
Understanding Risk (UR) is an open and global community of experts As a result of the 2012 UR Forum, participatory mapping projects have been
and practitioners in the field of disaster risk assessment. UR community
implemented in Nepal and Malawi, and open geospatial data platforms have
members include representatives of government agencies, the private sector,
been launched in the Horn of Africa, Haiti, and Sri Lanka. The 2012 forum also
multilateral organizations, nongovernmental organizations, community-
lead to the first national UR event, held in Brazil in November 2012. This event
based organizations, research institutions, and academia. Every two years,
brought together Brazilian experts and practitioners to discuss the challenges
the Global Facility for Disaster Reduction and Recovery convenes the UR
the country faces in understanding its disaster risk and to raise the profile of
Forum—a five-day event designed to showcase best practices and the
the topic nationally. In May 2014, Haiti will hold a national UR Forum to bring
latest technical know-how in disaster risk assessment. The forums provide
together nontraditional partners and tackle the challenge of economic, social,
organizations with the opportunity to highlight new activities and initiatives,
and environmental vulnerability in the country.
build new partnerships, and foster advances in the field.
The next global UR Forum, in London between June 30 and July 4, 2014,
The first UR Forum, held in Washington, DC, in June 2010, was attended by
500 practitioners representing 41 countries. The goal of the forum was to takes “Producing Actionable Information” as its theme; it will focus on how to
showcase progress in the field of disaster risk assessment and to promote translate and communicate scientific information into actionable decisions
the sharing of ideas and the exchange of knowledge through a series of on the ground. UR 2014 will continue to foster the growth of partnerships
technical sessions led by experts. During the forum, the GEM held its annual and spur the advances in risk assessment needed for achieving sustainable
outreach meeting, and Random Hacks of Kindness (RHoK)—a group that development and building resilience.
brings together software programmers to develop applications for DRM
challenges—organized its first global hackathon.(A) Based on the success of The UR Forums are clearly meeting a need. Participants report that the mix
the forum, the UR series was launched. of backgrounds, interests, and types of expertise they encounter, along with
the opportunity to share ideas and information, stimulate their thinking and
UR 2012, held in Cape Town from July 2 to July 6, was attended by 500 risk promote creative solutions to problems. Discussions taking place at the
assessment experts from more than 86 countries. The forum showcased new forums are being shared beyond the UR community by means of a post-
tools for decision makers, strengthened regional and global partnerships, and conference publication (Understanding Risk: Best Practices in Disaster Risk
built technical capacity in the Africa region through a series of training events.
Assessment). The UR community website (www.understandrisk.org) also
UR 2012 was also a testimony to the tremendous progress in understanding
serves as a platform for incubating innovation and forging partnerships in the
risk since 2010: crowdsourcing, a new topic in 2010, by 2012 was being
disaster risk assessment field. Membership in the community has grown from
mainstreamed and used to support risk assessment for financial applications
about 1,000 in 2010 to more than 3,000 in 2014.
intended to make governments, businesses, and households more financially
resilient to risk. A consensus about the need for data that are more open also (A) RHoK is a partnership of Google, Microsoft, Yahoo, the National
emerged, with many initiatives demonstrating that the trend toward open Aeronautics and Space Administration (NASA), and the World Bank. See the
data would be broadly beneficial. The forum also highlighted new tools and
website at http://www.rhok.org/.
methodologies for building resilience, and in particular called attention to the
extent to which these tools are now available to nonspecialists. Source: Emma Phillips (GFDRR).
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
67
68
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
The Global Earthquake Model suggests some
Creating Platforms and
of the benefits that arise when developing and
Partnerships to Enable applying knowledge is treated as a cooperative
endeavor.20 GEM was created specifically as a
the Development of public-private partnership because its founders
Risk Assessments judged that structure to be optimal for its purposes.
They recognized that risk holders reside in both
The move to collect, analyze, and produce risk sectors; that advocacy, models, and information are
information for current and future climates is necessary for mitigating earthquake risk; that the
gaining momentum among various actors at project could achieve its goals only by combining
various levels, from the individual to the global. funds from both sectors; and that the involvement
One consequence of this trend is a growing need of both sectors would lend the project credibility
for all actors involved with risk to cooperate, and momentum. GEM’s formal partners include
communicate, and form partnerships across 13 private companies, 15 public organizations
geographic, institutional, and disciplinary representing nations, and 9 international
boundaries. Fortunately, much progress has been organizations. Various other associate participants
and organizational members of international
made in this regard.
consortia also deliver global projects.
The recognition that cooperation and partnership
One notable aspect of GEM as a public-private
are crucial for building resiliency motivated the
partnership is its success in unifying diverse
formation in 2010 of the Understanding Risk
perspectives under a common interest. The
community, whose more than 3,000 members span
partnership works because both sectors seek
the globe and include experts and practitioners
the same outcome: credible, accessible risk
across many professions and disciplines (see box
information that is widely used and understood.
2-11 for more detail). Information sharing is critical At the same time, the two sectors have somewhat
to this community, which meets every two years to different focuses. Private sector partners generally
discuss best practices and promising innovations seek to reduce future financial losses (through
in disaster risk assessment and to give members an strict building codes and through open data that
opportunity to build and strengthen partnerships ensure common expectations of loss); to create
and spur further innovations. new markets for insurance products (requiring
worldwide intercomparable loss data and accessible Box 02—12 Willis Research Network
risk information); and to build customer demand The Willis Research Network was launched in 2006 to better integrate
(through increased engagement among trusted science, insurance, and resilience.(A) Starting with a partnership of seven UK
universities, the network has now grown to include more than 50 international
local experts and increased understanding of risk
research institutions, making the Willis Research Network one of the world’s
by the public). Public sector partners, including largest collaborations between science and the financial sector.
nongovernmental organizations, seek to reduce The network’s research program is organized across four pillars: economic
future casualties, economic loss, and disruptions capital and enterprise risk management; natural hazard and risk; man-made
liability risks; and core technologies and methods. A focus on accurately
(through DRM and land-use policies and retrofitting
quantifying natural hazard risk is a priority for Willis Re and the insurance
of public buildings); to implement policy (requiring sector as a whole, given that the solvency capital of most non-life insurance
companies is strongly influenced by their exposure to natural catastrophe risk.
broad awareness of risk and hence accessible data);
to base decisions on scientifically defensible hazard Research supported by the network has resulted in hundreds of peer-reviewed
academic articles; it has also led to improved insurance sector models,
and risk estimates; and to reduce the need for
methodologies, and transactions that enable the financial market to better
post-disaster aid (requiring free, open information understand and cover risk. Moreover, by openly sharing research findings,
the network has made it possible for other private and public institutions to
to support markets for financial risk transfer
improve their efforts to identify, evaluate, and manage disaster risk.
mechanisms and lower losses as a result of risk
The Willis Research Network’s principles and practices—its clear articulation
reduction).
of critical research requirements, its protection of academic and scientific
independence, and its recognition of the time frames consistent with academic
The perspectives and positions of the two sectors achievement—explain its ability to catalyze improvements in risk assessment,
do not differ as widely as GEM’s founders initially and exemplify the strengths of academic and private sector partnerships.
anticipated. In practice, differences in perspective (A) The network was formed to support the academic and analysis focus of
varied within each sector as much as or more than Willis Group Holdings.
they did across sectors. Source: Willis Research Network website (www.willisresearchnetwork.com),
©Willis Group Holdings. Used with permission; further permission required for
reuse.
Yet another collaboration that aims to build better
risk information is the Willis Research Network,
which links more than 50 international research
institutions to the expertise of the financial and
insurance sector in order to support scientists’
quantification of natural hazard risk. More detail on
the network is in box 2-12. For an account of another
kind of collaboration—one in which scientists,
engineers, and developers of building codes
collaborated with officials in planning, governance,
and public service to promote a more earthquake-
resilient city—see the account of participatory
earthquake risk assessment in Dhaka in box 2-13.
69
70
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
Box 02—13 Participatory Earthquake Risk Assessment in Dhaka
While Bangladesh can rightfully claim major accomplishments in flood and project and assessed the collective progress toward achieving project goals.
cyclone risk reduction, its urban earthquake risk has not been adequately Participants in the project were assigned to one of three groups depending
considered. Bangladesh lies on the seismically active northeastern Indian plate, on their job and type of expertise: a focus group, an advisory committee, or
which is subject to moderate- to large-magnitude earthquakes. The nearest a scientific consortium. Focus group members included representatives from
major fault line is believed to run less than 60km from the capital city of Dhaka. key national and local organizations involved in planning or in developing
Research suggests that an earthquake of up to magnitude 7.5 is possible in the and implementing construction codes; therefore their role involved engaging
area. Earthquake risk in Bangladesh is increasing with rapid and uncontrolled in data collection, analysis, and validation. The advisory committee is made
urbanization, particularly in and around Dhaka, which with 26,000 residents per up of policy makers and decision makers from various government and
square kilometer is one of the world’s densest cities.(A) nongovernment institutions who provide overall guidance and oversight to
project participants. The scientific consortium is made up of local experts
There has been no major earthquake in living memory, which has frustrated in earthquake engineering, geology and geophysics, land use and regional
efforts to build consensus around the need to invest in measures to increase planning, DRM, law and business administration, environmental management,
urban resilience to earthquake. Moreover, the governance of cities in and other closely related fields; collectively they provide guidance on
Bangladesh, particularly Dhaka, is very complex. Responsibility for urban scientific and technical matters.
planning, governance, and public service provision is spread out across many
Next steps include the development of multiyear process that will develop
different agencies. Agencies’ roles are not clear and often overlap. Moreover,
several decision-making tools for mitigating the impact of earthquake
political affiliations can affect capacity to implement policy and govern the
hazards by reducing structural and nonstructural vulnerability. Diverse
city. Thus any initiative intended to address Dhaka’s vulnerability to earthquake
working groups will mobilize resources and implement the project; existing
required engagement with multiple stakeholders and a common understanding
earthquake hazard and vulnerability data will be compiled; a uniform
of risk.
data platform will be developed; and an information, education, and
A participatory earthquake risk assessment over the last two years in communication program will be established. Building on this foundation, the
Bangladesh(B) has successfully built consensus on disaster risk across project will produce (a) an earthquake hazard, vulnerability, and risk analysis;
agencies, institutions, and technical experts in their pursuit of earthquake (b) an assessment of legal and institutional arrangements; and (c) a guide to
risk reduction and is now being leveraged to develop specif ic investments incorporating earthquake risk management into land-use planning.
to enhance urban resilience. The program has increased the collective
(A) Data are for Dhaka City Corporation; if the entire Dhaka Metropolitan Area
understanding of risk, promoted collaboration in identifying major disincentives
is taken into account, Dhaka’s population density is 13,500 residents per
for resilient development, supported planning for prevention, and has
square kilometer (World Bank 2012a).
gradually shifted the country toward a more proactive approach to resilient
development. (B) The assessment is called the Bangladesh Earthquake Risk Mitigation
Program and is a World Bank program supported by the GFDRR.
A successful aspect of this program involved ensuring that stakeholders
from over 40 different agencies working in Dhaka guided each step of the Source: Swarna Kazi (World Bank).
CHAPTER
02
Endnotes References
12 A 100-year event represents something with a
Adams, B., and C. Huyck. 2006. “Remote Sensing
Technologies for Disaster Emergency Response: A
probability of occurrence equal to 0.01 per year. In general,
Review.” In Acceptable Risk Processes: Lifelines and
an X-year event has a 1/X probability of occurrence per
Natural Hazards, edited by Craig E. Taylor and Erik
year. The number of years represented by X is termed the Vanmarcke. New York: ASCE Publications.
“X-year return period.”
Allen, T. I., D. J. Wald, P. S. Earle, K. D. Marano, A. J. Hotovec,
13 Information on the moment tensors for all earthquakes
K. Lin, and M. G. Hearne. 2009. “An Atlas of ShakeMaps
globally with moment magnitudes greater than 5 can be and Population Exposure Catalog for Earthquake Loss
Modeling.” Bulletin of Earthquake Engineering 7: 701–18.
obtained through the Global Centroid-Moment-Tensor
(CMT) Project (http://www.globalcmt.org). Best-track ATC (Applied Technology Council). 1985. ATC-13: Earthquake
information for tropical cyclones includes the location Damage Evaluation Data for California. Washington, DC:
(latitude and longitude), central pressure, and maximum ATC.
sustained wind at six-hour intervals for all tropical
Bal, Ihsan Engin, Julian J. Bommer, Peter J. Stafford,
cyclones. A collection of these data from a variety of
Helen Crowley, and Rui Pinho. 2010. “The Influence of
sources can be obtained from the IBTrACS archive (http:// Geographical Resolution of Urban Exposure Data in an
www.ncdc.noaa.gov/ibtracs/). Earthquake Loss Model for Istanbul.” Earthquake Spectra
26 (3): 619–34.
14 For more on LiDAR (Light Detection and Ranging), see
Bhaduri, B., E. Bright, P. Coleman, and M. L. Urban. 2007.
the National Oceanic and Atmospheric Administration
“LandScan USA: A High-resolution Geospatial and
website at http://oceanservice.noaa.gov/facts/lidar.html.
Temporal Modeling Approach for Population Distribution
and Dynamics.” GeoJournal 69: 103–17.
15 For more information, see Geoscience Australia, “New
Building Assessment Tool Supports Better Risk Analyis,” Boyd, E., M. Levitan, and I. van Heerden. 2010.
February 12, 2014, http://www.ga.gov.au/about-us/ “Improvements in Flood Fatality Estimation Techniques
news-media/news-2014/new-building-assessment-tool- Based on Flood Depths.” In Wind Storm and Storm
Surge Mitigation, edited by Nasim Uddin, 126–39. ADCE
supports-better-risk-analysis.html.
Libraries. doi:10.1061/9780784410813.ch11.
16 For a more in-depth discussion of how climate extremes
Brinkman, Jan Jaap, and Marco Hartman. 2008. “Jakarta
may change in the future, see IPCC (2012). Flood Hazard Mapping Framework.”World Bank.
(unpublished). http://www.hkv.nl/documenten/Jakarta_
17 The discussion of exposure here draws heavily on two
Flood_Hazard_Mapping_Framework_MH.pdf.
GAR15 input papers, both available at www.preventionweb.
Cardona, O. D., M. G. Ordaz, E. Reinoso, L. E. Yamin, and A.
net/gar: Massimiliano Pittore, Marc Wieland, and Kevin
H. Barbat. 2012. “CAPRA–Comprehensive Approach to
Fleming, “From Remote Sensing to Crowdsourcing:
Probabilistic Risk Assessment: International Initiative
Perspectives of a Global, Dynamic Exposure Model for for Risk Management Effectiveness.” Proceedings of
Georisk Assessment” and Rashmin Gunasekera, Christoph the 15th World Conference on Earthquake Engineering.
Aubrecht, Oscar Ishizawa, and Sergio Freire, “Global Lisbon, Portugal. http://www.iitk.ac.in/nicee/wcee/
Exposure Modelling: An End-User Perspective.” article/WCEE2012_0726.pdf.
Chapman, K. 2012. “Community Mapping for Exposure in
18 Many global exposure models make use of commercial
Indonesia.” Humanitarian OpenStreetMap Team. http://
available data sets such as LandScan (http://web.ornl.gov/ hot.openstreetmap.org/sites/default/files/CM4E-Indo-
sci/landscan/) and as a result the final exposure model en.pdf.
may not be completely open.
Chen, Keping, John McAneney, Russell Blong, Roy Leigh,
19 This section provides an overview of the results in
Laraine Hunter, and Christina Magill. 2004. “Defining
Area at Risk and Its Effect in Catastrophe Loss
Daniell (2014).
Estimation: A Dasymetric Mapping Approach.” Applied
Geography 24 (2): 97–117.
20 This account of GEM’s institutional structure was
provided by Helen Crowley, Nicole Keller, Sahar Safaie, and Coburn, A., and R. Spence. 2002. Earthquake Protection.
Kate Stillwell of GEM. Chichester, UK: Wiley.
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
71
72
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
Coburn A. W., R. J. S. Spence, and A. Pomonis. 1992. “Factors Literature Review towards a Roadmap.” Natural Hazards
Determining Human Casualty Levels in Earthquakes: 68 (1): 1–42.
Mortality Prediction in Building Collapse.” Proceedings
of the First International Forum on Earthquake-Related GFDRR (Global Facility for Disaster Reduction and
Casualties. Madrid, Spain. Recovery). 2011. “World Bank Distance Learning Natural
Disaster Risk Management Program.” World Bank and
Daniell, James, Alanna Simpson, Rick Murnane, Annegien GFDRR. https://www.gfdrr.org/sites/gfdrr.org/files/
Tijssen, Ariel Nunez, Vivien Deparday Rashmin publication/WB_Risk_Analysis.pdf.
Gunasekera, Abigail Baca, Oscar Ishizawa, and Andreas
Schäfer. 2014. Review of Open Source and Open Access Griffin, J. G., H. Latief, W. Kongko, S. Harig, N. Horspool,
Software Packages Available to Quantify Risk from R. Hanung, A. Rojali, N. Maher, L. Fountain, A. Fuchs, J.
Natural Hazards. Washington, DC: World Bank and Hossen, S. Upi, S. E. Dewanto, and P. R. Cummins. 2012.
Global Facility for Disaster Reduction and Recovery.` “An Evaluation of Onshore Digital Elevation Models for
Tsunami Inundation Modelling.” American Geophysical
Daniell, J. E. 2009. “Open Source Procedure for Assessment Union Fall Meeting Abstracts. http://fallmeeting.
of Loss Using Global Earthquake Modelling (OPAL agu.org/2012/files/2012/12/Griffin-et-al-AGU-2012-
Project).” CEDIM Earthquake Loss Estimation Series, NH21C-1598.pdf.
Research Report No. 09-01. CEDIM. Karlsruhe, Germany.
Gunasekera, R., O. A. Ishizawa, C. Aubrecht, G. Pita, A.
De Bono, A. 2013. “The Global Exposure Database for GAR Pomonis, K. Fane, S. Murray, and B. Blankespoor. 2014.
2013.” Background paper prepared for the 2013 Global “Developing an Adaptive Exposure Model to Support
Assessment Report on Disaster Risk Reduction, UNISDR, the Generation of Country Disaster Risk Profiles.”
Geneva. www.preventionweb.net/gar. Geophysical Research Abstracts 16: EGU2014-16168.
Dell’Acqua, F., P. Gamba, and K. Jaiswal. 2012. “Spatial Haraguchi, Masahiko, and Upmanu Lall. 2013. “Flood Risks
Aspects of Building and Population Exposure Data and Impacts: Future Research Questions and Implication
and Their Implications for Global Earthquake Exposure to Private Investment Decision-Making for Supply Chain
Modeling.” Natural Hazards 68: 1291–1309. Networks.” Background Paper prepared for the 2013
Global Assessment Report on Disaster Risk Reduction,
Dilley, M. 2005. Natural Disaster Hotspots: A Global Risk
UNISDR, Geneva.
Analysis. Washington, DC: World Bank.
IPCC (Intergovernmental Panel on Climate Change). 2012.
Ehrlich, D., and P. Tenerelli. 2013. “Optical Satellite Imagery
Managing the Risks of Extreme Events and Disasters to
for Quantifying Spatio-temporal Dimension of Physical
Advance Climate Change Adaptation. A special report
Exposure in Disaster Risk Assessments.” Natural
of Working Groups I and II of the Intergovernmental
Hazards 68 (3): 1271–89.
Panel on Climate Change. Cambridge and New York:
Elsner, J. B., and T. H. Jagger. 2006. “Prediction Models for Cambridge University Press.
Annual U.S. Hurricane Counts.” Journal of Climate 19:
Jaiswal, K. S., and D. J. Wald. 2010a. “Development of a
2935–52.
Semi-empirical Loss Model within the USGS Prompt
Esch, T., M. Thiel, A. Schenk, A. Roth, A. Müller, and S. Dech. Assessment of Global Earthquakes for Response
2010. “Delineation of Urban Footprints from TerraSAR-X (PAGER) System.” Proceedings of 9th US and 10th
Data by Analyzing Speckle Characteristics and Intensity Canadian Conference on Earthquake Engineering:
Information.” IEEE Transactions on Geoscience and Reaching Beyond Borders. Ottawa, Canada.
Remote Sensing 48: 900–916.
Jaiswal, Kishor, and David Wald. 2010b. “An Empirical Model
European Commission. 2010. “Risk Assessment and for Global Earthquake Fatality Estimation.” Earthquake
Mapping Guidelines for Disaster Management.” Spectra 26 (4): 1017–37.
Commission Staff Working Paper, SEC(2010) 1626 final,
European Commission, Brussels, Belgium. Jaiswal, K. S., D. J. Wald, and K. A. Porter. 2010. “A Global
Building Inventory for Earthquake Loss Estimation and
FEMA (Federal Emergency Management Agency). 2002. Risk Management.” Earthquake Spectra 26 (3): 731–48.
Rapid Visual Screening of Buildings for Potential Seismic
Hazards: A Handbook. FEMA 154. 2nd ed. https://www. Lavakare, Ajay, and Kenna Mawk. 2008. “Exposure
fema.gov/media-library/assets/documents/15212. Data Quality.” In The Review: A Guide to Catastrophe
Modelling, 12–14. London: Informa. http://static.rms.
Flay, S., and J. Nott. 2007. “Effect of ENSO on Queensland com/email/documents/liferisks/collateral/a-guide-to-
Seasonal Landfalling Tropical Cyclone Activity.” catastrophe-modeling-the-review.pdf.
International Journal of Climatology 27: 1327–34.
doi:10.1002/joc.1447. Lu, X., E. Wetter, N. Bharti, A. J. Tatem, and L. Bengtsson.
2013. “Approaching the Limit of Predictability in Human
Geiß, C., and H. Taubenböck. 2013. “Remote Sensing Mobility.” Nature Scientific Reports 3: Article no. 2923.
Contributing to Assess Earthquake Risk: From a doi:10.1038/srep02923.
CHAPTER
02
Martin-Ortega, J., and A. Markandya. 2009. “The Costs UNDP (United Nations Development Programme). 2013.
of Drought: The Exceptional 2007–2008 Case of “A Comparative Review of Country-level and Regional
Barcelona.” Basque Centre for Climate Change, Bilbao, Disaster Loss and Damage Databases.” Bureau for Crisis
Spain. Prevention and Recovery, UNDP. http://www.undp.org/
content/undp/en/home/librarypage/crisis-prevention-
McLaren, T. 2008. “MAEVIZ.” Presentation at EclipseCON, and-recovery/loss-and-damage-database/.
Santa Clara, CA, March 17–20.
UNISDR (United Nations Office for Disaster Risk Reduction).
Mondal, P., and A. J. Tatem. 2012. “Uncertainties in 2009. UNISDR Terminology on Disaster Risk Reduction.
Measuring Populations Potentially Impacted by Sea Geneva: UNISDR. http://www.preventionweb.net/
Level Rise and Coastal Flooding.” PloS ONE. doi:10.1371/ files/7817_UNISDRTerminologyEnglish.pdf.
journal.pone.0048191.
———. 2012. “Towards a Post-2015 Framework for Disaster
Müller, A., J. Reiter, and U. Weiland. 2011. “Assessment of Risk Reduction.” http://www.unisdr.org/we/inform/
Urban Vulnerability towards Floods Using an Indicator- publications/25129.
based Approach: A Case Study for Santiago de Chile.”
Natural Hazards and Earth System Science 11: 2107–23. UN (United Nations) Statistical Division. 2008. Principles
and Recommendations for Population and Housing
Peduzzi, P., H. Dao, C. Herold, and F. Mouton. 2009. Censuses. New York: United Nations Publications.
“Assessing Global Exposure and Vulnerability towards
Natural Hazards: The Disaster Risk Index.” Natural Van Westen, C. J. 2012. “Remote Sensing and GIS for
Hazards and Earth System Science 9: 1149–59. Natural Hazards Assessment and Disaster Risk
Management.” In Application of Space Technology for
Pesaresi, M., and M. Halkia. 2012. Global Human Settlement Disaster Risk Reduction: International Training Course
Layer Urban Atlas Integration: Feasibility Report. Lecture Notes, 307–75. Dehradun, India: Indian Institute
Luxembourg: Joint Research Centre European Union. of Remote Sensing and Centre for Space Science and
Technology Education in Asia and the Pacific.
Pittore, M., and M. Wieland. 2013. “Toward a Rapid
Probabilistic Seismic Vulnerability Assessment Using Wald, D. J., P. S. Earle, T. I. Allen, K. Jaiswal, K. Porter, and
Satellite and Ground-based Remote Sensing.” Natural M. Hearne. 2008. “Development of the US Geological
Hazards 68: 1–31. Survey’s PAGER System (Prompt Assessment of Global
Earthquakes for Response).” Proceedings of the 14th
Potere, D., A. Schneider, S. Angel, and D. Civco. 2009. World Conference on Earthquake Engineering. Beijing,
“Mapping Urban Areas on a Global Scale: Which of China.
the Eight Maps Now Available Is More Accurate?”
International Journal of Remote Sensing 30: 6531–58. Wesolowski, A., C. O. Buckee, D. K. Pindolia, N. Eagle, D.
L. Smith, A. J. Garcia, and A. J. Tatem. 2013. “The Use
Scawthorn, C. 1997. “Fires Following the Northridge and of Census Migration Data to Approximate Human
Kobe Earthquakes.” In Thirteenth Meeting of the UJNR Movement Patterns across Temporal Scales.” PLoS ONE.
Panel on Fire Research and Safety, March 13-20, 1996, doi:10.1371/journal.pone.0052971.
vol. 2, edited by K. A. Beall, 325–35. NISTIR 6030.
National Institute of Standards and Technology. World Bank. 2012a. Bangladesh: Towards Accelerated,
Inclusive, and Sustainable Growth—Opportunities and
Spence, Robin, Julian Bommer, Domenico Del Re, Juliet Challenges. Vol. 2. Washington, DC: Poverty Reduction
Bird, Nuray Aydinog Lu, and Shigeko Tabuchi. 2003. and Economic Management Sector Unit, South Asia
“Comparing Loss Estimation with Observed Damage: A Region, World Bank.
Study of the 1999 Kocaeli Earthquake in Turkey.” Bulletin
of Earthquake Engineering 1: 83–113. ———. 2012b. “Thai Flood 2011: Rapid Assessment for
Resilient Recovery and Reconstruction Planning.” World
Taubenböck, H., T. Esch, A. Felbier, M. Wiesner, A. Roth, and Bank, Washington, DC.
S. Dech, S. 2012. “Monitoring Urbanization in Mega Cities
from Space.” Remote Sensing of Environment 117: 162–76. Wu, M. C., W. L. Chang, and W. M. Leung. 2004. “Impacts of
El Niño–Southern Oscillation Events on Tropical Cyclone
Trésor-Economics. 2012. “The Impact of Japan’s Earthquake Landfalling Activity in the Western North Pacific.” Journal
on the Global Economy.” No. 100. April. http://www. of Climate 17: 1419–28. doi:http://dx.doi.org/10.1175/1520-
tresor.economie.gouv.fr/File/371985. 0442(2004)017<1419:IOENOE>2.0.CO;2.
Tsuji, Y., H. Matsutomi, F. Imamura, M. Takeo, Y. Kawata, M. Wyss, M., S. Tolis, P. Rosset, and F. Pacchiani. 2013.
Matsuyama, T. Takahashi, and P. Harjadi. 1995. “Damage “Approximate Model for Worldwide Building Stock in
to Coastal Villages Due to the 1992 Flores Island Three Size Categories of Settlements.” Background
Earthquake Tsunami.” Pure and Applied Geophysics 144 paper prepared for the 2013 Global Assessment Report
(3): 481–524. on Disaster Risk Reduction, UNISDR, Geneva.
PROGRESS, ACHIEVEMENTS, & REMAINING CHALLENGES
73
74
CHAPTER
03
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
CASE STUDIES
Highlighting Emerging
Best Practices
D emonstrated success is one of the best ways
to illustrate the benefits associated with
risk assessment and show how emerging efforts
are roughly grouped into those focused on data;
those focused on modelling; those that describe
specific risk assessment projects; those that focus
can contribute to further success. This section on participation, collaboration, and communication;
reviews a variety of case studies describing and those that address the future of risk. Given
ongoing and emerging open efforts that support that many case studies speak to some or all of these
risk assessments and successful examples of aspects, however, there is a fair amount of overlap
completed risk assessments. The contributions across categories.
Case Study Color Key
Data for Risk
Modelling Developments
Risk Assessment Case Studies
Participation, Collaboration, and Communication
Future of Risk
75
76
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-1. Open Data for Resilience Initiative (OpenDRI)
John Crowley, Vivien Deparday (GFDRR); Robert Soden, Abigail Baca,
Ariel Nunez (World Bank)
Risk assessments never start from a blank slate; partnership of institutions that was launched by
instead they build on existing data, analysis, and the Global Facility for Disaster Reduction and
historical experience. All too frequently, the data Recovery (GFDRR) and the World Bank in 2010,
sets that are required are incomplete, out-of-date, and designed to make data available to those who
and ill-suited to the analysis required. Moreover, need information about disaster risks in order to
data are often in forms that prevent them from make decisions. OpenDRI offers governments and
being shared widely, and they therefore remain
their partners a process for cataloging their existing
latent and inaccessible (even across ministries and
stocks of data and placing certain types of data
municipalities within the same country). Some
under open licenses that still enable ministries to
are blocked by technologies that lock data into
retain stewardship. The initiative also offers an
proprietary ecosystems. Most are stoppered by
policies that prevent release beyond small groups or
are simply fragmented into bureaucratic silos that
require significant investment to assemble back into
a whole picture.
Yet even fusing these existing data stocks into
a usable form is not enough, as the data need to
capture a dynamic reality. Rapid urbanization, CARIBBEAN GEONODES
population growth, and increasingly climate change http://www.dominode.net/
mean that the analysis of the potential impacts of http://haitidata.org/
natural hazards needs to updated more frequently http://geonode.data.gove.ag/
and at higher resolutions than ever before. In a time http://sling.gosl.gov.lc/
of economic hardship and unequal globalization,
few governments possess the resources to collate
existing data or collect new data, or to analyze data
COLUMBIA
and communicate the results to decision makers
http://geonode.columbiassh.org/
able to implement projects that get ahead of the
disaster cycle.
Because individual governments may not currently BOLIVIA
have the capacity to take on this work, however, http://geosinager.defensacivil.gob.bo/
does not mean that it cannot be accomplished.
The task of stewarding data about shared risks
should be understood as a collective effort, one
engaging governments, civil society, industry, and
individuals. That understanding is behind the Open
Data for Resilience Initiative (OpenDRI), a growing
CHAPTER
03
inexpensive method of engaging at-risk communities practices in the service of more effective disaster
in the process of mapping about their changing risk management (DRM) and climate change
exposure to natural hazards. Finally, it offers a way adaptation.
to build ecosystems of entrepreneurs, researchers,
and international institutions around data that a OpenDRI projects offer a menu of approaches for
nation manages for itself. building and using risk data and information:
The OpenDRI approach to managing risk data. • Collation and sharing of data and
Since 2010, the GFDRR has worked with the World information through open geospatial
Bank to implement OpenDRI in over 20 countries, catalogs. Here local partners are supported to
including Indonesia, Haiti, Nepal, Sri Lanka, and identify, prepare, and release existing hazard,
Malawi. The program is designed to build the exposure, and risk data via an online geospatial
necessary data for quantifying and mapping risk catalog. Recognizing a need to move away from
and for communicating the results to a wide range proprietary software platforms, GFDRR and
of decision makers at various levels, from national the World Bank have been active in leading and
to community. The OpenDRI team works with developing the open source platform GeoNode
governments to harness the value of open data (http://geonode.org/), which provides tools that
Figure 03—1
Examples of locations
of GeoNodes
supported by
the World Bank
KYRGYZSTAN and GFDRR.
http://geonode.caiag.kg/
Source: World Bank and
GFDRR.
PAKISTAN
HORN OF AFRICA http://disasterinfo.gov.pk/
http://horn.rcmrd.org/
SRI LANKA TYPHOON YOLANDA
http://riskinfo.lk/ http://yolandadata.org/
MALAWI
http://www.masdap.mw/ MOZAMBIQUE
http://moz.adapt.org/
PACIFIC
http://paris.sopac.org/
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
77
78
Box 03—1 Typhoon Yolanda GeoNode: An Example of the allow users to upload, visualize, and share data as
Collaborative Effort Possible under OpenDRI well as simply produce maps. The platform also
enables clients to federate multiple GeoNodes so
Super Typhoon Yolanda (international name Haiyan), with 305km/hr sustained
that each ministry can retain custody of the data
winds and 6m storm surge, made landfall in Guiuan (central Philippines)
and choose which data sets are made available
in November 2013 as one of the strongest cyclones on record. Yolanda
subsequently made landfall on four more islands before heading back to sea
through open licenses. Figure 3-1 highlights
and weakening into a tropical storm, eventually dissipating over China. GeoNodes supported by GFDRR and the World
Bank.
Damage across the central Philippines was severe. UN agencies estimate that
approximately 11 million people were displaced and over 6,200 killed. Entire • Collection of exposure data with
sections of cities were leveled by wind and water. Understanding the extent participatory mapping. Participatory mapping,
and magnitude of the damage was core to both the response effort and the also known as crowdsourcing and volunteer
planning for recovery and reconstruction.
geospatial information, provides a way for
Working together, the geographic information system (GIS) team from the countries and cities to create fundamental data
American Red Cross’s International Department and the team from the GFDRR on their infrastructure, including attributes
Labs set up a GeoNode data catalog to collect all geospatial data that were such as building vintage, construction materials,
technically and legally open. Over the course of three weeks, the Yolanda elevation, use, and number of stories—
GeoNode team collected over 72 layers of geospatial data, including damage
information critical for quantifying risk. Here
assessments performed by the EU Joint Research Centre, UNOSAT, the U.S.
support is provided to communities and
National Geospatial Intelligence, and the Humanitarian OpenStreetMap team.
governments to build this asset database from
The GeoNode also hosted hundreds of situation reports and PDFs from the
Red Cross and United Nations Office for the Coordination of Humanitarian
the bottom up, by (for example) collecting data
Affairs (OCHA), many of which contained geospatial data. Importantly, the through open platforms like OpenStreetMap
GeoNode also collated data from collective efforts of the OSM community, (OSM; described in box 1-2 above). Under this
which made over 4.5 million edits from 1,600 mappers working from 82 approach OpenDRI has sought to build the
countries. capacity of national OSM chapters and train them
to collect data about the exposure of the built
A technical team—BoundlessGeo and LMN Solutions, working under the U.S.
Army Corps of Engineers—developed a technique to extract footprints of environment to natural hazards. OpenDRI has
damaged buildings from these OSM data, placed them under version control supported the collection of data on millions of
in a tool called GeoGit, and made daily snapshots available. In the process, the buildings during its programs.
technical team prototyped new approaches to tracking the growing volumes
of damage assessment data generated by the OSM community. This technique • Catalyzing open data ecosystems. The
will continue to be explored for future efforts. development of a community around DRM data
is critical for fostering information sharing,
The Yolanda GeoNode is an example of a GeoNode for a specific event.
providing training, and creating the network of
This approach can be used to make specific subsets of data available to
decision makers who apply data to understanding
a community that needs them to support the specialized use cases of
response operations and recovery planning. Over the long term, the data in their risks from natural hazards and climate
event GeoNodes can be rolled back into national GeoNodes or databases, change. This work includes establishing a
allowing agencies to curate data for their general operations. This scenario community of technologists and organizers
recently played out with haitidata.org, which has been transferred to national who build applications and tools using risk data
government ownership. at “hackathons”—such as the 2014 Code for
Resilience, which builds on previous Random
Information hosted on GeoNode, in combination with hazard and exposure
data produced in the last 10 years, including a high-resolution risk assessment Hacks of Kindness activities.21 Moreover, there is
for disaster risk and financing purposes produced in 2013, is now being used to a realization that the OpenDRI program requires
inform recovery and reconstruction in the Philippines. many actors all striving toward a collective vision
and goal, so efforts to engage with a wide range
Additional information is available at Yolanda GeoNode, http://yolandadata.
org; OpenStreetMap Yolanda, http://wiki.openstreetmap.org/wiki/; Typhoon_
Haiyan, GeoGit Version Control for GeoData: http://geogit.org/.
of public, private, and academic stakeholders to make money (or gain other benefits) from it. (The
to meet collective challenges—for example, loss of revenue when data are shared is a concern
improving access to appropriate-resolution not only for governments, but also for the small GIS
digital elevation models—are a fundamental part consultancies that make a living selling their data to
of this program. local, provincial, and national government officials.)
• Creating tools for communication of risk. It Risk assessment and the need for data about
has long been recognized that the communication potential disasters represent an easier entry point
of risk results to different users is a significant into discussions about open data than many other
challenge in the global effort and one that has thematic areas (such as budget accountability),
received insufficient attention. Support to the because there are often more champions where
development of InaSAFE (described in section disasters are concerned, and it is easier to appeal to
3-22) is one example of efforts to overcome stakeholders' altruism. This ongoing work is rarely
this challenge. easy or straightforward. Opening data for wider use
can raise fears, create uncertainty, and break power
Box 3-1 offers an example of the collaborative effort
structures that control data flows. For this reason,
possible under OpenDRI—specifically, the efforts
OpenDRI works to empower local champions and
mobilized in the aftermath of Typhoon Yolanda
help them build a community of leaders to advance
(Haiyan).
the principles of open data, which in turn contribute
Challenges remain. Many governments have
/// ///
to making societies more resilient. An OpenDRI
worked with partners to aggregate and centralize field guide (GFDRR 2014), which captures the
some portion of the data that they generate through experiences and lessons learned over four years of
comprehensive stock takings. However, these efforts implementation and provides a practical guide for
have often failed or faltered, generally because other partners, was launched in March 2014. Section
governments perceive that sharing data means 3-2 offers a case study of the local application of
giving up control over it and losing the opportunity OpenDRI.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
79
80
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-2. Open Cities: Application of the Open Data for
Resilience Initiative in South Asia and the Lessons
Learned22
Robert Soden, Nama Raj Budhathoki, Marc Forni (World Bank); Vivien Deparday
(GFDRR)
South Asia is one of the most rapidly urbanizing exposed assets, which are expressed in monetary
regions in the world. Growing populations, terms. These data are insufficient for driving
unplanned settlements, and unsafe building specific investments to reduce disaster risk, because
practices all increase disaster risk in the region. individual assets are typically not accurately located,
As urban populations and vulnerability grow, described, and valued. By contrast, the Open Cities
promoting urban growth that is resilient to natural platform engages local expertise and stakeholders
hazards and the impacts of climate change becomes in identifying all building structures in a city and
an ever-greater challenge. assigning vulnerability attributes to each. In this
way, a risk assessment that identifies particular
The Open Cities project constitutes one effort
structures at risk can be completed. An assessment
to meet this challenge. Launched by the World
with this degree of precision is able to identify
Bank and the GFDRR in November 2012, it aims
structures based on importance and risk level, and
to create open data ecosystems that will facilitate
can therefore guide plans to reduce disaster and
data-driven urban planning and DRM in South
climate risk through physical investments.
Asian cities and builds on the practices and tools
developed under OpenDRI. Open Cities has Drawing upon experiences from Haiti and
///
brought together stakeholders from government, Indonesia. Open Cities was inspired by two
///
donor agencies, the private sector, universities, and other projects involving community mapping,
civil society groups to create usable information the OpenStreetMap response to the 2010 Haiti
through community mapping, build applications and earthquake (described in section 3-3) and the
tools to inform decision making, and develop the Community Mapping for Exposure effort by the
networks of trust and social capital necessary for Australian and Indonesian governments (described
these efforts to become sustainable. This process below in section 3-4). Like these efforts, Open
has been evolutionary, with opportunities for Cities made use of the OSM platform to harness
experimentation, learning, failure, and adaptation the power of crowd and community to create
incorporated into the project planning. accurate and up-to-date spatial data about the
location and characteristics of the built and natural
Open Cities approaches risk assessment differently environments.
from catastrophic risk modelling firms, whose data
are typically used by the insurance industry or for Using lessons learned from these projects in Haiti
specific portfolio analysis. Professional assessments and Indonesia, Open Cities employs a scalable
often involve computationally intensive modelling approach to understanding urban challenges and
analysis, but they also tend to rely on statistical disaster risk in South Asian cities. Three cities were
representations, proxies, or estimations of the chosen for the initial work: Batticaloa, Sri Lanka;
CHAPTER
03
Dhaka, Bangladesh; and Kathmandu, Nepal. These establish the close collaboration needed to carry
cities were chosen for their high levels of disaster out the actual mapping. But they were also meant
risk, the presence of World Bank activities related to ensure local understanding of and trust in the
to urban planning and disaster management that mapping process and in the data produced, so as to
would benefit from access to better data, and encourage local authorities to use the tools and data
the willingness of government counterparts to for their own DRM and urban planning projects.
participate in, and help guide, the interventions.
Open Cities has sought to support the creation A team of four technical experts (three recent GIS
of new data in each of these projects, but has and IT graduates and one experienced GIS analyst)
also supported broader ecosystems of open data was hired and trained in OSM techniques in order
production and use in the three cities. Leveraging to supervise and support the overall mapping
data to improve urban planning and DRM decisions process. Team members worked directly with the
requires not just high-quality information, but also staff from local partners, including the Batticaloa
the requisite tools, skills, and willingness to commit Municipal Council, the Batticaloa District, the
to a data-driven decision-making process. With Manmunai North Divisional Secretariat, and the
this in mind, Open Cities also sought to develop 48 Grama Niladhari that make up the Manmunai
partnerships across government ministries, donor North Divisional Secretariat. A small group began
agencies, universities, private sector technology by tracing all building outlines into OSM using
groups, and civil society organizations to ensure satellite imagery and then added landmarks, roads
broad acceptance of the data produced, facilitate and road names, and points of interests using local
data usage, and align investments in risk reduction paper maps provided by the divisional secretariat.
across projects and sectors. With the first phase of This effort created a solid reference map for the
Open Cities complete in each of the projects, these surveying work. The work was then split into two
partnerships will be critical for continuing the work components: buildings were surveyed by 48 recent
and expanding into new cities in the region. graduates hired to work on the Grama Niladhari
local planning and development, and surveyed data
Case study: Batticaloa, Sri Lanka. Batticaloa, a
/// ///
were entered by government workers who were
major city in Sri Lanka’s Eastern Province severely also responsible for fixing the maps and refining
affected by the Sri Lankan civil war and the 2004 the point of interests. Both groups were trained in
Indian Ocean tsunami, is located in a hazard- OSM and surveying techniques by the Open Cities
prone area that has suffered near-annual droughts, team, and all the staff involved in the data collection
floods, and cyclones. Some limited hazard maps
received a stipend for the extra work.
were available for the area, but no detailed digital
geographic data of the built environment were Data on basic characteristics (number of floors,
available for use in risk studies or for informing usage, and construction materials of walls and
potential infrastructure and risk mitigation projects. roof) were collected for all 30,000 buildings in the
To fill this gap, Open Cities started a pilot project to area. These data are now freely available in OSM
map the building stock, including critical assets of and in the government geospatial data-sharing
the Manmunai North Divisional Secretariat, which platform RiskInfo (www.riskinfo.lk) for easy use
covers an area of 68km and includes about 90,000
2 by many stakeholders. To publicize the benefits of
people around the town of Batticaloa. The work these techniques at the national level and promote
began with a series of meetings with the Batticaloa their adoption, high-level managers of the relevant
local authorities. In part, these were designed to national agencies were briefed regularly and given
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
81
82
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
final results when available. Two week-long training to the public—directly into OSM. This allowed the
courses, one dealing with OSM techniques and mapping team to focus on field surveying, in which
the other with use of data for decision making basic characteristics, such as building height, usage,
(specifically the combination of data with existing construction materials, and age were collected
hazard maps through GIS tools and the InaSAFE through visual survey of each building. The team
tool) were conducted at the national level with all also mapped road characteristics (width and surface
relevant national agencies. Discussions are ongoing type) along with important water and sanitation
with various ministries concerning the next phase infrastructure. The data were added to OSM during
of the project. There is a strong interest in scaling times when conditions prohibited field surveys (e.g.,
up the project to cover a greater geographic area poor weather conditions). Two weeks of training at
and in streamlining the use of the data in more DRM the beginning of the project and a final two weeks of
applications and sectors. data entry and quality assessment at the end of the
project left two months in the middle for fieldwork.
Case study: Dhaka, Bangladesh. Dhaka's Old
During this period, the team was able to finish
/// ///
City is a crowded and complex area of immense
complete maps of the three wards.
historical value and an important locus of social and
economic activity. In consultation with Dhaka Water In total, 8,500 buildings, 540 of which were deemed
and Sanitation, seismic risk experts from Bangladesh to have historical significance, were surveyed.
University of Engineering and Technology (BUET), Sections of roads measuring 43km and drainage
and a local nongovernmental organization (NGO) works measuring over 50km were also assessed. This
working on heritage preservation and restoration information is now available to the public through
in Old Dhaka, the Dhaka Open Cities pilot sought the OSM platform. Several training courses and
to create detailed maps of three of the Old City's 15 presentations on OSM were also given to university
wards. These maps would provide data useful for students, government partners, and private sector
planning evacuation routes, managing water and technology companies during the project period
sanitation infrastructure, and understanding the in order to help the OSM community in Dhaka
location and characteristics of heritage buildings. In grow. The results of the pilot were presented to the
partnership with BUET, which provided technical government and other key stakeholders in December
support and a working space, 20 engineering and 2013. Consultations are ongoing concerning the next
planning undergraduates were hired as mappers and phase of the project.
were trained in a series of workshops over a three-
Case study: Kathmandu, Nepal. Kathmandu,
month period. A local nonprofit GIS consulting
/// ///
the capital city of Nepal, has very high potential
organization, CEGIS, was contracted with to provide
for significant loss of human life during a major
management and quality control for the work. The
earthquake event.23 In November 2012, in
Humanitarian OpenStreetMap Team (HOT), a
partnership with the government of Nepal, the
nonprofit specializing in the use of OpenStreetMap
World Bank and GFDRR launched a project to
in development and humanitarian relief situations,
build seismic resilience in the Kathmandu Valley’s
also provided training and technical oversight to
education and health infrastructure, in part by
the project.
creating a disaster risk model to determine the
The effort began by importing building footprint relative vulnerability of the relevant buildings.
data for the three wards—created by CEGIS as part Once complete, the model will be used to prioritize
of a different project but until that point unavailable plans for retrofits of schools and health facilities
CHAPTER
03
to improve structural integrity in the face of made substantial contributions to the OSM project
earthquake. However, a critical input into this model in Kathmandu, suggesting the opportunities for
is building-related exposure data. partnerships between development organizations.
A local NGO called the Kathmandu Living Labs,
World Bank staff and consultants began the year-
staffed by participants in the first phase of the Open
long project by assembling a team of mappers and
Cities project, has been created in order to continue
community mobilizers. The team was responsible
the work.
for a variety of tasks, from field surveying to
software development to training of community Lessons learned and recommendations.
groups in OSM. The core team comprised six Although the Open Cities project is ongoing, several
graduates of Kathmandu University who were key lessons have already emerged that can be
recruited based on their prior contribution to applied to other initiatives.
Nepal’s then-nascent OSM community. They were
1. Government ownership is important.
paid full-time salaries at rates commensurate with
/// ///
the local salary structure for recent graduates in Although many Open Cities partners and
technical disciplines. The project also recruited six participants will be from civil society and the private
part-time interns from Kathmandu University and sector, government counterparts in line ministries
11 volunteers from Tribhuvan University. Office must be involved in projects’ development and
space for the team provided access to meeting execution. Engaging governments early in the
rooms, reliable Internet service, and opportunities planning process and ensuring close involvement
to interact with other technologists and throughout is an essential component of a successful
entrepreneurs, some of whom later became active Open Cities project. Governments are primary
in OpenStreetMap. stakeholders for many DRM and urban planning
projects and provide necessary legitimacy to Open
Open Cities Kathmandu surveyed 2,256 schools
Cities work. In Kathmandu, the involvement of
and 350 health facilities in the Kathmandu Valley.
the Department of Education in the mapping work
In addition to collecting a comprehensive list of
will be critical for developing the department’s
structural data for health and school facilities, the
confidence in and use of the data to prioritize
team worked to create a comprehensive base map of
seismic retrofitting activities. An official letter in
the valley by digitizing building footprints, mapping
support of the project carried by mapping team
the road network, and collecting information on
members helped them gain the access to schools and
other major points of interest. The Open Cities team
health facilities that was needed for conducting their
also conducted significant outreach to universities,
technical communities, and government in order assessments. In Sri Lanka, the project deliberately
to expand the OSM community. Over 2,300 involved local authorities directly in the mapping
individuals participated in OSM trainings or activities as a way to ensure government ownership
presentations during the first year of the project. of the project and the use of the data in various
The data have been used in plans to retrofit applications.
school and health facilities and in applications
2. Universities make good partners.
/// ///
for transportation planning; moreover, the U.S.
Agency for International Development (USAID) has Universities have been valuable allies during
incorporated the data into disaster preparedness the first year of Open Cities work. Outreach to
planning exercises. The American Red Cross has also university departments of engineering, geography,
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
83
84
computer science, and planning has provided
projects with critical connections and support. In
Dhaka and Kathmandu, university students have
played an important role in mapping activities and
software development. Students from technical
departments tend to learn OSM quickly, and some
students in Kathmandu fulfilled a requirement to
complete internships or volunteer projects through
participating in Open Cities. University faculty have
also provided useful support. In Dhaka, professors
from the BUET Civil Engineering Department
and Planning Department contributed to the
design of the mapping project. Professors in the
Geomatics Department at Kathmandu University
provided guidance to the project on quality control
techniques for surveying, and they also incorporated
OSM into their courses. Training future classes of
university students will help the OSM community
in Kathmandu continue to grow after the formal
project period has ended.
3. Access to imagery is critical.
/// ///
As the work of Haiti’s OSM community made
clear, access to high-resolution satellite imagery 4. Data must be trustworthy and credible.
/// ///
is extremely useful for efficient mapping of
Data quality is a frequently raised issue in
infrastructure. However, such imagery is often
community and volunteer mapping projects.
prohibitively expensive or available only under
licenses that prohibit digitization by the public. Numerous measures were taken by the Open
With this in mind, the U.S. State Department's Cities project to ensure that partners and intended
Humanitarian Information Unit launched an users of the data would trust the data’s accuracy
initiative in 2012 called Imagery to the Crowd, and completeness. In Kathmandu, partner
which makes high-resolution imagery owned by
organizations—including the National Society for
the U.S. government accessible to humanitarian
Earthquake Technology, a respected NGO working
organizations and the volunteer communities that
support them. Open Cities Kathmandu partnered on seismic resilience, and the Kathmandu University
with USAID and Imagery to the Crowd to release Geomatics Department—provided technical
2012 satellite photography for the Kathmandu guidance to the project as well as independent
Valley and to organize volunteers in Nepal, the quality assessments throughout the process to
United Kingdom, Germany, and the United States
provide credibility. In Dhaka, key stakeholders,
to digitize building footprints. The data created by
including BUET and representatives of government
these volunteers have been incorporated into USAID
disaster response planning, and they provided and civil society, were consulted throughout the
a solid foundation upon which the Nepali OSM project, and many were given basic training in OSM
community can continue to expand and improve. in order to familiarize them with the platform.
CHAPTER
03
5. Sustained engagement is required
///
to contribute to cultural and policy shifts within
for success. ///
technical groups and government that will prioritize
open data and broad participation in development
For these projects to be successful, sustained
challenges. When projects of this kind are planned,
engagement with local partners is necessary. Too
the parties involved must understand and commit to
often technology and data projects of this sort are
sustained investment in their success.
discrete and short-term endeavors. A workshop or
a weeklong training course is simply not enough In its early phase, Open Cities has demonstrated
time to trigger the kinds of change that Open Cities
success in engaging nontraditional institutions and
hopes to support. Although OSM makes mapping
community groups in the process of creating high-
more accessible to nonspecialists, collecting and
resolution spatial data that can be used in support
interacting with geographic information remains a
of urban planning and resilience-building programs.
complex technical undertaking, one that requires
There is still work to be done to establish direct
more training and involves a longer learning
links between the OSM data set and target users
process than is often assumed. It also takes time
to build technical communities of OSM mappers in and out of government, but the initial reception
and software developers who are familiar enough has been positive, and there is strong interest from
with the platform to comfortably deploy it in their a number of other development institutions in
own tools and applications—and creating these learning from the early experience and in partnering
communities is an important part of sustaining on future work. In the future, Open Cities will also
Open Cities projects. Finally, Open Cities seeks seek to scale through expansion of the range of
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
85
86
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-3. Preliminary Survey of Government Engagement with
Volunteered Geographic Information
Muki Haklay (University College London), Sofia Basiouka (National Technical
University of Athens), Vyron Antoniou (Hellenic Military Geographical Service),
Robert Soden (GFDRR)
When data and information are shared and part • Designing strategies to encourage governments
of open systems, they promote transparency and to engage with VGI efforts is not straightforward,
accountability, and ensure that a wide range of and we are still learning from early experience
actors can participate in the challenge of building what opportunities exist and what methodologies
resilience. Arguably, one of the greatest revolutions work well.
in this open data space has been the increasingly
The survey project focuses on cases that
active involvement of local people in geospatial data
demonstrate a synergy between government and
collection and maintenance—a process known as
citizens or civic society organizations. “Synergy”
volunteered geographic information (VGI).
means a government authority’s clear use of
A preliminary survey of government engagement contributed information to make decisions and take
with VGI was undertaken in order to strengthen actions. Four case studies are highlighted: Canada’s
governmental projects that incorporate voluntary interaction with OSM, Haiti’s interaction with the
and crowdsourced data collection and to provide Humanitarian OSM Team, Indonesia’s experience
information that can support wider adoption of VGI. with community mapping, and the U.S. State
The survey compiles and distributes lessons learned Department’s interaction with HOT.
and successful models from efforts by governments
Canada. In Canada, the main duty of National
/// ///
at different levels. The survey project began from
Mapping Agencies is to provide up-to-date
the following premises:
topographical maps and a range of spatial products
• Sources of VGI data such as OpenStreetMap are to public and private sector. Likewise, the role
growing increasingly important across a range of of the Mapping Information Branch at Natural
thematic areas and user communities. Resources Canada is to provide accurate geographic
information on landmass at the scale of 1:50,000.
• Concerns about the quality, consistency, and This task involves regularly covering an area of 10
completeness of VGI data have been assessed by million km2 divided into 13,200 map sheets.
a range of studies and overall have been found
Taking into account the results of ongoing research
not serious enough to prevent exploration of VGI
regarding VGI quality, Canadian authorities choose
data as a valuable data source.
to cooperate with the OSM community to see if
• For governments, interacting with VGI and how the updating process could profit from the
communities is different and potentially more evolution of VGI. As Beaulieu, Begin, and Genest
complex then interacting with typical sellers and (2010) describe, the first step to this synergy was
resellers of GIS data. made by the Centre for Topographic Information in
Sherbrooke, which released the digital topographic and effort on areas with identified changes. Given Figure 03—2
map of Canada in the native .osm format. This that the authoritative database had failed to update Change detection
move enabled further integration of the Canadian all the originally designed spatial entities, this using OSM.
authoritative data into OSM and gave the OSM contribution is valuable.
Source: Beaulieu, Begin, and
community a chance to interact with—that is, Genest 2010.
Engaging with OSM has also presented challenges.
complete, correct, or update—the authoritative data.
Among the issues that must be addressed are the Note: Gray = OSM road
In addition, authorities are now able to regularly
imperfect compatibility of the two data sets (in network; green = data
compare the OSM database with the original data missing from OSM; red
terms of semantics and attribution), the virtual
to pinpoint the differences (figure 3-2). Those = data missing from
nonexistence of metadata for OSM data, and the authoritative data.
differences are treated as potential changes and are
differences in coverage (OSM is concentrated in
verified using the authoritative channel at the field.
urban areas compared to the uniform authoritative
Verified changes are propagated to the authoritative
coverage). All these differences stem from the
database.
differences in the two geographic information–
On the positive side, the titanic work of keeping the generating processes—that is, the bottom-up
data sets up-to-date has been facilitated by the OSM and looser OSM process versus the top-down
community. Leveraging the OSM crowdsourcing authoritative process. Yet another issue involves
mechanism, the Canadian authorities have a conflict between license and use terms of OSM
developed a much-needed change-detection process, and the intellectual property rights of Canadian
which helps the authorities concentrate resources authorities.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
87
88
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
The Canadian and OSM synergy rests mainly on two available spatial data were poor in content and had
pillars: the authorities’ recognition that they have last been updated in the 1960s; moreover, the local
been unable to keep the national data up-to-date, mapping agency collapsed in the earthquake and
and their willingness to acknowledge and trust the many of the skilled employees were lost. An updated
quality of the OSM data. Another factor contributing map was urgently needed to enable distribution of
to the synergy is that Canadian authorities are supplies, attention to collapsed buildings, repair of
well organized and equipped and therefore have a damaged infrastructure, and provision of medical
standard process regarding spatial data collection, services.
change detection, and spatial data quality control
and quality assurance. They can easily handle the The Haiti disaster response constitutes an example
addition of OSM data in their processes, and the of a successful project in which geographic
results are visible, understandable, and tangible. information was released from partners to the
In other words, in this case, the context in which crowd for enhancement and then returned back to
authoritative and non-authoritative entities interact government for activation—although government
is an important influence on how easy it is to was rather reluctant to involve volunteers. Historical
integrate the two different spatial data sets. maps, CIA maps, and high-resolution imagery in
Yahoo were used for tracing in OSM so that the
The Canadian experience suggests several basic maps could be improved. Within 48 hours,
important lessons: new imagery from the World Bank, Google, and
others was also made available for tracing in OSM.
• An authority’s recognition that it needs
According to HOT, within a month, 600 volunteers
assistance to meet its target can trigger the turn
had added spatial information to OSM, and OSM
to VGI.
was used as a default base map for the response to
• VGI data sets can be used by authoritative and the Haiti earthquake.
governmental bodies to supplement or facilitate
Four factors explain the success of this project:
their standard operational procedures.
the quick creation of the data, the low cost, the
• Differences in structure and operation mean that numerous contributions of volunteers from the
updates to geographic information do not move OSM community, and the public release of high-
freely between the two systems. resolution satellite imagery. The first two factors
were summarized by the opinion that the United
• Different terms of use and license options for the
Nations “would have taken tens of thousands of
two data sets can create connectivity problems.
pounds and years to do what OpenStreetMap did
Haiti. Haiti was dramatically affected by the 7.0
/// ///
in 3 weeks.”25 The third factor was the remote
magnitude earthquake of January 12, 2010. Most volunteers who acted quickly, coordinated their
estimates of deaths range from 100,000 to 159,000, efforts, and disseminated the appeal for help all
with Haitian government reports of over 200,000 over the world. As Tim Waters puts it, “It is the first
fatalities. More than 250,000 residents were injured time where individuals from the comfort and safety
and more than 30,000 buildings were collapsed or of their own home can literally help other people
severely damaged. The Haitian government and the save lives in a disaster zone.”26 A final key factor
numerous nongovernmental organizations seeking in the success of the project was the willingness of
to respond to the disaster lacked accurate and partners to provide spatial data and imagery free of
up-to-date maps to help guide their work. The only license restrictions.
CHAPTER
03
Despite the project’s overall success, several Wibowo, and Nurwadjedil 2013). The community
challenges should be highlighted. First, despite the mapping component had clear leadership, specific
efforts of HOT and others, the Haitian national guidelines in data manipulation, and great
mapping agency (CNIGS) was never fully involved in coordination of the different contributors. The
the project. This represented a missed opportunity crowd was motivated to participate (driven by
to establish a richer connection between the a desire to improve disaster protection, win the
Haitian government and the OSM community. mapping competition, or other reasons) and was
Second, the number of volunteers involved in the supervised during the various stages of the process,
digitization and the speed with which it occurred and the process of data collection and manipulation
caused coordination difficulties, which in turn led to was well defined. A factor contributing to the
duplication of data and effort. project’s success was the evaluation of the data by
academics and project leaders.
Undeniably, what OpenStreetMap did in Haiti
changed both disaster response and perceptions Some limitations of the project involve the quality
of VGI forever.27 Overall, the Haitian experience of the results, which while acceptable overall and
suggests several important lessons: in some cases very good, was in some cases very
poor (Gadjah Mada University and HOT 2012).
• Crowdsourcing of mapping is a valuable ex post
There appeared to be many empty or wrong
disaster response.
records concerning the structure of buildings.
• Volunteers from the OSM community and the Some minor deficiencies were also noted during
access to high-resolution imagery made the the implementation, such as the use of time-
project a success. consuming technical methods (e.g., use of Excel
spreadsheets in data collection or manual methods
• Coordination among distributed volunteers
of data manipulation).
involved in mapping is a challenge that needs to
be addressed in order to ensure efficient use of The Indonesian experience suggests several
their time. important lessons:
Indonesia. The Indonesian community mapping
/// ///
• An ex post response can be focused on
of exposure project began in early 2011 and is appropriate models and parameters and can
still active (for more details, see section 3-4). The calculate the damages in case of a physical
project’s goal was to use OSM to collect previously disaster by using crowdsourced spatial data sets.
unavailable data, including structural data, for
both urban and rural buildings and use the data • Successful interaction between the VGI
in appropriate models to estimate the potential community and Indonesian government officials,
damage from natural hazards. The combination who evaluated the data used for scenario building
of these two components and the use of realistic as reliable, led to the project’s being continued
data led to the development of the InaSAFE tool and expanded past the initial phase.
(discussed in section 3-22).
• Risk managers and the local community can
The project was seen as successful from a human, combine local wisdom with scientific knowledge
technical, and financial point of view. It has enabled to produce realistic scenarios for numerous
local government to use spatial data to visualize different physical disasters that may occur at the
where people are most in danger (Chapman, area of interest.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
89
90
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
• The success of the project was due in part to the the digitization of basic infrastructure data into
coordination of volunteers and full use of human OSM in eight countries to support humanitarian
resources and technical innovations. response or disaster risk reduction.
• The mixed quality of the attribute data is an issue Following the Typhoon Haiyan disaster in the
of concern. Philippines in November 2013, Imagery to the Crowd
published images for Tacloban, Ormoc, Northern
Imagery to the Crowd. As shown in Haiti,
Cebu, and Carles. This imagery supported a massive
/// ///
facilitating the access of volunteer communities to
volunteer effort of over 1,600 mappers from the
high-quality aerial and satellite imagery can have
OSM community, coordinated by HOT, who
dramatic results. Such imagery is often prohibitively
contributed nearly 5 million changes to the map—
expensive, however, or available only under licenses
changes that provided detailed information on the
that would prevent digitization by the public.
location and extents of pre-event infrastructure as
With this in mind, the U.S. State Department’s
well as offering a preliminary damage assessment
Humanitarian Information Unit launched a new
(see box 1-2 for more detail).
initiative in 2012 called Imagery to the Crowd. This
program makes high-resolution imagery, purchased Technical and policy efforts are under way to
by the United States from providers like Digital increase the speed at which imagery can be released
Globe, accessible to humanitarian organizations and and to standardize and improve the process, but this
the volunteer communities that support them. Since new initiative has already achieved demonstrable
its inception, Imagery to the Crowd has facilitated results.
CHAPTER
03
3-4. Collection of Exposure Data to Underpin
Natural Hazard Risk Assessments in Indonesia and
the Philippines
A. T. Jones, K. Van Putten, M. Jakab (Geoscience Australia); M. L. Bautista, B.
C. Bautista, I. C. Narag (Philippine Institute of Volcanology and Seismology);
A. Wibowo (Badan Nasional Penanggulangan Bencana); K. Chapman
(Humanitarian OpenStreetMap Team)
Until recently, the scope and usefulness of risk for Natural Disasters (CSCAND)—to promote
assessments in the Asia-Pacific region were limited the goals of the Greater Metro Manila Area Risk
because the fundamental exposure data required Assessment Project.29 The GMMA RAP was one of
were either missing or incompatible with the level the first integrated multi-hazard probabilistic risk
of risk assessment required. But two projects in assessments ever undertaken for a megacity and
the region, one in Metro Manila and the other in included estimations of economic loss and potential
Indonesia, have each found a way to develop much- casualties. The project was intended to provide
needed exposure data. a better understanding of exposure databases,
including how to prepare them; to make exposure
In the Philippines, a state-of-the-art technological information available for analyzing natural hazard
approach to collecting exposure data was used on an risk and climate change impacts in the Greater
urban scale as part of the Greater Metro Manila Risk Metro Manila Area; to improve assessments of
Assessment Project (GMMA RAP). This approach the risk of and impacts from flood (in the Pasig-
incorporated data from high-resolution aerial Marikina River basin) and from tropical cyclone
imagery (cm resolution) and airborne LiDAR (giving severe wind; and to improve the understanding of
ground and building heights to mm accuracy) into earthquake risk in the Greater Manila Metro Area.
GIS data sets to provide needed information about
individual buildings’ location and size; it then added To achieve its goals, GMMA RAP needed to
further information about building construction address the challenge of gathering data in a heavily
type, land-use classification, and residential populated and highly complex urban environment.
population estimates. In Indonesia, a collaborative Attempting to acquire, manage, and maintain
and cost-effective approach—crowdsourcing exposure information for every significant feature
through OpenStreetMap—was used both to collect (there are over 1.5 million buildings, for example)
exposure data (including information on building was not practical given the limited available existing
type, building capacity, wall type, roof type, and data and the three-year time frame of the project,
number of stories) and to create a methodology that and given the dearth of risk analysis tools able to
could be replicated across the entire country. handle very large volumes of exposure data.
Philippines. With the Office of Civil Defense
/// /// The project team designed the GMMA RAP exposure
acting as a coordinating agency, Geoscience database to make use of existing methods and
Australia has worked with a group of government draw on lessons learned in preparing exposure
of Philippines technical agencies—known jointly as data for an earlier project on earthquake risk.30 The
Collective Strengthening of Community Awareness database was populated with a range of data from
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
91
92
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
other projects or already held by government of Several additional data sets were derived from these
Philippines agencies, by Local Government Units, LiDAR data, including a digital elevation model
and by other organizations, and these were then and a digital surface model. Both these data sets
enhanced with additional data. To support the were generated with a 1m horizontal resolution to
process of developing data and offer local expertise optimize them for spatial analysis. Where the digital
and knowledge, a technical working group of elevation model and digital surface model were
spatially coincident, the difference between their
specialists was established. Because acquiring and
elevation values was the height of features above the
managing highly complex data is so difficult, and
ground. After vegetated areas were isolated from the
because detailed exposure data were unavailable for
derived features through analysis of aerial imagery
some areas of the Greater Metro Manila Area, the
that accompanied the LiDAR data, a model of
project adopted an area-based approach to exposure
artificial elevated areas—i.e., buildings—was left.
data development. This approach allowed data to be
included in the database at a suitable level of detail The extents and heights of buildings determined
while offering the flexibility to move to a feature- from the LiDAR data were then used to estimate
based approach as data became available. the floor area of the buildings (which is ultimately
used to determine the amount of damage a building
Statistical information on population and building will suffer if a hazard event occurs). The vertical
type (e.g., from National Statistics Office Census distance between floors of buildings, also referred
data) was used to describe exposure characteristics to as the inter-story height, was assumed for each
Figure 03—3 relevant barangay, and this was used in conjunction
for broadly defined areas (in this case, barangays, the
LiDAR provides an with the areal extent of the building to calculate the
smallest administrative division in the Philippines,
opportunity to map floor area. Sample images are in figures 3-3 and 3-4.
equivalent to an inner-city neighborhood or suburb).
and visualize in detail
This information was then supplemented with Finally, the collected (census) data and calculated
the highly urbanized
exposure data derived through a novel technological (LiDAR) data were combined into statistical models
environment
approach developed at Geoscience Australia, in for individual barangays based on land use. These
of Manila.
which data from airborne LiDAR, which measures formed the basis of the GMMA RAP exposure
Source: Danilo Pinzon building heights very accurately, and high-resolution database and the economic loss calculations
(World Bank). aerial imagery were incorporated into GIS models.31 determined through the risk analysis.
Figure 03—4
Application of aerial
imagery, LiDAR
data, and land-use
mapping to develop
exposure database
(Taguig City).
Source: Bautista et al. 2014.
Note: High-resolution aerial
imagery over Taguig City is
shown in upper left; detailed
land-use mapping is show in
upper right; building heights
determined from LiDAR data
are shown in lower left (red
= high; blue = low); and the
number of stories is shown
in lower right (blue = low
rise; green = medium rise;
orange = high rise).
Crowdsourcing in Indonesia. The Australia-
/// ///
Team (Chapman 2012). OSM provides communities
Indonesia Facility for Disaster Reduction (AIFDR) with tools to quickly, simply, and easily map their
initiative, which is a key part of Australia’s environment; when mapping infrastructure, users
development program in Indonesia, collaborated can tag objects with information (for example,
with the Indonesian National Disaster Management about use, wall type, roof type, capacity, etc.). This
Agency (Badan Nasional Penanggulangan participatory mapping approach provides detailed,
Bencana, or BNPB), the GFDRR, and the World local-scale exposure information that can be used
Bank to develop InaSAFE (see section 3-22). The by governments and communities for developing
requirement to provide a spatially independent impact assessments. It minimizes access and
product that could be applied anywhere across usability issues by employing low-tech approaches
Indonesia meant it was not possible to underpin the that are easy to carry out (for example, it uses
risk models with a single exposure database, as was paper maps with digital imagery that can later be
the case in the GMMA RAP. Instead, a partnership uploaded into a database); and because it engages
was formed to obtain location-specific exposure communities in mapping their own vulnerability, it
information that was at the right scale, up-to-date, has the added benefit of increasing their sense of
and complete. ownership over resultant impact assessments.
To determine if OSM could be used to map exposure This pilot was a first attempt to use OSM to collect
in Indonesia—that is, provide exposure data for detailed exposure and vulnerability data and then
impact scenarios—a community mapping pilot was feed it into scientific models to determine how
developed through collaboration with the Australian a disaster would affect a specific location. An
aid–funded Australian Community Development evaluation of OSM data showed that for the 163,912
and Civil Society Strengthening Scheme (ACCESS) buildings mapped in Indonesia, results were not
Phase II and the Humanitarian OpenStreetMap significantly different from ground-truthed and
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
93
94
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
referenced data (Gadjah Mada University and are limited, and where existing data are limited as
HOT 2012). well. Within this context, approaches to acquiring
exposure data depend on the scale, purpose, and
Figure 3-5 shows the increase in exposure data over
end-users of the risk assessment being undertaken,
time for three locations in Indonesia being mapped
as well as on factors specific to the assessment’s
by OSM. The first location was Dompu, in Sumbawa
location. The examples suggest, however, that
(top row of figure). The second location was Jakarta
procedures for collecting data may be useful for a
(middle row), where AIFDR and HOT partnered
range of applications if they are well thought out and
with DKI–Jakarta Regional Disaster Management
based on a consultative process involving technical
Agency, UNOCHA, World Bank, and University of
experts, decision makers, and disaster managers.
Indonesia to map critical infrastructure through
district workshops that captured local knowledge Indeed, the project in Indonesia has since been
from urban village heads. The third location was used as a template for similar endeavors worldwide
Padang (bottom row), where HOT asked volunteers and as a model for coordinating and structuring a
to use its online tasking manager (developed to crowdsourcing project. It is also an example of how
help large groups of volunteers to map in one area developing countries can protect themselves from
without overlap or conflicting contributions) and or prevent natural disasters. The project succeeded
where in two months, volunteers mapped 100,000 because it was supported by the local government
buildings. with money and time depth; its methodology was
adapted to the nature of the mapping area (rural
Since the end of the pilot in March 2012, over 1.3
or urban); and it was well designed and defined in
million buildings have been mapped in Indonesia
terms of technical structure and human resources.
with OSM, over 900 Indonesians have been trained
Volunteers tended to remain involved with the
in the use of the software, and three universities
project both because they received incentives to
have begun to teach OSM within their GIS program.
continue their efforts, and because they could see
Conclusions. These two examples demonstrate
/// ///
the importance of their efforts—that is, see how the
Figure 03—5 different approaches to capturing exposure data new data they had collected could be combined with
Growth in exposure where budgets, time frames, and human resources hazard layers to determine potential disaster impact.
data through
crowdsourced (OSM)
mapping of buildings
and infrastructure
in three locations
in Indonesia.
Source: Australia-
Indonesia Facility for
Disaster Reduction.
Note: Top row shows
Sumbawa in January 2011,
January 2012, and December
2012. Middle row shows
Jakarta in January 2011,
January 2012, and December
2012. Bottom row shows
Padang in January 2011,
August 2011, and December
2012.
CHAPTER
03
3-5. International Collaboration of Space Agencies to
Support Disaster Risk Management Using Satellite Earth
Observation32
Philippe Bally (European Space Agency), Ivan Petiteville (European Space
Agency, CEOS Disasters Working Group), Andrew Eddy (Athena Global),
Francesco Gaetani (Group on Earth Observations Secretariat), Chu Ishida (Japan
Aerospace Exploration Agency), Steven Hosford (Centre National d’Etudes
Spatiales), Stuart Frye (NASA), Kerry Sawyer (CEOS), Guy Seguin (International
Space Consultant)
Working together in groups such as the Committee • Basic mapping. Nearly all the mapping
on Earth Observation Satellites (CEOS), national services provided by satellite EO to DRM and
space agencies are seeking to coordinate their humanitarian aid projects are underpinned by
efforts and resources to make large volumes of basic mapping. This base-layer information serves
earth observation (EO) data available for use in as a standardized geographical reference data set
risk management and disaster reduction. EO data that can be used to determine key geographical
attributes of a given area.
are currently used operationally in the context of
disaster response by the International Charter (see • Asset mapping. Asset mapping provides up-
box 3-2). to-date, synoptic, and objective infrastructure
information concerning the asset at risk. It can
The EO data come in various forms—medium-
also add to and improve knowledge about the
and high-resolution optical data; medium- and
potential impact of natural hazards in areas at
high-resolution microwave radar data (C, L, and risk.
X band); interferometric SAR (synthetic aperture
radar) data products; infrared and thermal data; and • Urban mapping. This service assesses the
meteorological data sets—and can serve as the basis structure of the built-up areas. In agglomerations
where urban expansion is progressing very
of regular, detailed updates on the status of hazards
rapidly and the territorial conditions are
globally, regionally, or nationally. Currently, much
extremely constrained, EO data help to create
EO data complements ground data, but where in
easily updatable baseline maps of urban assets
situ information is limited, EO data may be the only
while taking into account location of informal
source of information available.
settlements and their high vulnerability to
natural hazards such as floods and landslides.
EO data can be instrumental in risk assessment
and disaster reduction. These data can be used • Remote assessment of damage. This
for a range of applications, such as mapping service uses processed satellite data from before
hazards, evaluating asset exposure, and modelling and after a disaster to provide crisis mapping,
vulnerability: situation mapping, and damage assessment
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
95
96
Box 03—2 International Charter Space and Major Disasters for on-the-ground disaster response by
governments, first responders, and planners of
A good example of the potential of satellite EO can be seen in the International
resilient recovery.
Charter Space and Major Disasters (www.disastercharter.org), an international
collaboration among space agencies that uses space technology to aid in
• Flood risk analysis. This service provides
response to disasters. When a disaster occurs, the International Charter grants
information to support risk management and
access to satellite data at no cost and in a rapid fashion. The Charter aims
to help better organize, direct, and mobilize national disaster management water resources management. Depending on
resources during emergencies and to assist the international relief community input data and methodologies used, different
in situations requiring humanitarian assistance. The only users who can types of information can be extracted, such as
submit requests are Authorized Users, a predefined list of organizations with the classified distribution of the land cover and
a mandate related to DRM. The Charter is focused on hazards with rapid onset
socioeconomic units in areas at risk, or hazard
scenarios, in the immediate response phase, and aims to service operational
damage information based on measurements of
users wherever a disaster occurs. Since its inception in 2000 it has delivered
services over 400 times in well over 100 countries. water depth and/or flow velocity.
To cite the Charter and its dramatic evolution over the last decade as progress • Precise terrain deformation mapping.
toward risk assessment may be surprising, given the Charter’s response-only This service contributes to geohazard risk
focus. Yet the Charter remains a striking example of what space agencies assessment to support mitigation, prevention,
working together can achieve. By raising the profile of satellites in disaster
and preparedness. For a wide range of risk
response, the Charter has greatly increased the DRM community’s interest
assessments, including those concerned with
in EO satellite data and EO-based solutions. Satellite based geo-information
can contribute to the entire cycle of risk management, including mitigation, flood, seismic hazard, and climate change,
warning, response, and recovery. To date, much of the DRM effort of the EO terrain-motion information has direct relevance.
sector has been focused on disaster response and recovery, which by its
nature attracts more attention but also more resources than pre-crisis phases. • Landslide inventories and landslide
Stronger ties to end-users and increased collaboration with DRM practitioners monitoring. These services provide hazard
would increase the impact of EO-based response activities such as those of mapping information in landslide-prone areas
the Charter. At the same time, meeting the ongoing need for information by
and carry out repeat observations over large
supplying large volumes of data over large areas is very different from meeting
areas. (Locally, emergency monitoring of hot
the more limited needs arising during the response phase; and within the
context of existing systems, supplying EO data for disaster mitigation on a
spots typically is performed using ground-based
global basis represents a clear operational challenge for satellite agencies. radar as the primary source).
Box 03—3 Innovations in Earth Observation over the Coming Decade
The resolution and availability of earth observation satellites are much greater With a constellation of two operational satellites allowing a five-day geometric
now than they were a decade ago. It is still the case, however, that the use of revisit time, Sentinel-2 will provide systematic coverage of the overall land
satellite-based EO in DRM is constrained by the lack of observations for risk- surface. Other EO missions that will greatly enhance global observations
prone areas. for DRM applications include the ALOS-2 mission of JAXA (Japan Aerospace
Exploration Agency) and the Canadian Radarsat Constellation Mission (RCM).
Space agencies are addressing this issue by putting in place new data policies
Two new U.S. commercial optical satellites, Skybox and PlanetLab, will become
that will soon provide users with open and free access to agencies’ archives
available in the near future and will greatly enhance the accessibility of these
of images from the past 10 years, starting with SPOT images. They are also
high-resolution images.
developing complementary plans of observation. Two upcoming satellite
missions—Sentinel-1 and Sentinel-2, jointly developed by the European
In complement to the systematic and frequent coverage over wide areas
Commission and the European Space Agency and scheduled to launch in
made available by EO missions, detailed and up-to-date observations are
spring 2014—will make high-quality SAR and multispectral data freely available
being provided through very high-resolution systems operated by commercial
to end-users.
players and national space agencies. Relevant missions are the Pléiades
The SAR data generated by Sentinel-1 can be used for global, national, and mission of CNES (France’s National Center for Space Studies) and Astrium
local hazard assessments. The multispectral Sentinel-2 mission—for global Geo-Information Services; Cosmo-Skymed of ASI (Italian Space Agency) and
land observation at high resolution with high-revisit capability—will provide e-geos; TerraSAR/Tandem-X of DLR (German Aerospace Center) and Astrium
enhanced continuity of data so far provided by SPOT-5 and Landsat 7 and 8 Geo-Information Services; and Radarsat-2 of CSA (Canadian Space Agency)
and will offer data comparable to those provided by the U.S. Landsat system. and MDA Corporation.
CEOS has developed a long-term vision for how 3-3—and the volume of service delivery by current
it can expand its contributions to all phases of projects in DRM, users could consider how such
DRM. It anticipates contributions that are global volumes of data might be better exploited. Existing
in scope, even as they build on strong partnerships use for risk assessment and disaster preparedness
at local, national, or regional levels; that are user remains embryonic, despite evident potential.
driven; that address several hazard types; and that Further investment may be required to support
take into account all relevant EO-based capabilities new user communities and emerging partnerships.
available or under development. As part of this Looking at efforts to reduce disaster risk, existing
vision, and to demonstrate the benefits of EO data services have proved useful and have demonstrated
used in complement to more conventional data
the cost benefit of providing risk assessment based
sources, CEOS is implementing pilots defined
on satellite EO data. For some geo-information
with representatives of the user community
needs, additional research and development is
(scientists, civil protection agencies, local resources
required. For other needs the available products
management authorities, etc.) for floods, seismic
are mature, precise, and documented. However,
hazards, and volcanoes in 2014–2016.
currently it appears that the main obstacle to
Looking at the tremendous resources of new EO progress remains lack of awareness of what exists
missions—some innovations are described in box and what can be accomplished.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
97
98
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-6. Global Earthquake Model
Helen Crowley, Nicole Keller, Sahar Safaie, Kate Stillwell (GEM)
The Global Earthquake Model (http://www. • Collaboration. GEM is made up of people with
globalquakemodel.org/) is a collaborative effort a passion for contributing to the mitigation of
involving global scientists and public and private seismic risk, so collaborations have been built
stakeholders. Founded in 2009, GEM aims to build across sector, geography, and discipline.
greater public understanding and awareness of
Between 2009 and 2013, GEM made a significant
seismic risk, and to increase earthquake resilience
contribution toward advancing the science and
worldwide, by sharing data, models, and knowledge technology needed for global state-of-the-art seismic
through the OpenQuake platform; by applying GEM hazard and risk modelling, data collection, and
tools and software to inform decision making for risk assessment at the global, regional, national,
risk mitigation and management; and by expanding and local scales. These contributions include
the science and understanding of earthquakes. the following:
During the last five years, GEM has focused on four ISC-GEM Global Instrumental Earthquake
///
key pillars: Catalogue (released January 2013). This risk
///
assessment data set is a homogenous global catalog
• Trusted and credible science. Assessing of nearly 20,000 earthquakes. Archiving and
earthquake risk holistically requires reassessing records from 1900 to 2009, the catalog
multidisciplinary knowledge—seismology, represents the state-of-the art record for earthquake
geotechnical and structural engineering, locations and magnitudes.
economics, and social science—combined with
Historical Catalogue and Archive (released
the latest technology. GEM has brought this
/// ///
June 2013). This project archives almost a thousand
diverse scientific community together in various
earthquakes. Using the most detailed and up-to-date
scientific platforms which aim to achieve a
studies in the scientific literature, this archive spans
common language, while keeping discussion and
nearly a millennium, from the early Middle Ages
debate alive. (1000 CE) to the advent of instrumental recording
at the start of the 20th century (1903 CE). The
• Wide impact and public good. GEM has
catalog itself provides detailed parameters on 827
focused on trying to bridge gaps—both from
earthquakes of magnitude greater than 7 across the
science to practice, and from knowledge to
globe; see figure 3-6 for a sample image.
action.
Geodetic Strain Rate Model (released February
/// ///
• Openness and transparency. The OpenQuake
2014). This model estimates deformation rates on
platform is being designed to allow users to the Earth’s surface based on measurements from
evaluate the impact of any assumption on results, the global network of geodetic instruments using
implement alternative data or models, and the Global Positioning System (GPS). Building
explicitly account for uncertainty. Source code of upon a data set of more than 18,000 GPS velocity
the software and tools is publicly accessible. measurements worldwide, the GEM Global Geodetic
Strain Rate model represents a fivefold expansion Global Exposure Database will be a multi-scale, Figure 03—6
of data from its 2004 predecessor. It features multilevel database that will be an integral part of A fuller picture of
global coverage and high resolution in actively the OpenQuake platform. It has been designed to seismic history
deforming regions. accommodate data at four levels of resolution, from is obtained when
national to individual-building scales. instrumentally
Active Faults Database and Tools (expected
/// ///
recorded events
release November 2014). This database assembles Earthquake Consequences Database
/// ///
are combined with
available national, regional, and global active-fault (expected release November 2014). This database events from historical
databases worldwide within a common repository. A captures a full spectrum of consequences from records (in pink).
capture tool has been developed to allow local and earthquake-induced ground shaking, landslides,
Source: GEM Historical
regional geologists to feed data on local active faults liquefaction, tsunamis, and fire following 66 Catalogue, Global
into the common database. historical earthquakes between 1923 and 2011. Earthquake Model, http://
www.globalquakemodel.
Global Exposure Database (expected release
/// /// Physical Vulnerability Database (expected
/// ///
org/what/seismic-hazard/
November 2014). The first open database of global release November 2014). This data set contains historical-catalogue/.
buildings and population distribution is being more than 7,000 existing and new fragility and
built through the GED4GEM project. GEM’s vulnerability functions (“damage curves”) from
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
99
100
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
around the world, derived from empirical, analytical, across the globe. It features 13 building attributes,
and expert-opinion methods, and rated for quality. including building occupancy, roof, and wall
The functions form the basis for estimating material. Selected characteristics are those affecting
damage and loss in terms of fatalities and building the seismic performance of a building, and also
repair costs. those used to describe exposure. This “common
language” will facilitate global collaboration to
Socio-Economic Vulnerability and Resilience
///
understand the diversity and seismic vulnerability
Global Database (expected release November
///
of buildings.
2014). This global database contains indicators
measuring social vulnerability, resilience, and Physical Vulnerability Guidelines (expected
/// ///
economic vulnerability at various scales. The
release June 2014). These guidelines apply to the
data are structured and sub-structured according
development of empirical, analytical, and expert
to a taxonomy that accounts for eight major
opinion–based vulnerability functions.
categories (population, economics, education,
health, governance and institutional capacities, Inventory Data Capture Tools (released January
/// ///
the environment, infrastructure and lifelines, and 2014). This set of open source tools captures data
current indices). on buildings (inventory) both before and after
an earthquake. Tools range from those capable
Ground Motion Prediction Equations (released
/// ///
of extracting footprints from satellite photos, to
December 2013). This initiative conducted a critical
tablet or paper forms suitable for field use. After
appraisal of ground motion prediction equations
(GMPEs) in published scientific literature from validation, the captured data can contribute to the
around the globe. Defining a clear and reproducible Global Exposure Database or the Global Earthquake
process for the selection of ground motion models Consequences Database.
across all tectonic settings worldwide, the initiative
Socio-Economic Vulnerability and Resilience
proposed a set of 10 GMPEs for use in seismic
///
Tool Set (expected release November 2014). This
hazard analysis in subduction, active shallow crust,
///
set of tools assesses integrated earthquake risk by
and stable continental regions around the globe.
combining indices of physical risk with indices of
Building Taxonomy (released December 2013).
/// ///
socioeconomic vulnerability and resilience; the
This taxonomy categorizes buildings uniformly latter allows users to incorporate local knowledge.
CHAPTER
03
3-7. Global Probabilistic Risk Assessment: A Key Input
into Analysis for the 2013 and 2015 Global Assessment
Reports
Manuela Di Mauro (Risk Knowledge Section, United Nations Office for Disaster
Risk Reduction)
The Global Assessment Report on Disaster Risk measure underlying risk factors and drivers. This
Reduction (GAR) is the UN flagship publication on approach, however, had significant limitations; the
global disaster risk and disaster risk management. short historical record used meant that temporal
Building on the UNDP (2004) report on global and spatial information was limited, and records of
risk patterns and trends and on the World Bank’s consequences lacked detail.
report on natural disaster hot spots throughout
the world (Arnold et al. 2005), the GAR has been A probabilistic approach minimizes these
produced every two years since 2009 by the UN limitations. It uses historical events, expert
Office for Disaster Risk Reduction (UNISDR). Each knowledge, and theory to simulate events that
report is based on original research and a global can physically occur but are not represented in
assessment of risk from natural hazards. Since 2013, the historical record over the past few decades. A
this GAR global risk assessment has been carried out probabilistic approach can generate a catalog of all
following a fully probabilistic approach applied at possible events, the probability of occurrence of
global scale (UNISDR 2013a). The research carried each event, and their associated losses. For these
out for the 2013 assessment (UNISDR 2013b) and reasons, a probabilistic risk assessment approach
for the 2015 assessment involved contributions from was used for GAR13, which began development
world-leading institutions. From this research,
33 in late 2011, and it is being further developed
original data have been produced, new hazard for GAR15. This approach delivers a number of
models have been built, and existing hazard and key outputs:
risk modelling tools have been upgraded, with all
• Global stochastic hazard catalogs of earthquakes
outputs peer-reviewed.
and tropical cyclones that include their spatial,
Rationale for the probabilistic approach to temporal, and intensity characteristics, and their
risk assessment. The 2009 and 2011 GAR took an associated losses
historical approach to risk assessment. Researchers
• Regional probabilistic models for riverine flood
looked at hazardous events and their consequences
and agricultural drought
over the last 30 years and derived exposure and
vulnerability parameters (UNISDR 2009; UNISDR • A global exposure database
2011). They then used these parameters to estimate
losses for any given year from 1970 to 2010. These • Loss exceedance curves for each hazard at the
results were then used to produce a proxy of country level, which provide an estimation of
current risk and past trends by region. The main the average annual loss (AAL) and the probable
strength of this model was its capacity to reveal and maximum loss (PML) for a given return period
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
101
102
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
The flood, earthquake, and tropical cyclone risk their investment planning, and to work with
assessments were carried out using the CAPRA risk governments to reduce the risk for the country in
modelling suite (www.ecapra.org). which they plan to invest.
Applications of the global risk model results.
/// ///
• For organizations representing small and
The aim of the GAR global risk assessment is to medium enterprises (the commercial entities
produce an order of magnitude of the risk at global that are usually most affected by disasters),
scale as a basis for advocating for investments in results can offer a broad estimation of how major
disaster risk reduction. Thus the GAR global risk hazards would translate into direct losses. This
assessment’s results should not be downscaled
information can in turn encourage businesses to
to a local level and do not render other types of
assess their particular risk and governments to
risk assessments unnecessary. Instead, the GAR
adopt DRM strategies.
global risk assessment advocates for national and
subnational risk assessments using consistent The Global Exposure Database. The Global
/// ///
approaches and highlighting estimates of hazard, Exposure Database (GED)—with a 5km x 5km
exposure, and risk at national level. cell resolution (figure 3-7)—was developed for
GAR13 by CIMNE and Associates and United
The results from the GAR global risk modelling have
Nations Environment Programme–Global Resource
a variety of uses:
Information Database (UNEP-GRID). The GED
• They can be used by government officials and includes the economic value and number of
ministries as evidence to support the funding residents in dwellings, commercial and industrial
of higher-resolution risk assessments and can buildings, and hospitals and schools in urban
encourage countries to optimize their disaster agglomerations (De Bono 2013). The physical
risk management portfolios. areas were defined using an urban mask based on
MODIS land cover (Schneider, Friedl, and Potere
• For governments engaged in transboundary and 2009) and were divided into rural, minor urban,
regional partnerships implying mutual support and major urban areas. Population in urban areas
and collaboration in case of disasters (e.g., was extracted from LandScanTM (ORNL 2007).
ASEAN), they can be used to provide an overview Building classes and percentages for each country
of the risk levels of the partner countries. were derived from various sources, including the
World Housing Encyclopedia, detailed in WAPMERR
• They can show international organizations (2013). The economic value was calculated through
(international financing institutions, the UN, analysis of income levels and education levels, with
NGOs, etc.) how disasters are likely to affect downscaled nationally produced capital based on a
different countries, and can thus form the gross domestic product (GPD) proxy. Further details
basis for strategic definition, programmatic of the exposure analysis are in De Bono (2013);
prioritization and planning, budgeting, etc. WAPMERR (2013); and CIMNE et al. (2013).
• They can be used by investors to gain an For the 2015 release, the GED will be enhanced
understanding of the overall level of risk, and to enable inclusion of other initiatives, such
thus the potential losses, that a country faces as GED4GEM (see box 2-5 and section 3-6 for
from specific hazards. They can be a means more information), as well as future population
of encouraging investors to perform detailed distribution models, a building-type pilot study,
risk analysis, to budget for DRM as part of and critical facilities, should these become available
at a later stage. Additional improvements for 2015 • Improvements in the building class distribution Figure 03—7
include the following: at national level and for large countries (e.g., Example of the 5km x
China and United States) to subnational levels 5km grid constituting
• The ability to account for both urban and rural
(e.g., administrative level 1). the exposure
populations and buildings when calculating
database for GAR13.
human and economic losses. This will involve • System performance improvements in functions
new geospatial layers defining urban areas, such and algorithms that will support the increased Source: UNISDR.
as the global built-up area layer developed by the data volume.
European Union Joint Research Centre.
Earthquake. For GAR13, the stochastic earthquake
/// ///
• The flexibility to replace the LandScanTM event set (location, depth, frequency, and
data with gridded population supplied by an magnitude) was built considering principal seismic
alternative source. This makes it possible to avoid sources, tectonic regions and seismic provinces,
any constraints to data distribution linked to and historical earthquakes from the U.S. Geological
Survey National Earthquake Information Center
proprietary licenses.
catalog. Analysis was undertaken using the CRISIS
• Inclusion of socioeconomic parameters, based on 2012 earthquake modelling software (Ordaz et
income, employment, etc., to the most detailed al. 2012; CIMNE et al. 2013), which is compatible
level possible from subnational data. with the CAPRA modelling suite. The results are
expressed in terms of ground shaking (spectral
• A downscaled 1km x 1km GED in coastal areas acceleration) in a 5km x 5km grid for each event.
for the calculation of tsunami risk and the The combination of the modelled losses for each
integration of storm surges in the tropical building class in each cell of the exposure grid is
cyclone risk assessment. used to calculate the seismic risk for the cell.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
103
104
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
For the 2015 GAR global risk assessment, the the streamflow data. The calculated probabilistic
earthquake model will be improved using the discharges were introduced to river sections,
products developed by the GEM foundation, whose geometries were derived from topographic
including the new set of ground motion prediction data, and used with a simplified approach (based
equations and the new historical seismicity catalog; on Manning’s equation) to model water levels
for more detail on these products, see section 3-6. downstream.34
Tropical cyclone. GAR13 assessed tropical cyclone
/// ///
Improvements in the 2015 release include
risk using stochastic cyclone tracks generated from the following:
historical track information from the IBTrACS
database of the National Oceanic and Atmospheric • Updates to the Global Streamflow database,
Administration (NOAA). The track information was and definition of new approaches to extracting
integrated with data on global topography (derived hydrological and climatic information from
from NOAA) and terrain roughness (derived by the database
integrating European Space Agency GlobCover and
• Consideration of the influence of dams on the
Socioeconomic Data and Applications Center data
different streamflow conditions, with particular
sets) to estimate surface-level winds over land using
attention to extremes
the hurricane model of CAPRA (CIMNE et al. 2013).
• Updates to the model’s regionalization through
The tropical cyclone risk model for GAR13 did
not consider storm surge, even though this can a reworking of the concept of homogeneous
contribute substantially to the losses caused by region with respect to more detailed metrics
this hazard (as Typhoon Haiyan in the Philippines (e.g., reweighted area on the basis of rainfall
in 2013 made clear). Storm surge will therefore volume contribution, seasonality, and time
be included in global risk assessment for GAR15. series variance)
GAR15 will also aim to implement improvements
Tsunami hazard. The global tsunami modelling
in tropical cyclone modelling highlighted
/// ///
carried out for GAR13 constituted a significant
in a peer-review process lead by the World
improvement to the first global-scale tsunami
Meteorological Organization.
hazard and exposure assessment, carried out
Riverine flood. A new, fully probabilistic Global
/// ///
for GAR09. In comparison with the previous
Flood Model was developed for GAR15 by the CIMA study, GAR13 provides a more complete coverage
Foundation and UNEP-GRID. of tsunamigenic earthquake sources globally
(developed by the Norwegian Geotechnical Institute
The GAR13 flood model calculated flood discharges and Geoscience Australia).
associated with different return periods for each
of the world’s major river basins, based on flood The GAR13 model uses two methods, one based on
discharge statistics from 7,552 gauging stations scenario analysis and one based on a probabilistic
worldwide. Where time series of flow discharges method known as Probabilistic Tsunami Hazard
were too short or incomplete, they were improved Assessment (PTHA) (Burbidge et al. 2009). The first
with proxy data from stations located in the same method now uses better input data and is applied for
“homogeneous region.” Homogeneous regions more sources than in the GAR09 model. The second
were calculated taking into account information method has been applied for the Indian Ocean and
such as climatic zones, hydrological characteristics the southwest Pacific using research and analysis
of the catchments, and statistical parameters of undertaken by Geoscience Australia (Cummins
CHAPTER
03
2009; Thomas and Burbidge 2009). It calculates a The next advance will be to improve the set of
set of synthetic earthquakes to obtain a distribution vulnerability functions that capture regional
of possible run-up heights rather than using one variations in construction practices. For GAR15,
scenario per location, and it allows for a robust regional vulnerability curves will be adopted
determination of the return period. for East Asia, Oceania, and the Pacific Islands,
through consultation with local experts lead by
For GAR13, the tsunami hazard was calculated Geoscience Australia under its existing international
based on earthquakes with a 500-year return development programs (Sengara et al. 2010, 2013;
period—those earthquakes that are expected to Bautista et al. 2012; Pacheco et al. 2013).
contribute most significantly to tsunami risk. For
Risk assessment for earthquake, flood,
GAR15, a fully probabilistic model will be developed
///
and tropical cyclone. For each building class
through application of the PTHA method globally,
///
associated with a grid point, the risk is calculated
in partnership with Geoscience Australia and the
using CAPRA by assessing the damage caused by
Norwegian Geotechnical Institute.
each of the modelled hazard events.
Volcanic hazard. The Global Volcano Model
Because the model considers different events, each
/// ///
is working on an initial global assessment of
grid point can be associated with a probability
probabilistic volcanic ash hazard, using an updated
distribution of hazard intensity for certain return
version of the model developed at the University of
periods. As each point of the vulnerability curve
Bristol. The model employs stochastic simulation is itself a probability distribution, a different
techniques, producing a large number of potential probabilistic distribution of damages is calculated in
scenarios and their relative ash dispersal patterns each grid point for each event and for each building
(Jenkins et al. 2012a, 2012b). In addition, a regional- class. A distribution of losses is therefore calculated
scale probabilistic volcanic ash hazard assessment for each grid point, for each modelled event, and for
is being undertaken using an innovative approach each building class.
developed by Geoscience Australia. Building upon
This analysis produces an average annual loss
existing modelling methodologies (Bear-Crozier
metric, which estimates the loss likely every year
et al. 2012), this approach emulates hazard for
due to a specific hazard. As the GAR global risk
ash-producing volcanoes in the Asia-Pacific.35 A risk
assessment is performed at global scale, the AAL
calculation using the CAPRA platform will also be
assessed should be read as an order of magnitude
piloted; this approach combines the probabilistic
estimate for the potential recurrent extent of
volcanic hazard results and vulnerability models
losses in a country. The assessment also produces a
developed by Geoscience Australia with exposure
probable maximum loss metric, which estimates the
data from the GAR Global Exposure Database.
loss expected for long return periods—for example,
100, 200, or 500 years (depending on the hazard and
Vulnerability functions. The vulnerability
/// ///
the needs of the stakeholder). For GAR13, the return
functions used for the GAR13 global risk assessment
period of 250 years was used to assess the PML.
are based on those developed for the U.S. Federal
This corresponds not to a loss that will happen once
Emergency Management Agency’s Hazus-MH, also
every 250 years, but to an event that has 0.4 percent
taking into account different resistant construction
chance of occurring in any year.
qualities and the level of countries’ development
(which affects, for example, the completeness of and It should be recognized that all results are uncertain.
adherence to building codes). The uncertainty arises from assumptions and data
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
105
106
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
sets used in the assessment of the exposed value, the The deterministic approach developed for GAR13
simplifications necessary to model the hazards at analyzed the Normalized Difference Vegetation
global scale, and the use of vulnerability curves that Index, which is derived from 10 years of satellite
are not country-specific. However, for the purposes imagery. This data set, which combines data on land
of global-scale analysis and country-to-country use and agricultural information, provided a regional
comparisons, the level of uncertainty is considered assessment of drought frequency. This methodology
acceptable. These results should thus be considered is useful in that it draws on easily available data and
an initial step toward understanding the extent gives a general overview (Erian et al. 2012). Kenya
of disaster losses that a country might face and and Somalia will feature as case studies in 2015.
toward determining further actions, such as detailed
An alternative approach undertakes probabilistic
country and subnational risk assessments.
analysis of the relationship between crop losses and
Landslide hazard and risk. Analysis in GAR09
/// /// precipitation, temperature, and soil conditions. The
showed that 55 percent of global mortality risk technique is based on modelling the water content
from landslides is concentrated in the Comoros, needed by the soil to sustain vegetation, which is
Dominica, Nepal, Guatemala, Papua New Guinea, done by representing the relationship between water
the Solomon Islands, São Tomé and Príncipe, requirement, evapotranspiration, rainfall (satellite
Indonesia, Ethiopia, and the Philippines. These derived), soil water-holding capacity, etc. The
countries also account for 80 percent of the deficit in water content at critical times of the year
exposure at risk of landslide (Peduzzi et al. 2009). (i.e., when germination occurs) and for prolonged
The landslide susceptibility is a result of terrain periods of time translates into crop losses, which
slope, soil and geology type, soil moisture content are also determined stochastically by relating
(resulting from rainfall), and seismicity. Given known water deficits with data on crop losses. Once
the localized nature of this hazard, a probabilistic these relationships are established, it is possible to
approach at a global scale is problematic; however, produce a synthetic time series of crop losses.
a number of case studies of countries highly prone
This stochastic water content event set was used
to landslide were undertaken by the Norwegian
to determine average annual crop losses and
Geotechnical Institute (NGI 2013).
the probable maximum crop losses for different
Landslide risk in Indonesia and El Salvador was return periods (Jayanthi and Husak 2012). This
assessed in 2011 and 2013, respectively. The El probabilistic approach will be applied to other
Salvador model produced a detailed susceptibility countries, possibly including different regions
analysis, which was overlaid by population in Africa, and will be improved based on peer
distribution, to highlight high-risk areas. For 2015, reviewers’ comments. Future work will also include
the landslide hazard and risk will be calculated for climate change scenarios based on changes in
high-risk countries such as Italy and the Philippines, seawater temperatures.
and systematic improvements will be made in
To improve the transparency and the dissemination
the analysis.
of the results, the GAR global risk assessment
Agricultural drought hazard and risk. The
/// /// follows an open data policy. The results and data
GAR has used both deterministic and probabilistic produced within the GAR global assessment reports
approaches to analyze the complex phenomenon of are available for viewing and downloading at www.
agricultural drought. preventionweb.net/gar.
CHAPTER
03
3-8. Global Water-related Disaster Risk Indicators
Assessing Real Phenomena of Flood Disasters: Think
Locally, Act Globally
Toshio Okazumi, Sangeun Lee, Youngjoo Kwak, Gusyev Maksym, Daisuke
Kuribayashi, Nario Yasuda, Hisaya Sawano (International Centre for Water
Hazard and Risk Management)
Water-related disasters, including both flood and flood events often do not highlight the potential
drought, continue to pose threats globally. Although intensity and therefore potential impact of the
preventive strategies have been devised to address event.
this risk, especially in the years since the Hyogo
Framework for Action (HFA), important steps still 3. They must take into account the
///
need to be taken to guide DRM. effectiveness of water infrastructure. ///
Global-scale hydrological models generally ignore
Water-related risk assessments do exist, but none
the effectiveness of dams, reservoirs, levees, etc.
is without limitations. Credible water-related
disaster risk indicators need to meet five particular This practice produces inaccurate indicators
challenges (ICHARM 2013): and fails to emphasize governments’ efforts to
protect people from floods. (This issue is further
1. They must represent the real phenomena.
discussed in section 3-23.
/// ///
Categorizing data and proxies on an ordinal scale
creates indicators that lack transparency and 4. They must use meaningful proxies for
///
physical meaning. vulnerability. Using poverty-related proxies
///
2. They must evaluate flood hazard in terms
///
such as GDP per capita or a national wealth
of the frequency and intensity of the index to represent vulnerability assumes a clear
physical phenomenon. Hazard assessments
///
relationship between poverty and flood risk,
that examine the frequency of occurrence of though one has not been established. Nor does
PAMPANGA CHAO PHRAYA TONE Table 03—1
River length (km) 265 1,100 322 Basic Characteristics
Basin area (km2) 10,540 163,000 16,840 of the Three
Population 5.8 million 23 million 12 million River Basins
Percentage of national population 6.8 40 10
Sources: JICA (2011) for
River bed gradient in the midstream Pampanga; JICA (2013) for
1/1,000 to 1/2,500 1/11,000 to 1/12,000 1/4,000 to 1/6,000
area Chao Phraya; MLIT (2006)
Average annual temperature at key 27.5oC, CLSU Munoz 28oC, Nakhon Sawan for Tone.
15oC, Maebashi station
gauging stations station station
Average precipitation (mm/year) 2,100 1,487 1,300
Peak discharge at key gauging About 6,900 m3/s,
About 1,880 m3/s, Arayat About 9,200 m3/s,
stations during the recent largest Nakhon Sawan, station,
station, 2004 Yattajima station, 1998
107
flood 2011
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
108
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Table 03—2 PAMPANGA CHAO PHRAYA TONE
Historical Flood Date of disastrous flood Aug-04 Jul-11 Sep-47
Disasters in the Three Inundation area (km2) 1,151 28,000 440
River Basins Affected people (persons) 757,000 13,500,000 600,000
120 totally damaged 2,300 totally damaged 23,700 totally damaged
Sources: JICA (2011) for Damaged houses (numbers)
1,200 partly damaged 97,000 partly damaged 31,400 partly damaged
Pampanga; JICA (2013), data
from Philippine Bureau of Affected agricultural area (ha) 71,772 1,800,000 177,000
Agricultural Statistics (2013) Fatalities (persons) 14 660 1,100
for Chao Phraya; MLIT
(2006) for Tone.
this approach provide guidance on how to protect have not been a serious threat to the mainstream
people from flood disasters (Wisner et al. 2004). river, but the tributaries often experience floods.
Nevertheless, large floods (those with a 100-year
5. They must clearly identify risk hot spots.
or greater return period) are anticipated to pose
/// ///
Identifying large target areas is insufficient
a significant threat to the social and economic
because affected people and fatalities may be
systems, given the area’s high population density
concentrated in risk hot spots that are small
fractions of the target area. and many links with domestic and overseas
industries. Impacts of these historical floods in the
The discussion below focuses on the third issue, three river basins are summarized in table 3-2.
concerning the inclusion of water infrastructure in
regional or national flood risk assessments, using Hazard assessment. To assess the flood hazard,
/// ///
three case studies. All three river basins discussed we utilized a simplified modelling technology to
are heavily populated, located in or near capital produce flood inundation depth (Kwak et al. 2012)
cities, and suffer frequent floods from tropical based on flood river discharge simulated with the
cyclones and typhoons. Table 3-1 summarizes the distributed hydrologic Block-wise TOP (BTOP)
overall characteristics of the three river basins. model (Takeuchi, Ao, Ishidaira 1999). Using global
data sets, this enabled us to apply a standard hazard
In the delta area of the Pampanga River, the flow
assessment methodology to various river basins in
capacity is so small that even low river discharges,
such as those of floods with a five-year return different countries for inundations associated with a
period, can cause flooding. Over the whole river selected return period such as a 50-year flood. This
basin, floods happen almost every year. approach had a number of advantages:
In the Chao Phraya River basin, four tropical • Data sets used (for precipitation, temperature,
cyclones and Typhoon Nesat in 2011 caused floods topography, soils, land use, etc.) were
that broke levees at 20 locations. For the period globally available.
from July to November 2011, flooding damaged
• Visual comparisons of 1-in-50 year flood events
industrial parks and affected residents’ livelihoods
over large areas inside and outside Bangkok. and historical flood inundation maps were
possible.
The Tone River basin experienced tremendous
damage from Typhoon Kathleen in 1947. After this • Consideration of dam effectiveness made it
event, the Japanese government strived to improve possible to account for individual dams’ flood
levees and construct dams and retarding basins. control capacity, which in turn made it possible
Although middle-sized discharges are common, they to reduce the 50-year flood discharge.
CHAPTER
03
Figure 03—8
Effects of water
infrastructure in
reducing flood
inundation depths for
50-year floods.
Source: International Centre
for Water Hazard and
Risk Management.
109
110
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Table 03—3 PAMPANGA CHAO PHRAYA TONE
Potential Flood Infrastructure Dam Without dam Dam Without dam Levee Without levee
Inundation Areas Potentially
in the Three River inundated area 1,320 1,360 14,310 18,060 130 890
Basins (considering (km2)
or omitting dams and Percent change 3.2 21 86
flood protection)
Table 03—4 PAMPANGA CHAO PHRAYA TONE
People Potentially Infrastructure Dam Without dam Dam Without dam Levee Without levee
Affected by Potentially
Flood Inundation affected 935,000 993,000 4,342,000 8,301,000 59,000 487,000
(considering or persons
omitting dams and Percent change 6 48 88
flood protection)
• It was possible to consider levee effectiveness comprehensiveness of the water infrastructure,
when calculating overflow water level (the including super-levees designed to protect the highly
overflow water level is calculated as the populated Tokyo metropolitan area.
difference between the flood water level of
the 50-year flood discharge and the bankfull Table 3-3 presents the respective values of flood
water level) and inundation depth for each grid inundation area change in the three river basins
globally. 39 considering water infrastructure. The projected
flood inundation area due to a 50-year discharge
This hazard assessment calculated changes in decreases in response to both types of infrastructure
inundation with and without water infrastructure (dam and levee). Above all, the Tone River basin
such as dams and levees; see figure 3-8. The case is noticeable, in that the reduction is as high as
inundation changes due to dams with flood control 86 percent owing to the effect of levees.
capacity are shown in panel a for the Pampanga
River basin and panel b for the Chao Phraya River To assess flood exposure, we assumed a critical
basin; inundation changes due to levees are shown inundation depth of 0.1m in view of the minimum
in panel c for the Tone River basin. resolution of topographical data and models.
We used the Global Population Database of
In the Pampanga River basin (panel a), the LandScanTM as a digital population map in order to
Pantabangan Dam makes a large change to estimate potentially affected people, i.e., population
inundation in its downstream area (see the enlarged at grid cells where 50-year floods are likely to cause
area in panel a). In the Chao Phraya River basin inundations beyond the critical level.
(panel b), three dams have very large flood control
capacities and reduce the inundation. In the Tone Table 3-4 shows the respective values of flood
River basin (panel c), the water infrastructure exposure change considering water infrastructure.
does not affect the headwaters but creates drastic The number of affected people decreases in
changes in the downstream area. This dramatic response to both dams and levees. The dams’
inundation change can be explained by the flood control capacity in the Pampanga River basin
resulted in a small decrease in flood inundation basin, levee infrastructure has the potential to
depths and areas, implying a 6 percent decrease significantly decrease inundation depths, implying
in the number of affected people. The decrease a sharp decrease—88 percent—in the number of
in affected people was much more noticeable affected people.
in the zone managed by the Pantabangan Dam
This analysis has clearly shown the importance
(see the enlarged areas figure 3-8, panel a). The of including water infrastructure in a flood risk
number of affected people was reduced by about assessment. Global and regional flood analysis that
30 percent. Dams in the Chao Phraya River basin fails to consider water infrastructure should be
could moderately decrease flood inundation depths treated with caution, as this type of analysis will
and areas, implying a 48 percent decrease in the inevitably result in an overestimation of both the
number of affected people. In the Tone River flood extent and impact to communities.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
111
112
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-9. Government-to-Government Risk Assessment
Capacity Building in Australasia40
A. T. Jones, J. Griffin, D. Robinson, P. Cummins (Geoscience Australia); R. U.
Solidum Jr. (Philippine Institute of Volcanology and Seismology); M. V. De
Guzman, A. Orquiza (Department of Foreign Affairs and Trade, Manila); S.
Hidayati (Badan Geologi); I. Meilano (Bandung Institute of Technology); J.
Murjaya (Indonesian Agency for Meteorology, Climatology and Geophysics)
During the last five years, Australia’s development and Geoscience Australia staff based in Canberra. As
cooperation program has supported a series of a result of these activities, the Greater Metro Manila
successful capacity-building activities for natural Risk Assessment Project (GMMA RAP) has produced
disaster risk assessment within neighboring one of the world’s first noncommercial multi-hazard
Southeast Asian countries. Although the modality of risk assessments for a megacity on this scale. (See
engagement between the agencies has varied in each the section 3-4 for more information about this
country context, the successes have been uniformly project.)
underpinned by strong, long-term bilateral
Background. The Australian government has
government-to-government (G2G) relationships
/// ///
invested in a variety of DRM activities, including
between Geoscience Australia and partner technical
efforts to strengthen the capacity of partner
agencies.
government technical agencies to map risks from
In Indonesia, the Jakarta-based Australia- natural hazards. The Australian development
Indonesia Facility for Disaster Reduction provides cooperation program draws on the technical
a forum for ongoing interactions between risk expertise of Australian government departments
assessment practitioners from the government to help developing country partners build their
of Indonesia, technical agencies, and Australian capacity to reduce disaster risk.
risk and vulnerability experts posted in Indonesia.
Geoscience Australia, the Australian government’s
Earthquake, tsunami, and volcanic hazard modelling
national geoscience agency, provides geoscientific
activities have increased government capacity to
advice and information to support governmental
understand the country’s natural hazard risk profile,
priorities. Geoscience Australia has had a long
and these gains have in turn informed significant
engagement in disaster mitigation and preparedness,
policy directives at the national level (e.g., the 2012
primarily through the quantitative modelling of the
Indonesian Presidential Master Plan for Tsunami
potential risks posed by natural hazards in Australia.
Disaster Risk Reduction).
Geoscience Australia has accumulated important
In the Philippines, capacity-buildings activities research, tools, and experience over the past 15 years
have been facilitated through remote bilateral as part of efforts to mitigate and prepare for the
relationships between the government of risks to Australian communities from earthquakes,
Philippines Collective Strengthening of Community tsunami, severe wind, flood, and volcanoes. This
Awareness on Natural Disasters (CSCAND) agencies work has included the development of open source
Figure 03—9
Earthquake
hazard map of
central Sulawesi
Province, developed
collaboratively by
Badan Geologi
High prone zone and AIFDR.
Moderate prone zone
Source: R. Robiana, A.
Low prone zone Cipta, A. Solikhin, J. Griffin,
and N. Horspool (Badan
Province area
Geologi and Australia-
Indonesia Facility for
Disaster Reduction).
software that can be used in quantitative modelling engagement with technical partners in Indonesia
of these hazards and risks (see “Hazard and Risk and the Philippines and explores the common
Assessment Tools” in part 2 for a review of relevant factors that have led to significant gains in capacity
software packages). Examples include the EQRM for in the region.
earthquake hazard and risk modelling (http://code.
google.com/p/eqrm/; Robinson, Dhu, and Schneider Indonesia. The AIFDR, in operation since 2009,
/// ///
2006) and the ANUGA for flood and tsunami represents the Australian government’s largest
inundation modelling (https://anuga.anu.edu.au/). bilateral commitment to reducing the impact of
For the past six years, as part of the Australian disasters and is a key part of Australia’s development
development cooperation program, Geoscience cooperation program in Indonesia.41 The program
Australia has been actively applying these hazard aims to strengthen national and local capacity in
and risk modelling tools and experience to capacity- disaster management in Indonesia and promote
building activities with partner technical agencies in
a more disaster-resilient region. Through its Risk
the Asia-Pacific region.
and Vulnerability work stream—led by Geoscience
Two of Geoscience Australia’s official development Australia—the AIFDR facilitates partnerships
assistance programs, with the governments of between Australian and Indonesian scientists to
Indonesia and the Philippines, have strengthened develop and demonstrate risk assessment methods,
the capacity of partner technical agencies to tools, and information for a range of natural hazards.
undertake natural hazard and risk modelling.
Though the two programs faced different challenges Two activities undertaken between 2009 and 2013
and were delivered through different modalities of illustrate this style of partnership: the Indonesian
engagement, both have been considered successful. earthquake hazard project, and a volcanic ash
This case study outlines Geoscience Australia’s modelling project.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
113
114
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
The earthquake project aimed to build the capacity In addition, significant improvements were made
of the Indonesian government to understand and in earthquake education and research; notably, the
quantify Indonesia’s earthquake hazard, including program for Graduate Research in Earthquakes and
earthquakes’ likely location, size, and frequency. A Active Tectonics was established at the Bandung
sample hazard map developed under this project is Institute of Technology. This program has become
shown in figure 3-9. Achievements include a revised a crucial resource for the government of Indonesia,
national earthquake hazard map for Indonesia, providing it with opportunities for earthquake-
designed for use within Indonesia’s building codes related education and collaborative research as well
as well as for more general risk assessment; the as independent scientific expertise.
capacity to maintain and update this hazard map in
A mixture of modalities was used in this program.
the future; and the production of over 160 real-
The primary form of technical assistance was direct
time ShakeMaps and impact forecasts to inform
training and mentoring of Indonesian scientists
emergency earthquake response. by Australian scientists who were based in Jakarta.
The project was implemented by a partnership of These were supplemented with additional technical
support from Canberra-based scientists through
Figure 03—10 Indonesian and Australian government science
short-term (one- to three-week) missions. Funding
Badan Geologi and agencies and academic institutions with additional
was also provided to allow Indonesian students to
Geoscience Australia technical and management support from AIFDR
study in Australia, and to allow Indonesian students
staff working staff (Indonesian and Australian scientists are
and academics to undertake research in Indonesia.
collaboratively on shown working together in figure 3-10). The major
probabilistic seismic deliverables were produced collaboratively with The second activity designed to build the risk
hazard maps for five key Indonesian agencies;42 and the interagency modelling capacity of Indonesian technical agencies
Indonesia. memorandum of understanding developed focused on volcanic ash modelling. The activity’s
among these agencies represented the first formal specific goal was to develop the capacity of Badan
Source: I. Maemunah
agreement on roles and responsibilities for Geologi to undertake probabilistic volcanic ash
(Bandung Institute of
Technology). understanding and managing earthquake hazard modelling using open source modelling tools. This
analysis in Indonesia. capacity allows the government of Indonesia to
rapidly assess the potential volcanic ash risk from intensive probabilistic and near-real-time Figure 03—11
Indonesian volcanoes. forecasted volcanic ash modelling into the future, Badan Geologi and
the government of Indonesia has invested in Geoscience Australia
The first phase of the activity focused on testing
high-performance computing equipment. staff collect volcanic
and assessing existing volcanic ash dispersal
ash samples from a
models and identifying the most suitable model • Badan Geologi has determined that further
roadside agricultural
for adaptation and use in Indonesia. The second engagement with Geoscience Australia and
plot of land
phase involved validating the chosen model against the AIFDR in volcanic ash modelling would be
approximately 10km
historical eruptions in Indonesia in order to assess highly beneficial. This work would likely focus
from the summit of
the accuracy and uncertainty in the simulations, and on building Badan Geologi’s capacity to produce
Ciremai volcano, West
implementing the model as part of a case study of regional and national scale map products from
Java, in 2010.
four volcanoes located in West Java. (Field work is volcanic ash modelling.
shown in figure 3-11.) The final phase of the activity Source:A. Bear-Crozier
primarily focused on building the capability to The success of this program was demonstrated (Geoscience Australia).
undertake near-real-time volcanic ash forecasting in early 2013, when Gunung Guntur erupted in
using the existing model. West Java. After increased seismicity was detected,
Indonesian volcanologists at the Volcanology and
All phases of this activity were successfully Geological Disaster Mitigation Centre assumed
completed, with the following results: responsibility for using the volcanic ash dispersal
models to gain some insight into how wind
• Badan Geologi has the capacity to use volcanic
conditions over the coming days could affect
ash modelling tools in Indonesia.
ash dispersal. Figure 3-12 shows the center’s ash
• The government of Indonesia has probabilistic dispersal model for the last historical eruption of
volcanic ash hazard information available for Guntur, in 1840.
four West Javan volcanoes and near-real-time
The volcanic ash modelling activity was
forecasting information available for two North
implemented almost entirely through short-term
Sulawesi volcanoes.
missions, conducted as a series of workshops
• Badan Geologi has the capacity to apply the hosted by both Badan Geologi and Geoscience
volcanic ash dispersion model using standard Australia. These workshops provided an important
computers. To undertake more computationally capacity-building environment for knowledge
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
115
116
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
transfer and intensive skill building. In the months of Metro Manila Communities to Natural Disaster
between workshops, Geoscience Australia staff and Climate Change Impacts) program, which
provided ongoing remote technical support to aimed to reduce the vulnerability and enhance the
Badan Geologi via email, telephone, social media, resilience of Metro Manila and selected neighboring
and videoconference. areas to the impacts of natural disasters and climate
change. As part of this larger program, Geoscience
Philippines. In 2008, a partnership between Australia worked with CSCAND agencies on the
Australia and the Philippines was formed with the Greater Metro Manila Area Risk Assessment
aim of reducing disaster risk. During the initial years Project43 (described in detail in section 3-4). This
of this engagement, Geoscience Australia worked collaboration contributed to the overall aims of the
with government of Philippines technical agencies program by increasing the capacity of Philippine
(the CSCAND agencies) on a project to strengthen government technical experts to understand how
natural hazard risk assessment capacity in the the potential risks and impacts of natural hazards in
Philippines. the Philippines can be assessed.
In 2010, Australia and the Philippines developed In contrast to the Indonesia initiative, the work in
the BRACE (Building the Resilience and Awareness the Philippines involved a multi-hazard probabilistic
Figure 03—12
The dispersal of
volcanic ash from
the last historical
eruption of Guntur in
1840, as produced by
the Volcanology and
Geological Disaster
Mitigation Centre.
Source: Adapted from Bear-
Crozier and Simpson 2011.
Note: A combination of
field data and volcanic
ash dispersion modelling
was used to calibrate
the dispersion model
for forecasting possible
future eruptions.
Figure 03—13
Modelled depths for
a flood equivalent
to that experienced
in Manila during
Typhoon Ketsana
in 2009.
Source: Adapted from
Badilla et al. 2014.
Note: The colored points
are measured depths for
comparison. Areas outside
the model region are
shaded semi-transparently.
risk assessment for a single megacity (Manila) produce, exposure databases and exposure
that included estimations of economic loss and information for analyzing natural hazard risk to
potential casualties. Significant coordination from and climate change impacts on the Greater Metro
the Philippine Office of Civil Defense and associated Manila Area.
agencies was needed to bring together the disparate
• Scientists within government technical agencies
agencies working on different hazards for the
are better able to assess the risk and impacts
same area.
from flood (figure 3-13), cyclone, and earthquake,
The key outcomes of the project are these: and better understand these risks in the Greater
Metro Manila Area.
• Manila and national government authorities have
base data sets (such as high-resolution digital The risk maps and models developed collaboratively
elevation models, captured through LiDAR) by the government of Philippines CSCAND
available for analyzing natural hazard risk and agencies and Geoscience Australia were delivered
to the mayors and planning officials of the Greater
climate change impacts.
Metro Manila Area and selected neighboring areas
• Government of Philippines technical specialists to inform their decisions about planning and
better understand, and are better able to mitigation for natural hazards.44
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
117
118
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Like the Indonesia volcanic ash modelling activity, common factors—the presence of trust and use of
the GMMA RAP was implemented almost entirely a catalytic approach—led to significant capacity-
through short-term missions comprising workshops building gains. However, neither of these factors
hosted by staff from both Geoscience Australia in is achievable without the right experts: building
Canberra and CSCAND in Manila. Evaluation of technical capacity through a G2G relationship
these programs has identified key success factors for requires individuals with the right combination of
building capacity (box 3-4). specific technical and social skills. The success of
Conclusions. Geoscience Australia’s long-standing
/// ///
these projects has relied upon credible, capable,
engagement in official development cooperation and committed professional staff members whose
programs with the governments of Indonesia and interest in their work goes beyond the purely
the Philippines has strengthened the capacity of technical issues to be resolved and includes an
partner technical agencies to undertake natural understanding of partner countries’ systems,
hazard and risk modelling. In both countries, cultures, and languages.
Box 03—4 Factors Leading to Successful Technical Capacity Building
The success of the programs involving collaboration by Geoscience Australia capacity. It does so specifically by showing the technical capacity the project
and the governments of Indonesia and the Philippines demonstrates that delivers; by demonstrating the added value of science; and by serving ad hoc
government-to-government cooperation is an effective mechanism for needs of counterparts. The catalytic approach fosters improvement in processes
technical capacity building. This observation is supported by recent research and cooperation between partners through ongoing successful activities of
that indicates G2G capacity building is more effective and sustainable than mutual benefit.
postgraduate training, learning by doing, and centers of excellence.(A) Two
broad factors led to successful capacity building in the G2G partnerships A critical first step in using the catalytic approach is for the agencies within
between Australia and Indonesia/Philippines: the presence of trust and use of a which capacity is being developed to identify their own capacity gaps
catalytic approach. (Simpson and Dhu 2009). Once these gaps are known, it becomes possible
to showcase the potential impact of science in addressing them—without
The G2G projects showed repeatedly the importance of trust as a foundation for taking on a structural role or starting work that in the long run should be
working relationships between technical experts. These projects suggest that done by the recipient agency. The initial steps should always involve gaining
trust develops for a variety of reasons: an understanding of how the existing system works or should work, so that
capacity-building efforts can focus on realizing or strengthening this system.
• Experts’ knowledge and skill make them credible. Technical experts’ ability
to communicate with and speak the same technical language as recipient Capacity-building interventions require a long-term, consistent, and predictable
partners—the language of science and engineering—is a critical first step in investment that facilitates repeated application of improvements, reinforcing
building credibility, which in turn is the basis for developing relationships of changes until they are sustainable. Strengthening public sector systems
trust. is complex and involves individual, institutional, and sectoral capacity.
Unavoidably, unforeseen complications emerge when systems are strengthened
• Government scientists have shared experience. Their common
or changed. These complications can be discovered only by working in line with
understanding of government operations and the science-to-policy cycle can
anticipated systems, and resolving challenges in line. The system is sustainable
solidify foundations of trust built through scientific expertise.
when it has been operating long enough for each step in the process to become
• G2G relationships are institutional and national. As such, they can be an standard and routine.
effective basis for long-term cooperation, diplomacy, and trust between
The focus for each of the activities outlined above is on realizing systems that
partner countries.
produce ever-improving DRM outcomes in some of the world’s most hazard-
• Personal agendas are absent. Officials solely delivering to a government prone nations. Capacity building is a long-term effort in this context, but a
mandate feel less pressure to seek a high profile or to publish project catalytic approach ensures that local capacity is enhanced and not replaced
findings under their own name, and are more willing to maintain a supportive or displaced.
role in the background.
(A) Scholarships are more effective at the individual level and centers of
The catalytic approach exemplified in the G2G projects described above focuses excellence are more effective at the national level, but G2G has proven to be
not on replacing or displacing capacity, but on building or strengthening most effective overall. See Lansang and Dennis (2004).
CHAPTER
3-10. Informing Disaster Risk Management Plans in 03
Aqaba, Jordan, through Urban Seismic Risk Mapping
Jianping Yan, Kamal Kishore, Zubair Murshed (United Nations
Development Programme)
Seismological and archaeological studies indicate Aqaba’s new status was expected to increase its
that Aqaba, Jordan’s only coastal city, is at seismic risk. To minimize the potential human and
significant risk of intensive earthquakes (figure financial losses from seismic hazards, the Aqaba
3-14 shows historical seismicity for the country as a Special Economic Zone Authority (ASEZA), the
whole). As many as 50 major events have occurred United Nations Development Programme (UNDP),
in the last 2,500 years, including one as recent as and the Swiss Agency for Development and
November 1995. At that time, DRM considerations
45
Cooperation launched a project to integrate seismic
were not included in city plans. risk reduction considerations into Aqaba’s economic
development in 2009.
In 2001, Aqaba was declared a special economic
zone, which opened the door for investment, Assessing risks and using risk information.
/// ///
especially in tourism- and trade-related services. Under this partnership, the Jordanian Royal
The anticipated urban growth associated with Scientific Society conducted a seismic hazard risk
Figure 03—14
Historical seismicity
in Jordan.
Source: Daniell et al. 2011;
Ambraseys 2009, 4333–55.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
119
120
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
assessment. In addition to producing tools for A key finding was the potential impact of the
quantifying the level of seismic risk affecting the earthquake on Jordan’s only seaport, through
city (usable by both scientists and legislators), the which most imports and exports pass. For example,
project supplied the evidence for an earthquake risk disruption of port activities for three months due to
management master plan and served as the basis damage or due to a focus on humanitarian activities
for an operational framework for earthquake risk could amount to US$420 million. This loss would be
reduction. nearly equaled by the predicted US$300 million loss
associated with a reduction in tourism.
The seismic hazard risk analysis focused on two
potential sources of earthquake hazard to Aqaba, the This earthquake scenario made clear that unless
first from the fault system that runs from the Wadi DRM considerations were better accounted for in
Araba fault, through the Aqaba fault to the Gulf of city planning, the potential impacts of an earthquake
Aqaba fault, and the second from an earthquake on would be serious indeed. In response, ASEZA took
the Dead Sea fault system (figure 3-15). steps to strengthen DRM in the city Aqaba. Among
the improvements that were made are the following:
A deterministic (impact) scenario from a maximum
magnitude earthquake of 7.5 on the Aqaba fault A new DRM master plan was prepared for the city.
section was produced, showing the impact on
A DRM Unit and multi-stakeholder coordination
people, buildings, and the economy. Key results are
committee were established within the ASEZA
presented in table 3-5. This analysis was developed
to ensure that all development work takes risk
from data on building distribution provided by the
reduction into account.
Aqaba Department of Statistics, Population and
Housing Census. Through this city assessment, the Jordanian Royal
Scientific Society strengthened its risk assessment
Analysis also pointed to temporal elevated changes
capacity and is now able to carry out seismic risk
in the risk associated with the tourist peak season,
assessments for other parts of the country, including
weekend, and/or Ramadan. Moreover, the hospital
the Irbid Governorate.
capacity at the time of the analysis was 206 beds
among three hospitals—a figure that clearly Using the plausible seismic risk scenarios, ASEZA
highlights challenges that would be encountered in has also established and trained community-level
the aftermath of an earthquake event, given that the emergency response teams, including search and
scenario predicted more than 1,900 people requiring rescue teams, to save lives in the event of a disaster.
treatment. The study also made estimates of the
restoration times for critical infrastructure and The Aqaba Development Company, a partner of the
transport systems, and determined that main and ASEZA, is now using the findings of the seismic risk
secondary roads would likely be disrupted for more assessment to make decisions about construction
than 40 days, and wastewater systems disrupted for projects and about allocation of land to new
almost a month. businesses.
The DRM Unit is now a focal point for coordinating
Economic analysis undertaken at Hashemite
stakeholders and integrating DRM into all policies
University (Al Waked 2011) provided a
and development planning. In partnership with
comprehensive view of the direct, indirect, and
UNDP, the DRM Unit has trained more than 200
secondary effects of this earthquake scenario.
officials to improve its capacity to plan, coordinate,
Findings are summarized in table 3-6.
CHAPTER
03
EFFECT ON BUILDINGS
Table 03—5
Building damage state Number of buildings Share of the total (%)
Seismic Risk
None 2,500 20
Scenario for Aqaba
Slight 3,600 30
(maximum magnitude
Moderate 2,300 20
7.5 earthquake)
Severe 2,500 20
Complete collapse 1,200 10 Source: Based on analysis
Total (in 2010) 12,100 100 of data from Aqaba
Department of Statistics,
Population and Housing
EFFECT ON PEOPLE
Census.
Human casualty class Number of people
Minor injury 2,500
Medium injury 1,300
Severe injury 600
Dead 600
Total casualties 5,000
Total affected population (in 2010) 106,000
and implement DRM responses more efficiently. The Figure 03—15
SYRIA
DRM Unit has also implemented a school awareness Jordan’s fault system.
campaign to educate students about personal safety
mediterranean sea Source: Institute for
in earthquakes. These initiatives are being replicated
Geophysics, University of
in other Jordanian cities to improve capacities Texas at Austin, http://www.
The Occupied
of local authorities to protect trade, tourism, Palestinean ig.utexas.edu/research/
Territories
and culture. projects/plates/data.htm.
JORDAN
Because of these achievements and its overall
progress in reducing disaster risk, the city of Aqaba ISRAEL
was recognized by UNISDR as a role model city
at the First Arab Conference on Disaster Risk SC 2: Wadi Araba
Reduction, held in Jordan in March 2013.
Lessons learned through this process to
///
SINAI
understand seismic risk in Aqaba. Five factors
///
were observed to contribute to the success of
this project: SC 1: Aqaba Fault
• A focus on decision making in risk assessment
SC 3: Gulf of Aqaba
• Use of evidence-based risk assessments
SAUDI ARABIA
• Use of local expertise to ensure the sustainability
and ownership of risk assessment activities
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
121
122
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
• Communication of the risk findings over the Several challenges yet remain, including the
course of the project implementation following: managing and collecting data about
natural hazards; applying microzonation maps to
• Extensive stakeholder engagement, and
specifically the use of stakeholder workshops to urban land-use planning; and continuing to build
disseminate knowledge and raise awareness of institutional capacity to analyze, assess, and manage
seismic risk in Aqaba disaster risks.
Table 03—6 IMPACT INDICATORS LOSS (MILLION US$) SHARE OF 2010 GDP (%)
Economic and Direct losses (wealth, compensation for
856 2.8
Financial Impacts death and disability)
of Earthquake Indirect losses (impact on output,
694 2.5
Scenario (magnitude emergency assistance)
7.5 earthquake) Secondary effects (account balance,
715 2.6
fiscal impact)
Source: Al Waked 2011. Total 2,265 7.9
CHAPTER
03
3-11. Tsunami Risk Reduction: Are We Better Prepared
Today Than in 2004?
Finn Løvholt, Carl B. Harbitz, Farrokh Nadim (Norwegian Geotechnical Institute);
Joern Birkmann, Neysa J. Setiadi, Claudia Bach (UNU-EHS); Nishara Fernando
(University of Colombo)
The Indian Ocean tsunami of December 26, 2004, magnitude to about 8, one order of magnitude lower
which was responsible for over 220,000 deaths, than the 2011 event (Geller 2011). Recent analyses
remains one of the deadliest disasters triggered have shown that a tsunami of this size may have a
by a natural hazard event (MunichRe 2013). It return period of about 500 years and should not
demonstrated the need for more research, improved have been a surprise (Kagan and Jackson 2013).
planning activities, awareness raising, and early
Today, from a scientific point of view, many of the
warning systems (UNISDR 2005). It also provided
tools for tsunami risk assessment are available, but
important lessons for developing the HFA and
it remains unclear whether they are actually used in
sharpened the commitment for its implementation
national and regional DRM efforts. This case study
(UNISDR 2009).
reviews the application of DRM methodologies for
In hindsight, the 2004 Indian Ocean tsunami should tsunami risk, with a focus on Southeast Asia, and
not have come as a surprise (Satake and Atwater in particular Indonesia and Sri Lanka, which were
2007). Events occurring two centuries ago provided severely affected by the 2004 Indian Ocean tsunami.
a warning sign that was remarked by scientists a
Progress in tsunami hazard assessment.
short time before the disaster hit (Cummins and
/// ///
Before the Indian Ocean tsunami occurred, and for
Leonard 2004). Recent paleotsunami deposits
a few years afterward, tsunami hazard assessment
provide evidence for past events in prehistorical
was mainly based on worst-case scenario analysis.
times (Jankaew et al. 2008). The 2004 Indian
As tsunamis having long return periods are believed
Ocean tsunami did introduce a paradigm change
to dominate the risk (Nadim and Glade 2006), the
in the sense that previous models for constraining
worst-case-scenario approaches may sometimes be
earthquake magnitudes along fault zones are now
appropriate, given the large uncertainty linked to
refuted (Stein and Okal 2007). As a consequence,
events having return periods of hundreds or even
mega-thrust earthquakes emerging from any of the
thousands of years. Furthermore, such scenarios are
large subduction zones in the world could no longer
often useful in areas that have a complex tectonic
be ruled out.
or geological setting, and that lack the information
The tsunamis that hit the Mentawai Islands in needed to conduct a proper probabilistic analysis
2010 and Japan in 2011 also revealed weaknesses (Løvholt et al. 2012a).
in the way society deals with tsunami hazard. The
The common metric associated with tsunami
2011 Tohoku tsunami was stronger than the design
hazard is usually the run-up height of the tsunami
standards of the tsunami barriers (Cyranowski
along a coastline. However, other metrics should
2011), and it revealed inadequacies in the Japanese
be considered. The Tsunami Pilot Study Working
hazard maps, which were largely based on historical
Group (2006) lists the following tsunami impact
earthquake records limiting the earthquake moment
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
123
124
Box 03—5 The Challenge of Multiple Tsunami Hazard Maps in metrics (intensity measures) that may be entered
Padang, Indonesia as parameters in tsunami models for assessment
of mortality, building damage, and forces on
The city of Padang, Indonesia, is a hazard-prone area, where the potential for
structures: tsunami flow depth; wave current speed;
a major earthquake and tsunami is well established. As part of the tsunami
risk reduction efforts in the city, international scientific groups as well as
wave current acceleration; wave current inertia
local institutions developed tsunami hazard maps as a basis for mitigation component (product of acceleration and flow
and evacuation planning. The maps’ information on hazard zones, however, depth); and momentum flux (product of squared
differed significantly due to the different approaches and data used by the wave current speed and flow depth and in many
mappers. As of August 2008, at least eight different hazard maps had been
circumstances the best damage indicator).
created.(A)
For hazard assessments, tsunami hazard modellers
To help stakeholders reach agreement on the most acceptable hazard scenario
and mapping approach for the city, the so-called Padang consensus meetings
take different approaches (even if all consider a
were convened. The scientists and local decision makers who attended the worst-case scenario), and assessments typically
meetings reached agreement on the following major issues: earthquake rely on different data sources for topography,
source scenario (e.g., most plausible worst case, multi-scenario probability bathymetry, and/or seismicity. These differences
approach), basis data (topographical, bathymetry), and modelling parameters
can result in users being provided with multiple
(e.g., consideration of roughness coefficient, consideration of buildings that
different tsunami hazard maps by different entities,
modify the tsunami wave energy, and potentially inundated areas). Although
some issues have yet to be resolved, the process has provided an opportunity
as is described in box 3-5. There is also a growing
to reconcile various state-of-the-art scientific findings and to showcase a recognition of the limitations of tsunami hazard
science-policy platform for advancing tsunami hazard information. mapping in areas with coarse resolution digital
elevation and bathymetry data sets; see box 2-4 for
(A) The figure is based on personal communication with GTZ, 2008.
discussion of this challenge.
Over the last decade, probabilistic methods
for estimating tsunami hazard have become
increasingly available. One important approach
is the Probabilistic Tsunami Hazard Assessment
(PTHA) method, which is largely based on the
well-documented approach to probabilistic seismic
hazard analysis originally proposed by Cornell
(1968). In recent years, PTHA has been used to
quantify tsunami risk in a number of areas, including
Japan, Australia, the West Coast of the United
States, and the Mediterranean (Annaka et al. 2007;
Burbidge et al. 2008; Parsons and Geist 2009;
Gonzalez et al. 2009; Thio, Somerville, and Polet
2010; Sørensen et al. 2012).
A crucial element in PTHA is the estimation of the
frequency of occurrence and maximum magnitudes
of large tsunami-generating earthquakes in each
source region. As the historical record for mega-
thrusts and other large earthquakes is very short
relative to their long recurrence times, it is not
CHAPTER
03
possible to constrain the occurrence and maximum In social sciences, the term vulnerability refers to
magnitudes of intense tsunamigenic earthquakes societal vulnerability, which is related to a society’s
directly using observed seismicity. Recent events exposure, susceptibility, and fragility, as well
such as the large 2004 Indian Ocean tsunami and its capacity to react to a hazardous event. A fair
the 2011 Tohoku tsunami demonstrate the reality of amount of progress has been made in recent years
tsunami risk. Past mega-thrust events along other in understanding the factors that influence societal
faults zones (such as those in 1960 in Chile and 1964 vulnerability and in developing relevant assessment
in Alaska) provide additional reminders of the need methodologies. For example, important vulnerability
for precautionary actions. factors were revealed by the Indian Ocean tsunami
in 2004, which devastated Indonesia’s Aceh Province
Progress in understanding exposure to
///
and many coastal districts of Sri Lanka. The
tsunamis. Mapping exposure in various hazard
///
especially high number of victims was due to the
zones exploits remote sensing data, geo-information near absence of preparedness measures appropriate
systems, and existing data for population, buildings, for such an extreme event.
critical facilities, etc. Population data are typically
obtained from available statistical data (population Populations need to be educated about tsunamis
census) at the lowest administrative level, while data and to be aware of hazard zones if evacuations are to
at the building level is normally obtained through be safe and effective. There was little knowledge of
remote sensing analysis (e.g., Taubenböck et al. tsunamis in the affected areas in Indonesia and Sri
2008). (A more detailed description of exposure Lanka prior to the 2004 tsunami. An Asian Disaster
data collection is in part 2 above.) Reduction Center survey (ADRC 2006) conducted
in October–December 2005 showed that most of the
In Padang the approach to exposure also considered Aceh population (88.50 percent) had never heard of
population groups with different evacuation tsunamis before the 2004 event. The others (11.50
(physical) capabilities. The data included an percent) said that they had heard of a big sea wave
activity diary that was part of household surveys, coming to land (recounted in Islamic storytelling)
as well as local statistics and building data from from family, friends, books, school, or television.
remote sensing (Setiadi et al. 2010). The analysis In Sri Lanka, less than 10 percent of respondents
emphasized differentiated exposure related to the reported having had any knowledge about tsunamis
spatial distribution of the city functions (building before 2004 (Jayasinghem and Birkmann 2007).
uses) and characteristics of the population, and This lack of knowledge led to what was identified
included factors such as work activities, gender, and as a main reason for the high number of fatalities:
income groups (Setiadi 2014). a lack of preparedness for such an extreme event
(Amarasinghe 2007). In addition, many people
Progress in understanding and assessing
ran to the beach to watch the setback of the sea
vulnerability to tsunamis. Vulnerability is a
(Amarasinghe 2007).
multifaceted concept that has different definitions
depending on the context and discipline. In natural Gaps and recommendations. In the actual
/// ///
sciences and engineering, vulnerability often planning of tsunami risk reduction activities, limited
refers to the physical vulnerability of the exposed use of hazard information (hazard maps) for buffer
population or elements at risk. Few reliable models zones and evacuation maps was identified. More
of physical vulnerability to tsunamis currently exist, advanced methodologies encompassing vulnerability
though substantial progress toward such models is factors have not been fully integrated into risk
being made. management activities. Continuous monitoring of
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
125
126
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
vulnerability to tsunamis is hampered by the lack While from a methodological perspective, important
of a centralized database, absence of information progress has been made in the last decade, the new
sharing among different agencies and local and methodologies are not widely applied in practice.
regional institutions, and lack of standardized Hazard maps, for example, are too often used only
common guidelines on tsunami vulnerability for establishing buffer zones when they could also
assessment. Furthermore, tsunami risk reduction aid in planning of construction and development
planning tends to focus on hard measures—for and in determining evacuation routes. More work
example, physical construction of evacuation is needed to develop indicators and criteria that
shelters—but seldom considers soft measure, determine the use of vulnerability information in
such as evacuation behavior and utilization of DRM, as well as to assess the effectiveness of key
facilities. Second-order vulnerabilities (in the case strategies and tools (like people-centered early
of relocation) also call for a detailed analysis and warning systems). These indicators and criteria
careful implementation of DRM, taking into account will ensure the application of the most recent
factors like the lack of land title and information findings on disaster risk and assist in choosing the
about resettlement decisions. appropriate risk reduction strategies.
CHAPTER
03
3-12. World Bank Probabilistic Risk Assessment
(CAPRA) Program for Latin America and the Caribbean:
Experiences and Lessons Learned
Fernando Ramírez-Cortés, Oscar A. Ishizawa, Juan Carlos Lam, Niels B. Holm-
Nielsen (World Bank, Latin America and Caribbean Regional Disaster Risk
Management and Urban Unit)
Urbanization in Latin America and the Caribbean urban planning. To address these challenges, the
has been dramatic; between 1950 and 2010, the Probabilistic Risk Assessment (CAPRA) Program
population living in urban areas increased by was developed by the World Bank (initially as the
approximately 600 percent. This increase is more Central America Probabilistic Risk Assessment
than twice the population growth experienced in the Initiative) in partnership with the Inter-
entire region (UN-HABITAT 2010). Urbanization American Development Bank, the UNISDR, and
has resulted in a greater concentration of people CEPREDENAC (Central America Coordination
and assets in areas exposed to several natural Center for Natural Disaster Prevention). The case
hazards, and has placed low-income groups study described here focuses on the experiences
disproportionately at risk (Lall and Deichmann and lessons learned during the implementation of
2009). By 2050, 150 million people in Latin America Technical Assistance Projects (TAPs) carried out
and the Caribbean region are expected to live in under the World Bank CAPRA Program from 2010
urban areas exposed to earthquakes. to 2013.
Decision makers, considering the combined During the first phase of CAPRA, which began in
effects of climate change, disaster risks, and 2008, the activities mainly focused on developing
rapid urbanization, are increasingly citing a the CAPRA software platform, a free and modular
lack of required information and awareness as a
risk modelling platform, through integrating
barrier to managing risk and fostering sustainable
existing software and developing new modules
development. Indeed, among decision makers
under a unified methodological approach (see
recently surveyed, 30 percent cited financial
Yamin et al. 2013). As part of the development and
considerations as a barrier to working on climate
testing of the CAPRA platform, more than 20 risk
change adaptation in their cities; 20 percent cited
assessment exercises were undertaken in Belize,
lack of awareness; and 20 percent cited a lack of
Costa Rica, El Salvador, Guatemala, Honduras, and
reliable information and knowledge (Fraser and
Nicaragua.46 The original objective of the CAPRA
Lima 2012).
Program was to transfer ownership of hazard and
Unfortunately, national and local governments risk information generated by consulting firms
continue to face significant challenges in generating to country governments for use in DRM policy
trusted, accurate, and targeted disaster risk and program design. It quickly became apparent,
information that can be readily understood and however, that risk information would be integrated
integrated into sustainable development and into decision making only if government institutions
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
127
128
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
were engaged more deeply and led the whole risk and volcanic eruption, and as a result have difficulty
assessment process. investing in and implementing DRM plans and
policies.
Thus in the second phase, which began in 2010,
the focus of the program shifted to supporting Volcanic risk—often overlooked because eruptions
government agencies in building their own are relatively infrequent, though the risk is
institutional capacity to generate, manage, and use significant for exposed populations—was prioritized
disaster risk information. This level of engagement by the Colombia National Planning Department for
was accomplished through the implementation a TAP in partnership with the World Bank. Galeras
of Technical Assistance Projects. Through a Volcano, one of Colombia’s 25 active volcanoes
partnership between government institutions and and the focus of the TAP, poses a significant risk to
the World Bank, and with the financing of donors neighboring towns. Three hazard zones around the
through the GFDRR and the Spanish Fund for Latin volcano cover a total of 888 km2. In the high-hazard
America and the Caribbean, technical agencies zone, there is more than 20 percent probability
leading the development of a TAP were trained that pyroclastic flows would completely destroy
in risk modelling and analysis using the CAPRA all property and kill any residents who did not
platform, and also received technical advisory evacuate. In the middle- and low-hazard zones, the
services for generating, managing, and using hazard probabilities are 10 to 20 percent and 10 percent,
and risk information. The scope for each TAP was respectively.
defined by the needs and priorities of each of the
A recent cycle of volcanic activity in Galeras took
institutions involved in the project. Under this
place between 1987 and 2010, with eruptions in 2010
approach, a lead government agency establishes
forcing the evacuation of 8,000 people. Despite
an interdisciplinary and cross-agency team for
this exposure, a number of municipal settlements
undertaking the risk assessment and discussing the
stretch into the high-hazard zone. The Colombian
results before using the generated information to
government is attempting to reduce this risk
inform specific DRM policies and/or programs.
through resettlement of populations living in areas
TAPs foster a hands-on approach to generating, at highest risk, but the success of this effort will
understanding, managing, and using risk depend on effective communication of trusted risk
information, and thus promote ownership of the information.
process and the results of the assessment. Between
Starting in March 2011, the TAP aimed to
2010 and 2013, eight TAPs were implemented in
complement the deterministic volcanic hazard
Chile, Colombia, Costa Rica, El Salvador, Panama,
analysis on Galeras, undertaken by the Colombia
and Peru, each focused on answering a different
Geologic Service (Servicio Geológico Colombiano),
risk-related question. Key features of three TAPs are
with additional vulnerability and risk evaluation.
described below.
Pyroclastic flows and volcanic ash were the focus
Understanding volcanic risk at Galeras
/// of the modelling activity. Modelling was based on
Volcano (Colombia). Colombia has a distinguished
/// a compilation of data on historical events, a newly
reputation for leading efforts to reduce the impacts developed exposure database, and vulnerability
of disasters, with significant progress made in the functions. The exposure database included
last 25 years. Despite these efforts, however, many information on population, essential buildings,
Colombian municipalities are struggling to analyze public services, and housing, among others, all of
the risks from hazards such as earthquake, flood, which was compiled into a GIS database.
CHAPTER
03
The program also delivered a series of technical All hazard information produced under this
workshops designed to introduce specialists to the TAP is being integrated into the National Public
CAPRA platform and to provide hands-on training in Investment System (Sistema Nacional de Inversión
developing and carrying out comparative analysis of Pública) database. This critical step facilitates the
the deterministic and probabilistic pyroclastic flows sharing of findings with the scientific community,
and volcanic ash risk assessment results. Experts government authorities, and the general public.
in charge of monitoring the Nevado del Huila and This information will be essential in general urban
Machín volcanoes, both of which remain active, also development planning and specifically in the design
participated in the training activities. and construction of infrastructure, schools, and
hospitals, as well as in mining. Moreover, local
Consolidating the national seismic hazard
engineers and researchers trained in the use of
///
model and understanding the risk of
CAPRA’s seismic and tsunami hazard module47 will
earthquake to schools and hospitals in Lima
be able to use and update the hazard model and
(Peru). Peru has a long history of seismic activity,
incorporate their finding in future analysis.
///
with historical records telling of an earthquake in
1582 that destroyed most of the city of Arequipa. An Under the second TAP, a seismic probabilistic risk
earthquake and associated tsunami in 1746 destroyed assessment was carried out for 1,540 schools and
the city of Callao and resulted in more than 5,000 42 hospitals in Lima and Callao. Currently, the
fatalities. A number of subsequent events have results of this study are being used by the Ministry
underscored the seismic risk in the country, with the of Education to complement the countrywide
most recent events—in 2007—causing significant infrastructure census and to design the National
damage and disrupting transportation, electrical, School Infrastructure Plan. Under this process,
and communication networks. the World Bank is providing technical assistance
to (a) extend the seismic risk assessment to other
In Peru, two TAPs since 2010 have addressed
cities; (b) design a structural retrofitting program;
different needs. The first TAP developed a
(c) conduct a cost-benefit analysis of existing
seismic hazard model at the national level and
structural retrofitting alternatives; and (d) define
was completed in 2012. Under the second TAP,
short- and medium-term investment for the
the seismic risk assessment focused on essential
infrastructure rehabilitation.
services and in particular on a probabilistic seismic
risk assessment for schools and hospitals in the The outcomes of the TAPs in Peru confirmed the
Lima Metropolitan Area. importance of institutional engagement throughout
the whole modelling process: they showed that the
The national seismic hazard model was developed
greater the level of engagement, the more likely it
by a team of researchers and engineers from the
was that targeted and strategic risk information
National Seismological Service of the Peruvian
informed DRM decision making.
Geophysical Institute (Instituto Geofísico del Perú).
Team members collected, generated, and analyzed Understanding and managing the risk to
///
historical seismicity data and tectonic data, and also water and sanitation systems (Costa Rica). ///
tested different attenuation models. These results Decision makers in Costa Rica have prioritized the
are currently considered as key inputs into the analysis of natural disaster impacts on infrastructure
updates of the national building codes and standards systems—that is, their focus is identifying the most
led by Peru’s National Committee for Building vulnerable parts of a system, realistically assessing
Codes and Norms. the expected damage at different locations and
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
129
130
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
the impact on populations, and setting investment investments for protecting access to water and
priorities with limited financial resources. The Costa sanitation after an earthquake. They also provide
Rican Water and Sanitation Institute (Instituto an evidence base to guide design and siting of new
Costarricense de Acueductos y Alcantarillados) has infrastructure. Moreover, under Presidential Decree
been working in partnership with the World Bank No. 36721-MP-PLAN, CAPRA has been established
to preserve and protect the water supply and to as the standard tool for DRM purposes and
establish a system that restores water and sanitation provides for an active government-sponsored risk
as soon as possible after an earthquake. Not only management approach.
does reducing interruption to water and sanitation
reduce costs after an event, it can also reduce the Lessons learned from the CAPRA Program
///
prevalence of waterborne diseases. experience about effectively developing,
communicating, and using risk information. ///
This TAP focuses on seismic risks to water and The CAPRA Program has continually evolved and
sanitation systems in the San José Metropolitan developed to incorporate lessons learned about
Area, the San Isidro area, and the Higuito area. the effective development, communication, and
Because these three systems differ in their demand use of risk information. Specifically, it takes into
levels and complexity, the project team had to account the need for risk information to be targeted,
consider a flexible approach that could work strategic, interdisciplinary, dynamic, accessible, and
anywhere in Costa Rica. For example, the San José formal. These characteristics are explained below.
Metropolitan Area includes 1.2 million residents;
draws water from riverine, spring, and artesian well Risk information is targeted and strategic when the
sources; and has primary and secondary pipework scope and specific objectives of the risk assessment
of 570km and 2,610km, respectively, as well as are consistent with the institutional needs and the
numerous water treatment plants, storage tanks, surrounding context (e.g., existing programs and
and pumping stations. The San José wastewater policies). The use of the resulting information from
system covers 85km of piping, pumping stations, risk assessment will define the level of detail of the
and treatment plants. In contrast, the Higuito area model and the resolution to be used.
is serviced by two streams, a small treatment plant,
Entailing as it does the involvement of many
eight storage tanks, and no wastewater facilities.
different institutions, disaster risk assessment is
The TAP began by collecting the input data sets a complex technical and institutional process that
required to understand seismic hazard, inventorying requires an interdisciplinary and cross-institutional
and categorizing water and wastewater systems framework.
and components, and defining appropriate
Risk information should be dynamic: it should take
vulnerability functions. The next step was to
advantage of new available data from hazard models
analyze scenario earthquake events; this made it
and should include changes in exposure from the
possible to understand what could happen to the
urban environment and sectoral infrastructure. Risk
system, highlight the most vulnerable sections
information must remain accessible to support
or components, and provide estimations of the
decision-making processes in each institution
maximum probable physical and economic losses.
leading a risk assessment, even as institutional
These results provided a baseline for the needs evolve. Moreover, good practice requires
formulation of a risk reduction program that that the owners of the risk information clearly
articulated short-, medium- , and long-term communicate with information users. They need to
explain their understanding of the main hypothesis, • When institutions participate in and lead risk
limitations, and uncertainties associated with the assessment processes, they are more likely to take
assessment, and they need to highlight input data ownership of the information and to be aware of
the information’s characteristics and limitations.
and information gaps and limitations in resolution
(so that the assessment may be improved upon). • The formal/official dimension of risk information
encourages institutional endorsement, which in
Information is formal when it is generated under an turn supports links between risk management
established institutional and legal framework. This policies and policies that address the risk’s
is a critical condition for the effective use of risk financial, social, and institutional impacts.
information in the design of public policies and risk
The CAPRA Program has found that well-targeted
reduction programs. Where information is formal
programs can help individual institutions strengthen
and has an official and legal status, decision makers their own capacity to use risk information and
are more likely to promote its use and application take decisions around it. However, from a broader
for specific purposes. perspective, the lack of technical capacities for
generating, understanding, and integrating risk
Experience proves the following: information poses a complex problem. Experience
in Latin America and the Caribbean reveals
• When created under an official legal and
that government agencies and institutions need
institutional framework, risk information is considerably more technical support in order to
considered legitimate for use in policy design and undertake risk assessments and produce needed risk
decision making in DRM. information.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
131
132
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-13. Detailed Island Risk Assessment in Maldives
to Inform Disaster Risk Reduction and Climate
Change Adaptation
Jianping Yan, Kamal Kishore (United Nations Development Programme)
With sea levels expected to rise and extreme communities living on distant islands. This event
weather events expected to increase in intensity, caused severe damage to physical infrastructure
Maldives, located in the central Indian Ocean, is of many islands and set back development. The
considered one of the world’s most vulnerable total damages were estimated at US$470 million,
countries. Eighty percent of all the islands that amounting to 62 percent of GDP. Of these, direct
make up Maldives are small, low-lying, and highly losses totaled US$298 million, which is 80 percent
prone to flooding and coastal erosion. More than of the replacement cost of the national capital
44 percent of settlements—home to 42 percent of stock. Most of the islands that were destroyed
the population—and more than 70 percent of all in the tsunami were highly exposed, with little
critical infrastructure is located within 100m of the or no coastal protection. The tsunami led
shoreline. As coastal erosion and pressure on scarce Maldives officials to seek financially sustainable
land resources increase, the physical vulnerability and ecologically safe settlement planning and
of island populations, infrastructure, and livelihood socioeconomic development of atolls, and to
assets will increase as well. integrate safety considerations into planning and
development.
The most significant driver of increasing
vulnerability to natural hazards and climate change Toward this end, the Safe Island Programme was
in Maldives is the absence of systematic adaptation established in 2006. Its goals were to protect the
planning and practice. Climatic risks and long-term islands from natural and other hazards; to rebuild
resilience are not adequately integrated into island and improve existing infrastructure and economic
land-use planning or into coastal development and facilities; and to build community resilience
protection policies and practice. to disasters through improved planning and
implementation of risk reduction investments. The
Safe Island Programme. In order to reduce the
/// ///
program emphasized that it was a multi-sectoral
environmental, economic, and social vulnerability effort and that it was to be seen as integral to all
of the widely dispersed population, in 2002 the development and planning (that is, not optional).
government of Maldives initiated a program to It held that decision making should be based on
encourage voluntary migration to larger islands. widespread consultation and participation, and
The program’s long-term objective was to reduce that human activities that damage the natural
the number of inhabited islands and consolidate the environment should be minimized and existing
population in fewer settlements across an identified damage rectified.
number of islands.
A key step in achieving the goals of the Safe Island
The 2004 Indian Ocean tsunami underlined Programme involved producing a short list of
the urgency of providing safe zones for isolated potential safe islands through consultation, using
Cross Section of an Island with Enhanced Mitigation Features
Figure 03—16
The safe
Mean sea level +1.3m +1.4m
island concept.
±00
Source: Ministry of Planning
Outer Harbour and National Development
Reef EPZ Elevated zone Elevated zone
+2.4m Maldives.
Drainage
Note: Elevated areas
are distributed across
the island and can be
used for emergency
Atoll evacuation; schools and
public buildings up to two
OU
stories in height can also
TE
be constructed in these
R
RE
areas. EPZ = Environment
EF
Protection Zone.
HA
EPZ
RB
OU
R
Drainage Area
IN
Elevated Area
NE
R
LA
Cross Section
GO
ON
both subjective and objective criteria. Once the particularly those related to health, communication,
short list of potential safe islands was agreed to, and transport, and would have a buffer stock of basic
detailed island-level assessments were planned and food and safe drinking water. Some of the enhanced
carried out. These assessments aimed at filling gaps mitigation features of safe islands are shown in
in knowledge and engaging with island officials and figure 3-16.
the general public.
Identifying Safe Islands. Detailed risk
/// ///
The goal was for islands developed under the assessments were undertaken for 10 islands
program to have appropriate coastal protection; short-listed for development as safe islands (see
improved communication and transportation figure 3-17). The assessments, carried out with
facilities; improved housing, infrastructure, and technical and financial assistance from UNDP,
social services; and adequate capacity/preparedness aimed to produce risk information that would be
to manage emergencies and disasters. For example, used to recommend specific mitigation options.
safe islands developed under the program would Key outcomes of the risk assessment included
have access to all basic services in an emergency, the following:
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
133
134
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
• Design and development of a risk information infrastructures, and the most important
process to generate critical inputs for the Safe economic sectors (mainly tourism and fisheries)
Island Programme
The project was carried out in three phases, starting
• Mapping of the selected islands’ overall hazard in January 2007:
context, including hazard event scenarios, their
Phase 1 involved hazard assessments of tsunamis,
probability of occurrence, and their geospatial
swells or high tides, wind storms, heavy rainfall,
extent, based on geological and historical disaster
storm surges, droughts, and earthquakes. These
data and simulated hazard data
were conducted for return periods of 25, 50, and 100
• Assessment of the islands’ full range of years for 10 islands (UNDP and RMSI 2006).
vulnerabilities (environmental, physical,
An environmental vulnerability assessment was
economic, social), with reference to multiple
undertaken at the same time. It examined the effects
hazard events and relocation
of coastal erosion and compiled available data
• Creation of comprehensive risk information on coastal erosion and hazards as well as related
for coastal ecological systems, building stocks, parameters. The assessment also included mapping
of coastal vegetation.
The exposure of buildings and infrastructure to
Figure 03—17
different hazards was calculated and “safe” buildings
The islands selected
on each island identified. This effort included
for detailed
determining the capacity of safe buildings to serve as
multi-hazard risk
shelters, and identifying where public infrastructure
assessment.
required retrofitting.49
Source: Ministry of Planning
and National Development
In the second phase, hazard data from phase 1
Maldives. were used to determine the vulnerability of fishery,
tourism, agriculture, small business, and home-
based industry sectors. This effort also included a
comparative analysis of livelihood opportunities and
relocation costs. A social vulnerability assessment
was undertaken that (among other things)
considered communities’ feelings about integrating
outsiders (since development of safe islands
requires relocating people).50
The third phase integrated all the information and
made recommendations for island-specific disaster
risk mitigation measures based on a cost-benefit
analysis. 51
Using risk information. The 2011 Strategic
/// ///
National Action Plan, which has been fully endorsed
by the government of Maldives, built on the
recommendations of the risk information and cost-
CHAPTER
03
benefit analysis. The risk information is providing • The islands were far apart from one another.
key inputs into the development of risk-sensitive Arranging the field survey across 10 dispersed
national building codes. The risk outputs have islands posed challenges for physical access as
been used to design and develop a national training well as information sharing.
program and to promote a national public awareness
• Data acquisition was not straightforward.
campaign for disaster risk reduction, early warnings,
Like risk assessments undertaken in other
and response actions. Launched in 2009 by the
developing countries, the assessment in Maldives
National Disaster Management Centre and Maldives
found data collection problematic. Maldives
Meteorological Service in partnership with the
lacked certain necessary data, including base
UNDP, the “Rakkaavethibiyya—Dhivehiraajje” (“Be
maps, long-term climatologic data, and historical
aware—Be prepared”) campaign was the country’s
event data; some necessary data were available
first public awareness campaign addressing disaster
but had to be purchased. For acquisition of
risk.
exposure data, field surveys were the only option.
There are still challenges to integrating risk
• Capacity and institutionalization were
information into the Safe Islands Programme, and
limited. The government of Maldives has
these have hindered progress toward the original
limited staff with the requisite skills and/or
vision. Specifically:
qualifications. Moreover, there is no institution
• The cost-benefit analysis showed that mitigation or organization specifically responsible for risk
investments must be approached with caution information and no unified data management
because there is significant uncertainty in the mechanism in place.
analysis and because the benefit-to-cost ratios
Lessons learned. The work in Maldives on risk
are not consistently positive or indeed very
/// ///
suggested the following lessons:
high. Therefore any change in the underlying
assumptions could result in a net loss • Evidence-based hazard risk profiles are critical
on investment. for carrying out cost-benefit studies of disaster
risk mitigation measures and for communicating
• A significant shift in focus needs to take place
risks to national stakeholders.
toward softer protection measures (e.g.,
mangroves) and other options to increase • Risk information can be an effective means of
resilience. engaging national stakeholders and decision
makers, and maintaining engagement from
There were also challenges encountered during the
the start to finish will increase the buy-in of
implementation of the risk assessment activities:
the results.
• Insufficient time was planned for project
• It is important to systematically document data
implementation. The duration of four months
collected and produced over the course of the
for project implementation was not sufficient,
project, including the implementation plans,
given the complexity of the analysis.
methodological framework, data and databases,
• Identifying local technical specialists was etc. This documentation provides critical inputs
difficult. The project struggled to recruit a to the institutionalization of the National
local structural engineer, resulting in significant Disaster Management Centre and lays down
reallocation of responsibilities, including the a solid foundation for the establishment of a
diversion of staff from other UNDP programs. national risk information system in the future.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
135
136
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-14. Malawi: How Risk Information Guides an Integrated
Flood Management Action Plan
Francis Nkoka, Pieter Waalewijn (World Bank)
Natural and man-made hazards cumulatively National assessment of drought and flood
///
affected 25 million people in Malawi between risk. Following the Standard Precipitation Index
///
1974 and 2003, with weather-related disasters methodology (McKee, Doesken, and Kleist 1993),
occurring on average once a year over the last the drought risk assessment measured daily
40 years (Government of Malawi 2010). Disaster rainfall from 45 meteorological stations in Malawi
risk in Malawi arises from a combination of to determine the precipitation time series. This
tectonic activity, erratic rainfall, environmental historical series was used to generate a 500-year
factors, and socioeconomic vulnerability driven by stochastic weather event set, which was in turn
widespread dependence on rain-fed agriculture, a
embedded in an agro-meteorological model to
narrow economic base, and extensive rural poverty
ascertain long-term drought frequency. The crops
(Government of Malawi 2011). With climate change,
considered most exposed to drought included
population growth, urbanization, and environmental
three types of maize and one type of tobacco.
degradation, the trend is toward more frequent and
Economic crop production (and losses) leveraged
more intense disasters.
data collected and shared by the Malawi Ministry of
The government of Malawi recognizes that Economic Planning and Development.
improved management of the natural hazard risk
The analysis, completed in January 2011, revealed
can lead to intensified, yet sustainable, agricultural
production, better transport links, and more secure that the central region of Malawi had the greatest
homes and livelihoods. With this vision of the potential for losses, and that losses associated with
country’s potential, the government of Malawi LMZ (local) maize were the highest for any crop;
partnered with the World Bank and GFDRR to the 50-year return period loss of LMZ maize in
undertake a national risk assessment (RMSI 2011). central Malawi was US$34 million. Across the entire
This proactive, evidence-based analysis sought country, the loss for this maize at this return period
to determine, quantify, and map Malawi’s flood was as high as US$62 million, and the average annual
and drought hazard potential both historically loss for this maize was US$6 million. Composite
and probabilistically, using average annual loss maize was found to be the most drought-resistant.
and probable maximum direct and indirect loss Losses associated with tobacco were considerably
as metrics. It was recognized that improved flood lower, with an average annual loss of US$1 million.
management in the Shire River Valley in particular
could significantly reduce entrenched poverty Flood hazard analysis used daily flow discharges
and potentially make the Shire Valley a national from 13 Malawi river stations over different two-
economic hub. With this in mind, the government year time periods, with ~90m resolution digital
of Malawi also commissioned a detailed flood elevation model, a digital river network, and HEC-
analysis of the Shire basin (Atkins 2012). This staged RAS flood modelling software. The Dartmouth
approach to understanding risk in Malawi— national Flood Observatory satellite images of the January
to local level—highlights the need for understanding 2003 flood event were used to calibrate the flood
of risk at many levels and for many purposes. extent. Flood extent maps were produced to show
CHAPTER
03
return periods of 2, 5, 10, 20, 50, 100, 200, and works currently in place are now considered
500 years. Exposure data consisting of population unsafe or unsustainable due to poor engineering
and dwellings (households)52, roads, railway, and practices. The Lower Shire River is the site of
agriculture (maize and tobacco) were then overlaid flood disasters nearly every year, and these cause
on the flood extent maps. Results reveal that, on damage to infrastructure that is never successfully
average, about 26,000 people and 6,000 dwellings repaired. These disasters require significant flood
are inundated each year at a cost of US$6.5 million, aid and other relief support to a region that is the
with the district of Chikwawa most affected.53 poorest in the country, and that already struggles
The average annual loss to roads, railways, and
with inadequate sanitation and limited access to
agriculture was found to be US$38,000, US$61,000,
clean water.
and US$19 million, respectively.
The Shire River is economically and environmentally
Economic analysis reveals that Malawi loses about
very important. It is the site of hydroelectric
1 percent of GDP per year as a result of drought,
schemes that generate 98 percent of Malawi’s
though during a 1-in-25-year drought, GDP can
electricity; it contains extensive fisheries and
contract as much as 10 percent. A 1-in-25-year
wildlife conservation areas; and it provides
drought can also significantly exacerbate income
poverty—that is, can cause an almost 17 percent freshwater for irrigated agriculture and for industrial
increase in poverty, which is equivalent to an and domestic uses. A better understanding of flood
additional 2.1 million people falling below the risk, and the mitigation of risk through targeted
poverty line. Malawi loses 0.7 percent of GDP per measures based on the findings of the assessment,
year as a result of flooding in the south—the part would help to improve agricultural production and
of the country where flooding is most severe. Since generally aid the population that lives in the area.
farmers in other parts of the country and export
The integrated flood risk analysis aimed to achieve
farmers typically benefit from higher prices during
the following:
southern flood events, the 0.7 percent contraction
in national GDP really does not reveal the significant • Construction and calibration of a hydrodynamic
localized impacts from flood.
model of the catchment capable of accurately
Lower Shire River basin study.54 Following
/// ///
predicting inundation of the floodplain for
the national-level study and other analysis (Shela extreme fluvial flooding. This model was
et al. 2008), a decision was made to undertake a developed so that it can be updated in the
comprehensive flood analysis of the Shire River future to improve accuracy and reliability as
basin. Approximately half a million people live in better data become available and can assess
the Lower Shire valley and are regularly affected the effectiveness of potential interventions to
by flooding and water pollution. The highest-risk mitigate flood impact.
areas in the Shire Basin are Chikwawa and Nsanje
districts, which are located in the lower section of • Simulation of floodplain inundation for 5-, 10-,
the basin, and Mangochi district, just downstream of 20-, 50-, 75-, 100-, and 500-year return period
the outflow from Lake Malawi in the upper section flood events, and for 100-year return period
of the basin, where flooding is caused when lake inundation considering change in rainfall
levels are high. patterns with climate change.
Flooding in the Lower Shire River often occurs • Production of flood maps of the catchment for
without warning, and some flood protection each of these design modelling scenarios.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
137
138
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
• Development of a framework for flood Physical data sets on topography, land use, geology,
forecasting and an early warning system in and soil type, as well as time series data, were used
Figure 03—18 (Left)
the basin. in the flood analysis. A variety of improvements is
1-in-100-year flood
being made to these data for future analysis:
extent (in pale blue)
• Development of guidelines for flood
around the Elephant • For topography, SRTM data were used, but these
mitigation measures.
Marshes of the Lower have inadequate vertical accuracy and spatial
Shire Valley, Malawi. • Building capacity of stakeholders involved in resolution to serve as the basis for detailed flood
flood management and development of an modelling and mapping. Higher resolution digital
Source: Atkins 2012.
institutional development plan. elevation is being developed for the catchment,
and the integration of these data will result in
Figure 03—19 (Right) The objectives were achieved by developing a Soil substantial improvements in model accuracy.
Flood zoning in the Conservation Service rainfall-runoff model (SCS
area of the Elephant • For flow and level data, sub-daily rainfall and flow
1986) using time varying rainfall data for different
Marshes based on data are now being used to improve hydrological
return periods (derived from depth-frequency
different return period modelling.
statistical analysis of daily rainfall), with input and
flood events.
flow data, where available, used to calibrate the • Observed water level on the Shire and its
Source: Atkins 2012. model. A sample flood map is in figure 3-18. tributaries should be used to provide calibration
CHAPTER
03
• Functional floodplain: land between the river at
data. Once limitations in the location of gauges
normal flow levels and the 20-year flood event
within the basin are addressed, better calibration
of the model will be possible. After defining flood zones, the assessment then
provided guidelines for risk-sensitive development
An assessment of the baseline flood risk to high-risk
within the different zones: for example, emergency
villages was used in conjunction with the economic
and other essential services should be located in
assessment of flood damage to assess the likely low flood hazard zones, water-compatible or less
benefits of implementing flood protection measures vulnerable development should be in high hazard
such as defenses, catchment improvement through zones, and a minimum of development should
reforestation, and flood storage. Key findings from occur within the functional floodplain. However,
this analysis include the following: agriculture could be promoted within the highly
productive floodplain area that was found to be dry
• Increase in forest cover to reduce flood depth
during five-year flood events.
in catchments should be applied on a case-by-
case basis, since the measure is not effective in Additional analysis and consultation based on this
every catchment. analysis led to development of the Shire Integrated
Flood Risk Management Action Plan. The plan is
• Flood storage options were found to be guided by three principles:
impractical and ineffective for events larger
than those having a 10-year return period. 1. Flooding is a natural process and a
These options appeared to reduce flooding in development issue. The action plan will work
toward a more detailed and robust understanding
more-frequent events, but the analysis was not
of flooding through improvements in input data.
conclusive and would benefit from analysis of
It will also identify where human development
higher-resolution LiDAR data.
and activities intersect with high flood risk
• Predicted changes associated with climate—such areas and implement measures (both structural
as a 12 percent increase in river flow—did not and nonstructural) that protect populations
result in a significant change in flood inundation from flooding and ensure effective response
along the river. However, changes may be to flooding.
more apparent with a higher-resolution digital 2. Flood management requires a whole-of-
elevation model. government/country approach and entails
partnerships between government agencies,
Based on the flood hazard and inundation maps,
donors, communities, land owners, and private
flood zones (figure 3-19) were defined with the
sector players. The action plan creates an
following zoning categories for the Shire River basin:
improved institutional structure and aims to
• Low flood hazard zone: land inundated in a 500- equip all stakeholders with the skills needed
to contribute to a holistic approach to flood
year flood event
risk management.
• Moderate flood hazard zone: land inundated in
3. A pragmatic and integrated approach
100- to 500-year flood events
to flooding includes flood management,
• The floodplain: land inundated in 100-year risk reduction, preparedness, response,
flood events and recovery. The action plan has identified
approximately 100 intervention measures under
• High flood hazard zone: land inundated in 20- to four main themes. Several sample interventions
100-year flood events are highlighted here.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
139
140
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
4. Improving the hydrodynamic modelling overall, and consideration of improvements in
framework that was produced in the light of flood risk assessment.
first phase of analysis, in recognition of
the limitations and uncertainties of this risk 7. Building institutional capacity through a
assessment. Key activities include channel comprehensive training package on collecting
topographic surveys to extend the model to hydrometeorological data, running the
tributaries and improve the accuracy of the hydrodynamic model, and building institutions.
model, improvement of data-sharing procedures
and protocols, and additional modelling of factors As a step toward implementing the action plan,
contributing to flood such as sedimentation. and specifically with the goal of improving data
sharing across government agencies, in November
5. Investing in structural interventions. These
2012 the Malawi government launched the Malawi
focus on flood protection for villages found to be
Spatial Data Portal (MASDAP http://www.masdap.
most at risk, catchment improvements through
reforestation, maintenance of culverts and mw/about/). This GeoNode already hosts 123 spatial
bridges to improve flow capacity, considerations layers,56 including infrastructure, OSM layers,
of flood storage options, and a feasibility analysis flood outlines from a 2012 Atkins study, elevation
of a plan to flood-proof existing buildings to act and other data, and data sets on soil type. (For
as flood shelters. more on the development and use of GeoNodes,
see the section 3-1 and box 3-1 above). It is part
6. Supporting improvements to flood
forecasting and early warning systems of the Malawi government’s effort to open data,
through review of past programs and support community mapping activities, and develop
interventions, improvements to monitoring decision support tools that leverage open data for
systems, assessment of the monitoring system contingency and land-use planning activities.
CHAPTER
03
3-15. Reducing Seismic Risk to Public Buildings in Turkey
Elif Ayhan, Joaquin Toro (World Bank)
Seismic risk in Turkey is substantial. Estimates buildings to earthquake shaking, and found they had
suggest that in the 76 earthquakes that have a high potential for collapse.
occurred in Turkey since 1900, 90,000 lives have
This risk assessment made the following high-
been lost, 7 million people have been affected, and
US$25 billion in direct damages have been incurred priority recommendations:
(Erdik 2013). The 1999 Izmit-Kocaeli and Duzce
• 635 hospitals should be urgently prioritized for
earthquakes were vivid reminders of this risk. They
detailed assessment and retrofitting.
prompted scientific analysis that emphasized the
increased risk to Istanbul arising from the nature of • Almost 2,000 schools should be urgently
the North Anatolian fault zone (Parsons et al. 2000). reviewed and retrofitted to prevent “pancake-
Indeed, this analysis suggested that Istanbul’s 1 like” collapse during an earthquake.
million buildings have a 2–4 percent chance of heavy
damage and a 20–34 percent chance of moderate • 24 bridges with a high probability of collapse and
damage from a scenario earthquake event. two viaduct bridges should be urgently reviewed
and retrofitted to prevent collapse during an
In response to the heightened concern, the Istanbul earthquake.
Metropolitan Municipality, in cooperation with the
Japan International Cooperation Agency (JICA), • To reduce the risks of secondary fires and
prepared a microzonation study with various explosions, systems that would automatically
seismic scenarios (Pacific Consultants International shut down the gas distribution network after an
et al. 2002). This analysis involved developing earthquake should be considered.
fundamental data sets on the seismology and
• A disaster management center should be
ground conditions that could amplify earthquake
established, and a campaign to raise awareness of
shaking.57 It also involved deriving exposure data—
disaster prevention should be conducted.
including data on public and private buildings,
land use, hazardous facilities, lifelines, and road The Istanbul Metropolitan Municipality took these
networks—from a variety of sources such as recommendations into consideration in developing
census and cadastral records, and then compiling the Istanbul Earthquake Master Plan.58 This plan was
them into a GIS database. Impact analyses were ultimately funded under a government of Turkey
undertaken for four scenario earthquakes, ranging and World Bank risk reduction program known as
in magnitude from 6.9 to 7.7, which were selected Istanbul Seismic Risk Mitigation and Emergency
in partnership with researchers from the Turkish Preparedness Project (ISMEP).59
scientific committee. The results suggested that 7–8
percent of buildings would have heavy damage, as Implementation of this program has improved
many as 87,000 people would be killed, and 135,000 emergency preparedness, reduced risk to existing
would be severely injured—significantly greater public facilities, and resulted in some improvement
damage than was found by the previous analysis. to building code enforcement across Istanbul—
The newer analysis also highlighted the vulnerability achievements that have collectively increased
of Istanbul’s schools, hospitals, and other public Instanbul’s seismic resilience. Highlights of
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
141
142
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
progress achieved under ISMEP by 2012 include the carried out for the Archeological Museum, Hagia
following: 60
Irene Museum, and Mecidiye Kiosk, including
development of recommendations about
• Seismic risk evaluation was carried out for 1,515
structural reinforcement.
public buildings associated with 749 schools, 31
hospitals, 57 health centers, and 51 other public This series of risk assessment studies, development
facilities. of risk reduction plans, and implementation of
• Work was done to retrofit or restore 658 investments to reduce seismic risk in Turkey
buildings associated with 451 schools, 8 hospitals, constitute a remarkable example of how risk
10 health centers, and 31 other public facilities. information can influence and trigger actual on-
the-ground risk reduction. Turkey’s achievements
• Reconstruction was performed for 95 schools
came about because of (a) strong relationships
deemed not suitable for retrofitting (where
between those developing the risk information and
estimates gave a total retrofit cost ratio higher
the decision makers using the information; (b) clear
than 40 percent of the value of the building).
actionable recommendations from risk assessment;
• Inventories were made of 176 historical buildings (c) strong political will to invest in risk reduction
in 26 complexes, and seismic evaluations were (driven by the devastation associated with the 1999
Figure 03—20 Prioritization of Public Buildings in Promoting Seismic Safety
Prioritization of Settlements with High Risks
methodology for
high seismic risk Inventory of Public Buildings in Settlements and Spatial Analyses
public buildings.
Source: World Bank 2012b. Attributes Contributing in Emergencies Attributes Contributing to Safety
of Building
• Use and Functions
• Capacity • Geological Properties of Site
• Share of Services Provided • Infrastructure Dependence
• Accessbility • Hazardous Neighbors
• Buildings to be Protected
Spatial Systems
Determination of Building of Priority According
to Their Contribution to City Safety
Re-Prioritization of Public Buildings According
to Engineering and Economic Efficiency Criteria
CHAPTER
03
ATTRIBUTE WEIGHT CLASSIFICATION DETAILS
5: vital buildings such as hospitals Table 03—7
Current and emergency use 20% 4: schools, major public buildings, etc. Building
3, 2, and 1: less important buildings Classifications Used
Service role (who and what relies on 5: a single facility that serves the entire region or city in Prioritization
20%
this building) 1: facility for which there is reasonable redundancy Methodology
5: a building that, if damaged, will cause physical damage to
Urban context 20% surrounding buildings, fires, infrastructure problems, or other Source: World Bank 2012b.
problems in its vicinity
Note: For brevity, only
5: an accessible building reachable by many roads or methods
Accessibility 15% levels 5 and 1 are described,
1: a building likely to be inaccessible in a disaster
although each attribute can
5: a building on poor soils
Geologic properties of sitea 10% earn a score of 1 to 5. For
1: a building on better soils
certain attributes, there are
5: a building totally dependent on local infrastructure multiple proposed methods
Infrastructure dependence 10% 1: a building that can operate independently for at least two for assigning values, such
weeks without external services as based on the number of
5: an historically important building students in a school.
Historical and cultural value 5%
1: not historically important
earthquakes); and (d) the prioritization of financial Turkey’s Disaster and Emergency Management
resources to invest in risk reduction. Presidency, with support from the World Bank and
GFDRR, has developed a preliminary methodology
These achievements notwithstanding, seismic risk for prioritization (World Bank 2012b). This
in Istanbul continues to increase—mainly because approach involves the development of an inventory
of population growth, urbanization, overcrowding, of public buildings, an evaluation of the relevant
and challenges associated with enforcement of importance of different buildings, and an assessment
land-use plans and construction policies. Moreover, of the elements of the building construction that
other cities in Turkey have made less progress than make them more or less likely to be damaged in an
Istanbul in reducing seismic risk. earthquake. This broad assessment methodology is
described in figure 3-20.
In light of the remaining seismic risk across the
country, the government of Turkey is seeking to This methodology is used to distinguish building
build on the success of the ISMEP project and significance levels, which included low, moderate,
extend it nationwide, focusing on public buildings significant, and high importance. Some of the
(schools, hospitals, administrative buildings, attributes used to classify buildings’ importance are
described in table 3-7.
emergency response centers, and other public
buildings with important life-safety or emergency The estimation of the earthquake performance of
response functions). Given the immense scale buildings by experienced earthquake engineers was
of this task, however, robust and objective based on building geometry and number of stories;
prioritization of buildings for retrofitting or construction quality and material properties; and
reconstruction is required. geotechnical and geological maps. This information
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
143
144
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Table 03—8 BUILDING SIGNIFICANCE
STRUCTURAL VULNERABILITY CLASSES
Prioritization for LEVELS
Reconstruction and Low Medium High
Rebuilding Class I P5 P4 P1
Class II P5 P4 P1
Source: World Bank 2012b.
Class III P5 P3 P1
Class IV P3 P2 P1
is used to determine the structural vulnerability A pilot application of this method was completed
class of low, medium, or high collapse potential. 61 in Tokat Province of Turkey in 2013. The selection
of Tokat was based on its proximity to the highly
Based on a synthesis of both these criteria, buildings active North Anatolia fault and building stock largely
for reconstruction/rebuilding were prioritized characteristic of the country. Among a sample of 12
buildings, two buildings were found to be priority
using the priorities defined in table 3-8. Under this
1 and therefore require urgent retrofitting and/or
methodology, all buildings that have a high collapse reconstruction; one building was a priority 2, seven
potential, irrespective of the building’s significance were priority 3 buildings, and two were priority 4
level as defined by its class, were allocated a priority buildings. This methodology is now forming the
basis for ongoing dialogue between the government
1 (P1). Buildings with low structural vulnerability
of Turkey—specifically the National Disaster and
were assigned the lowest priority, P5, except for
Emergency Management Presidency—and the
class IV buildings, which were assigned a priority World Bank on the design of future disaster risk
of P3. reduction investments.
CHAPTER
03
3-16. Applying Multi-Hazard Risk Assessment to the
Development of a Seismic Retrofit Program for Public
Schools in Metro Manila, Philippines
H. Kit Miyamoto, Amir S. J. Gilani (Miyamoto International); Jolanta Kryspin-
Watson, Artessa Saldivar-Sali, Abigail C. Baca (World Bank)
The Philippines is among the top global disaster and built-up area have increased in Metro Manila
hot spots, ranking eighth among countries most during the past 10 years, the projected loss of life
exposed to multiple hazards and 13th among those for a given scenario earthquake will likely increase
at high economic risk to natural disasters (Dilley significantly as well.
et al. 2005b). Two events in 2013—the magnitude
The World Bank’s partnership with the Philippines
7.2 Bohol earthquake on October 15 and Super
Typhoon Yolanda on November 8—suggest the focuses on improving the resilience of public
country’s particular vulnerability to earthquakes facilities to natural disasters by working with
and typhoons. The Philippines is also vulnerable to counterparts in the Department of Public Works
nontropical cyclone precipitation, floods, volcanic and Highways, the Department of Education, the
activity, and tsunamis. These natural hazard events Department of Health, and other line agencies
are harmful not only at the human level, but at the responsible for the construction and maintenance
economic level; it is estimated that 85 percent of of critical infrastructure. While past projects
economic activity associated with national GDP successfully raised awareness of hazards and risk,
occurs in at-risk areas (Dilley et al. 2005b). The need the next step is to help the implementing agencies
for a robust natural hazards risk reduction program prioritize risk reduction investments given limited
is great. budget resources. Using existing hazard and risk
assessment data, the current study—based on one
The National Capital Region, Metro Manila, is component of the Safe and Resilient Infrastructure
home to approximately 13 percent of the country’s program—has sought to develop a prioritization
population and generates 30 percent of its GDP. The methodology for seismic upgrading and retrofitting.
PHIVOLCS, JICA, and MMDA (2004) Metro Manila Preliminary results from a pilot analysis in Metro
Earthquake Impact Reduction study (the so-called Manila show that systematically strengthening
MMEIRS study) estimated that 10 percent of the and upgrading the most vulnerable public school
public schools in Metro Manila would incur heavy buildings would greatly reduce the number of
damage or collapse from a magnitude 7.2 West Valley projected fatalities from a magnitude 7.2 scenario
Fault earthquake, endangering 210,000 students. earthquake on the West Valley fault.
This study also found that over 50 percent of the
area’s public school buildings are at high risk from Prioritization methodology for seismic
///
earthquakes. These estimates are expected to be upgrade of schools. The prioritization method
///
updated by the findings of the Greater Metro Manila for determining which public school buildings were
Area Risk Assessment Program (GMMA RAP) being most in need of a seismic upgrade was based on the
led by the Philippine government (see section 3-4 expected number of fatalities under the magnitude
for more information). Given that both population 7.2 scenario earthquake described in the MMEIRS
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
145
146
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
study. The method took the status quo (no retrofit) • Damage functions. Using fatality rates based
as the baseline, and quantified both the benefits on Hazus (modified for Metro Manila) and
derived from and the costs associated with a seismic number of occupants, the number of fatalities for
retrofit program. each building was estimated.
The number of fatalities associated with • Cost estimates. Using the structural damage
each school building was estimated using a data, and based on the buildings’ floor area and
probabilistic risk analysis platform. Both the cost estimates obtained from local building
direct cost components (structural upgrade and/ contractors in Metro Manila, the cost (in 2013
or replacement) and indirect cost components U.S. dollars) for replacement as well as upgrade
(fatalities) were considered. In other words, the was obtained.
analysis considered both probable maximum loss
• Aggregation. Fatalities and upgrade costs were
and probable maximum death. The procedure used
integrated to identify the optimal number of
hazard, exposure, and building vulnerability as input
buildings to be selected for the first phase of
parameters as follows:
seismic upgrade.
• Seismic hazard. The seismic hazard data
The prioritization procedure can be summarized
(earthquake intensity, proximity to earthquake
as follows:
fault, and soil condition at the site) were input
as a layered map for analysis. Data were based on • The probability of the building experiencing any
the provisions of the National Structural Code of of the damage states was computed using the
the Philippines (ASEP 2010). fragility functions corresponding to the building
construction type, lateral load framing system,
• Exposure. The Department of Education
number of stories, and construction era.
provided the project team with a database that
lists the number of occupants (teachers, students, • The fatality ratio for each building was computed
etc.) for the facilities under consideration. using the fatality rate for each damage state and
Field surveys were conducted and data from the probability of exceeding that damage state.
these surveys were used to augment and modify
• The number of fatalities for each building was
the database.
computed using the fatality ratio and building
• Building vulnerability. The risk analysis occupancy.
platform provided fragility information for
• The seismic upgrade cost for each building was
buildings of various types (for example,
computed using the cost estimate per square
reinforced concrete moment frame) and vintages
meter and the building floor area.
(for example, constructed prior to adoption
of modern seismic codes). In this study, the • The fatalities and costs were aggregated for the
recommended values of FEMA’s (2003) Hazus 186 most vulnerable buildings.
model were used and modified for Metro Manila.
The costs to replace and to upgrade schools were
Following evaluation, probabilistic estimates gathered from a survey of several Metro Manila
of structural damage to a given building were contractors. A new or replacement school would
determined. These data were then used to obtain cost approximately 25,000 pesos (US$580) per
the following: square meter. Upgrading, which includes earthquake
120
Figure 03—21
Estimated Metro
Manila student
100 fatalities per school
building for a
magnitude 7.2 West
Valley fault scenario
80
earthquake occurring
in the daytime.
Fatalitiy
Source: Miyamoto
60
International and
World Bank.
18% of fatalities
40
25% of fatalities
20
80% of fatalities
0
500 1,000 1,500 2,000 2,500 3,000 3,500 4,000
Building number
strengthening and functional upgrades (for example, school was in session), it would result in an
bathrooms), would cost approximately 5,200 to estimated 24,400 student deaths given the current
11,000 pesos (US$120 to US$260) per square student population in Metro Manila. Over 25 percent
meter, depending on the number of stories and the (6,385) of these fatalities would occur in only 5
site requirements. percent (186) of the buildings, and 18 percent (4,320)
would occur in the most vulnerable 100 buildings
Key findings of school retrofit prioritization
(3 percent) (see figure 3-21). By strengthening
///
study. Key findings of the prioritization study
40 percent (1,500) of the most vulnerable school
///
indicate that, because of the use of older seismic
buildings, potential student fatalities could be
design codes or poor-quality detailing and/or
reduced by 80 percent (over 19,000 student lives
construction, multistory reinforced concrete
saved).
construction of the variety typically found in Metro
Manila public schools is especially vulnerable to The corresponding cost analysis showed that the
earthquake damage or collapse. The study further cost of strengthening and upgrading a typical school
determined that if a 7.2 magnitude scenario building is between 20 percent and 40 percent
earthquake event occurred in the daytime (while of the cost of new construction. Using the 20
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
147
148
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
percent figure, Metro Manila could strengthen and US$40 million and US$80 million (depending on the
renovate five school buildings for the cost of one extent of functional upgrades).
new building.
It should be noted that there were limitations in the
The Philippine Department of Public Works and exposure and hazard data that were available for
Highways has made the decision to implement the demonstration seismic prioritization analysis.
When more accurate data become available, the
the cornerstone phase (retrofit of 200 school
prioritization should be refined to identify the
buildings) of the Safe and Resilient Infrastructure
highest-risk candidate buildings for strengthening.
program in Metro Manila, with a view to eventual
Factors other than those related to structural
scale-up to other sectors (including lifeline
vulnerability (such as the need to replace certain
infrastructure) and geographic locations as well as schools to meet modern standards for educational
institutionalization of quality assurance systems. delivery) should also be considered when developing
Using local construction cost estimates, the cost to the final prioritized list of buildings for upgrading.
strengthen the most dangerous 5 percent (186) of These factors will be incorporated through
the vulnerable buildings is estimated to be between consultations with the Department of Education.
CHAPTER
03
3-17. Morocco Comprehensive Risk Assessment Study62
Axel Baeumler (World Bank); Charles Scawthorn (Kyoto University, emeritus),
Erwann Michel-Kerjan (Wharton Business School, University of Pennsylvania)
In recognition of the accelerating series of global The historical record of disasters in Morocco is
shocks—financial crises, commodity volatility, relatively short and incomplete. However, it is clear
and natural disasters—officials in the government that hydrometeorological hazard has affected the
of Morocco proactively developed and adopted a most people and created the most economic loss,
national strategy for integrated risk management whereas earthquakes have resulted in the most
(IRM). Working in partnership with the World fatalities (12,000 people were killed by the 1960
Bank, Morocco is using this strategy to reduce the magnitude 5.7 Agadir earthquake) and have also
potential impacts of future crises, to increase its been a major source of economic loss. 63Given that
Morocco’s urban population is expected to increase
resilience and responsiveness when crises occur, and
15 percent by 2025, seismic and flood risk will
to support decision making on resource allocation
likewise increase unless well managed. In addition,
and prioritization. This effort followed initial
the country already experiences more intense
investment in preliminary risk profiles (see box 3-6).
and frequent droughts and floods resulting from
This integrated risk approach was viewed as critical climate change, and increasingly scarce freshwater
because not all risks are equal across the public availability.64
sector; thus any risk management strategy must be
Probabilistic disaster risk assessment. As part
appropriately targeted. An IRM approach avoids
/// ///
of the IRM project, a probabilistic GIS analysis tool,
the tendency of risk management to be undertaken MnhPRA (Morocco natural hazards Probabilistic
in “silos” and is a rare example of enterprise risk Risk Assessment), was developed and used to assess
management—that is, the process of quantifying current earthquake, flood, tsunami, drought, and
risks, comparing them with one another, and landslide risk in Morocco (World Bank 2013).65
managing them in a coordinated manner—beginning This software package enables users to inventory
to be applied in the public sector. Morocco’s assets at risk, determine the hazard
characteristics and assign vulnerability functions,
The IRM initiative was launched in 2008 with
and estimate the impacts of these hazards on
financial support from the GFDRR and the Swiss
the assets in a robust and quantitative manner.
Agency for Cooperation and Development. It has
The impacts can be determined as estimates
focused on three key risk areas: (a) natural disasters, of the fatalities, injuries, and direct economic
specifically earthquake, tsunami, flood, and drought consequences of all possible hazard occurrences—
events; (b) commodity (energy) price volatility; and ranging from rare and potentially catastrophic
(c) agricultural risks, comprised of drought, pests events to frequent, lower-impact events. Loss
and diseases, and market price volatility. Of these, estimates are provided in detailed tables at the
natural disaster risk has been the most extensively commune level; in summary tables at the province,
assessed, and the results of these assessments are region, and national levels; and as maps. Risk can
discussed here in greatest detail. be assessed under current conditions and for future
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
149
150
Box 03—6 Risk Assessments as an Advocacy Tool for DRM in points in time considering growth and urbanization
the Middle East and North Africa as well as alternative public policies.
In 2008, GFDRR provided seed funding to help scale up DRM engagements MnhPRA used input-output and computable
in the Middle East and North Africa. Djibouti, Morocco, and Yemen received general equilibrium modelling to measure the
US$70,000, US$100,000, and US$150,000, respectively, to fund rapid risk
indirect economic costs of disasters (how the
profiling and assessment. These projects enabled each country to better
economy adjusts to the shock, including the effects
understand and more effectively communicate risk, and they sparked new
cooperation in risk management across ministries. With additional funding for
on household income and consumption). These
risk mitigation in the housing, infrastructure, energy, and education sectors, models, which were developed in conjunction with
government leaders partnered with the UN and European Union to carry out the government’s High Commission for Planning,
post-disaster needs assessments in Djibouti (for the 2011 drought, with funding capture the interdependencies between all sectors
of $60 million) and Yemen (for flooding in 2008, with funding of $30 million). of the economy as well as the ex ante and ex post
In all three countries, risk assessments were used as an advocacy tool. That
macroeconomic decisions of the government.
is, the assessment results showing the potential average annual losses
The project built a comprehensive exposure model
arising from a disaster were used to sensitize finance ministers to the need
for DRM. With finance ministers aware of the cost of inaction, technical
for Morocco covering residential, commercial,
assistance was expanded to multi-sectoral programmatic risk management: industrial, and public infrastructure and agricultural
early warning systems, risk management laboratories, and knowledge assets. The exposure model was compiled through
centers were established, and risk reduction information was integrated into a combination of existing data sets (collected from
development plans and strategies. Following the success of this approach, risk government institutions), satellite imagery, site
assessments were initiated by government authorities in Algeria and Saudi
visits, and statistical modelling. The project found
Arabia with the aim of sensitizing relevant ministries to the importance of DRM,
that the total value of the built environment in
influencing vulnerability reduction strategies and financial disaster risk transfer
instruments, and leveraging best practices. Partly as a result of getting finance
Morocco—that is, the replacement value of houses,
ministries to recognize DRM’s importance, most countries in the Middle businesses, factories, roads, bridges, ports, vehicles,
East and North Africa have made progress in DRM in recent years. Especially electrical networks, and other assets—is DH 2.7
notable is the shift in these countries away from reactive response to disaster trillion (US$ 330 billion), or around DH 90,000
to more proactive DRM—a shift that signals increased commitment to HFA (US$11,000) per capita.
objectives and priorities.
Earthquake risk was found to be concentrated in
Source: Andrea Zanon (World Bank).
the north of the country and in the seismically
active area between Fez, Marrakech, and Agadir—
essentially the mountainous belts formed by the
collision of the African and Eurasian plates. Five
provinces (Nador, Al-Hoceima, Berkane, Taza,
Tetouan) were found to account for 34 percent of
the estimated average annual loss from earthquake
despite having only 8 percent of the national
building exposure. These findings highlight the
government’s opportunity to significantly reduce
seismic risk in these provinces through focused
investments that increase earthquake resiliency.
Floods are a chronic disaster management challenge
for Morocco. Analyses showed that a significant
fraction of Morocco’s total exposure is at risk from
CHAPTER
03
flood, but that four provinces contribute 60 percent mitigation options. The BCR for these scenarios
of the total flood loss with respect to average annual ranged from 54.0 to 1.1 (the higher the BCR, the
loss. These findings provide a clear target for future more benefits for the money spent), with some
flood mitigation investments; they also indicate specific ratios as follows:
which areas should give greater consideration to
• Flood warning systems for the Ouregha subbasin:
flood risk in future urban and land-use planning.
BCR = 54.0
The analyses also highlighted the effects of flood
on the economy—evident, for example, in the • Culverts on railway lines in the Gharb plains:
vulnerability of the main railway line in the Gharb BCR = 34.6
plains, which when damaged significantly reduces
the flow of goods across Morocco. • Strengthening of hospital buildings in Nador
Province: BCR = 5.8
Tsunami events were found to represent a rare but
potentially devastating risk to Morocco’s Atlantic • Risk assessment for proposed new schools in the
and Mediterranean coastlines, with waves as high country: BCR = 5.7
as 10m possible in Casablanca, Morocco’s largest
• Seismic strengthening of schools in Nador
port. Not much attention is paid to tsunami risk,
Province: BCR = 3.6
particularly in the Atlantic basin. But tsunami
caused significant loss of life in Morocco after the These BCR analyses provide a quantitative measure
1755 earthquake (better known for its catastrophic that promotes efficient resource allocation.
effects in Lisbon).
A risk assessment also provides a higher-level
Drought is an insidious and significant risk to the perspective on the cost of various portfolio
agricultural sector in Morocco, which currently investment choices. For example, the cost to
employs about 40 percent of the nation’s work force. strengthen the seismic resilience of all schools
Especially at risk are the lowlands where cereal and hospitals in high-risk provinces was estimated
crops are grown, which are subject to considerable at DH 1.7 billion (US$207 million) and DH 700
variation in annual precipitation. Indeed, on average, million (US$85 million), respectively. For flood,
drought occurs every third year in Morocco, creating early warning systems in three regions would
volatility in agricultural production that is the main involve a capital outlay of about DH 400 million
constraint to expansion in the sector. (US$49 million), with annual operating costs of DH
40 million (US$4.9 million). Overall, total losses
Cost-benefit analysis provided a key tool in associated with a disaster event were typically
communicating the costs and benefits of different found to be 25 to 30 percent higher than the direct
risk reduction and mitigation actions. While benefits losses calculated through physical loss modelling
can be derived by increasing mitigation efforts, these (Government of Morocco 2012).
efforts come with an increasing cost. Hence it is
critical to determine, through cost-benefit analysis, Conclusions of IRM study. The probabilistic
/// ///
the optimal level of mitigation—that is, the point risk assessment revealed that natural disasters
where decreasing loss equals the increasing cost will cost Morocco DH 5.0 billion (US$611 million)
of mitigation. annually on average, of which flood contributes
the greatest loss. However, the average annual
For Morocco, the comprehensive probabilistic loss does not fully characterize Morocco’s risk. An
risk assessment allowed benefit-cost ratio (BCR) extreme event, such as an earthquake striking a
analyses to rank the effectiveness of 51 potential major population center, could have direct costs
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
151
152
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
on the order of DH 100 billion (US$12 billion), Quantifying these risks will help Morocco to move
equivalent to 5 percent of GDP, or 23 percent of the toward the next phase of managing the risks,
national budget. This amount is substantially higher mainly through dedicated investment programs
if indirect socioeconomic costs are considered, targeting both physical and fiscal risks. Using risk
such as the ripple effects on other sectors of the analysis, the government of Morocco has begun
to prioritize key short- and medium-term actions
economy. While the government would not bear the
across all three risk categories (natural disaster,
full cost of the damage to residential assets, there is
commodity price volatility, and agricultural
an implicit liability attached to this sector, and it is
risks). For natural disasters, short-term priorities
likely that government aid for asset reconstruction
include establishing early warning systems for
and livelihood support would be significant. flood, tsunami, and earthquake events; carrying
out additional hazard and risk analyses; enhancing
The loss from disasters, however, is not the sole
building code compliance; mounting an education
risk for Morocco. In 2011, oil volatility in Morocco
campaign around the need for seismic retrofits
resulted in a DH 30 billion (US$3.6 billion) negative
in the most seismically at risk areas of Morocco;
impact on the national budget, a result of the
and establishing a national catastrophic insurance
country’s existing fuel subsidy system. In 2008, the program for private assets. Lastly, MnhPRA has been
country’s agricultural risks cost an estimated DH installed in government ministries, with the aim that
75 billion (US$9 billion), and projections suggest it will become an ongoing tool for monitoring and
that these costs could rise as high as DH 185 billion managing exposure and risk at both the national and
(US$22.6 billion) by 2020. local level.
CHAPTER
03
3-18. Risk Assessment for Financial Resilience: The
Approach of the World Bank
World Bank/GFDRR Disaster Risk Financing and Insurance Program
Risk assessment is the first step in managing decision makers. With these data as a foundation,
disaster risk. Understanding and quantifying the governments can develop effective strategies that
risk allows policy makers to estimate the potential build financial resilience across society, increase the
direct physical and human losses from adverse financial response capacity of the state, and protect
natural events. This information can in turn help long-term fiscal balances.
governments, communities, and individuals make
informed decisions to strategically manage their The level of application and detail of the catastrophe
risks. Like other efforts to manage risk, financial risk model will depend on the decision to be made
protection strategies through disaster risk financing and the availability of data. Risk models for use in
and insurance (DRFI) rely on risk information. financial risk-transfer applications require high-
Financial risk assessment and financial diagnostics resolution and high-quality data sets that can
build on this information to help decision makers
withstand scrutiny by international finance and
understand financial and fiscal exposure to disaster
insurance institutions. They also require robust
risk.
reporting as well as methodologies that effectively
Experience has demonstrated that different DRFI convey the nature and uncertainty surrounding risk.
questions require different types and resolutions
of disaster risk information. For example, a What DRFI decision making requires from
///
national disaster risk profile undertaken at a coarse catastrophe risk models. The financial
///
resolution could be the starting point for a policy analysis enabled through simulated catastrophe
dialogue on DRM within a country, and could be risk data empowers policy makers to take more
used to raise public awareness of disaster risks. informed financial decisions in the public
It could also provide momentum for the more financial management of natural disasters.
resource-intensive and detailed risk assessments While sophisticated financial decision making
needed to guide specific financial decisions about requires highly detailed and granular outputs, risk
risk reduction investments. modelling provides many useful applications even
An analysis of historical loss information can in the absence of such detailed data. For example,
inform initial thinking on DRFI. The next step in comparatively coarse and incomplete data can still
developing a robust financial and fiscal protection be sufficient for showing governments the relative
strategy should be a quantitative risk assessment importance of different risk layers.
with detailed probabilistic modelling. Historical
loss data and simulated loss data from catastrophe But to provide the necessary level of granularity of
risk models can be used as the basis of financial outputs for the most complex financial decision
decision making (see figure 3-22). Financial risk making, catastrophe risk models require high-
analytics helps translate technical risk information quality, high-resolution inputs of their own.
into financial analysis that is useful to nontechnical Specifically, they require the following:
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
153
154
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
• A database of assets at risk (exposure yet physically plausible events (even if these have
module). A high-resolution exposure database a very low likelihood of occurrence).
comprised of the assets at risk to natural
• A database of asset fragility curves (vulnerability
hazards is essential in informing DRFI decision
module) that make the translation from hazard
making. At a minimum, individual risks should
and exposure to damage and loss. A high-
be identified in terms of their georeferenced
resolution vulnerability database is crucial for
location, value (economic replacement cost),
linking the physical characteristics of the assets
usage (school, office, hospital, etc.), and
at risk with the local intensity of the hazards to
construction type.
determine damage and loss estimates. Fragility
• A probabilistic hazard module comprising curves are described as mean damage ratios
and will vary by building use, construction,
synthetic representations of all possible
height, and age. The vulnerability component
hazard types. The hazard module of a
of a catastrophe risk model must reflect the
catastrophe risk model comprises a stochastic
impact of these key asset components, as well as
event catalog, which contains simulated
geographical changes across a country, such as
hypothetical events of different magnitudes.
those due to variations in regional construction
Events are modelled with a geographic footprint
codes and practices.
of hazard values represented at high resolution,
and take into account local site conditions such Commercial (vendor-built) catastrophe risk models
Figure 03—22 as soil type, surface roughness, or elevation. It is that are used in the private insurance industry also
Flowchart for financial important that the event catalog is well calibrated generate estimates of the possible broader sectoral
decision making. to historical records, but also allows for extreme impacts of disasters. Some models can apply
Collect historical loss data
and simulated loss data
Specify objectives
from catastrophe
risk models
Interpret data
and make assumptions Specify potential strategies
Monitor experience
about the economic for analysis
environment
Conduct comparative analysis of potential strategies
Make a decision and implement the preferred strategy
CHAPTER
03
adjustments to loss calculations—either based on catastrophe risk models play an important role: they
projections of inflation in labor costs and building allow analysts to identify the potential economic
materials during the post-disaster reconstruction impacts of natural disasters over different time
phase arising from increased demand, or based on frames so that analysis can test potential approaches
increases in the cost of food affecting government’s to risk retention and risk transfer before a severe
contingent liability to food security response. event occurs.
Particularly sophisticated catastrophe risk analyses
also attempt to include potential inflation mitigation Technical information generated by detailed risk
effects, such as the flow of labor and materials from models enables decision makers to carry out a range
unaffected regions (increased supply) and the use of of important tasks:
public work forces.
• Model and evaluate the cost-benefit ratio of
The outputs generated by such catastrophe risk complex financial instruments, such as (re)
models feed into the DRFI decision-making process. insurance contracts and catastrophe risk (CAT)
Typically these probabilistic models produce 10,000 bonds when applied as the basis of financial
or more years of simulated event losses and are the analytics tools
basis for metrics such as average annual losses—an
• Understand potential losses due to
estimate of the average annual losses that a portfolio
extreme events
of risks would be expected to incur from the hazards
modelled—and probable maximum losses—the • Quantify AALs and PMLs
maximum probable losses that could be expected
given the model inputs. PMLs are often described in • Model different sovereign DRFI strategies, which
terms of either a return period of occurrence (e.g., blend risk-retention, risk-transfer, and budgetary
a loss expected to occur, on average, once every 100 mechanisms, to compare the protection offered
years) or an annual probability of occurrence (e.g., and associated cost
a loss expected to occur, on average, with an annual
• Understand how key economic assumptions in
probability of 1 percent).
the models (such as inflation and interest rates)
Deterministic (also known as “scenario” or “what affect the losses
if?”) catastrophe model outputs are also useful to
governments because they allow analysis to focus on AAL and PML metrics are particularly useful for
the financial impact of single, defined events. This feeding into financial analytical tools, to both inform
approach is particularly beneficial if the country in and test prototype DRFI strategies. Financial risk
question has a history of severe natural disasters analysis allows decision makers to take the raw
(one or more of which may still be fresh in residents’ risk information and model complete financial
memory) or has neighbors that have recently protection strategies, and in this way to understand
experienced a catastrophic event. government’s average cost as well as probable
maximum retained cost.
How this information is used. Countries starting
/// ///
a DRFI engagement require a robust process to AAL and PML metrics enable complementary
understand the financial risks they face and to assess aspects of financial risk analytics to inform
and evaluate potential DRFI strategies. This process decisions. The AAL metric, calculated from all
includes the statistical analysis of historical losses, possible hazards affecting a country, places the
case studies, and simulated risk data. Probabilistic focus on the likely annual financial cost of natural
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
155
156
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
disasters. Once this number (or range) is identified, which depends on the clear identification of asset
it can be used to inform decisions, such as what classes in the underlying exposure databases,
the size of a national disaster reserve fund, and the ensures that only risks that the government
potential annual budgetary allocations to it, should considers to represent contingent liabilities are used
be. Graphical representations of the contributions of in the financial analysis and evaluation of potential
factors such as hazard type, geography, and affected DRFI strategies. For example, a recent preliminary
asset classes to the AAL across a territory can help exposure database developed in Colombia for the
decision makers understand which factors cause cities of Bogota, Medellin, and Cali identified the
most of the expected loss.
following asset divisions: residential (low, medium,
PML metrics at different return periods help and high socioeconomic classes), commercial,
to identify potential financial requirements for industrial, health (public and private), education
catastrophic events with a low annual probability (public and private), and institutional (public and
of occurrence. Five-to-ten year PMLs can inform private). Information like this allows governments
decisions about the size of potential short- to mid- to identify the contingent liabilities that should
term financing instruments, such as contingent inform DRFI decision making.
lines of credit. Similarly, low annual probability
The risk information generated by financial risk
PMLs (e.g., 100-year or 250-year return periods)
assessment and modelling is not only valuable
can inform the size of financial protection
instrumentation for the purpose of transferring for developing comprehensive sovereign DRFI
sovereign risk to the international capital and (re) strategies. Given their high level of detail, the data
insurance markets. sets can in some cases be adapted, often quickly
and at low cost, to inform local-level planning. The
An important component in DRFI is clarifying Pacific Catastrophe Risk and Financing Initiative,
contingent liabilities of the state. Disaster risks for example, has adapted data sets in this way. (For
create implicit and explicit contingent liabilities
more information, see section 3-19).
to the government budget, though these are
generally not well defined in law, making fiscal risk Limitations and challenges in risk modelling
///
assessment complex. Beyond explicit contingent for DRFI. The use of risk assessments’ quantitative
///
liabilities and associated spending needs, such as the outputs for DRFI purposes is constrained by a
reconstruction of public assets and infrastructure, number of challenges. First, low- and middle-income
governments may in cases of disaster have a moral countries tend to lack the technical understanding
and social responsibility (implicit contingent needed to perceive the importance of ex ante DRFI
liability) to offer their populations emergency initiatives and the potential gains arising from
assistance (such as food, shelter, and medication) ex ante DRFI programs. Countries often lack the
and to finance recovery/reconstruction activities
capacity, resources, and experience to properly
(e.g., through stimulus grants for rebuilding low-
use existing products. Globally, countries and
income housing stock).
international donors invest significant resources in
Suitable granularity of catastrophe risk modelling data collection and risk modelling. But the resulting
output is crucial for determining the elements technical risk information (simulated losses, average
driving the state’s liability—that is, the key asset annual losses, probable maximum losses, etc.) is
classes, the location of vulnerable populations, and difficult to understand for policy makers and often
responsibility for food security. This granularity, unsuitable for use in financial analysis.
Second, appropriate risk modelling tools are still Box 03—7 R-FONDEN: The Financial Catastrophe Risk Model
lacking in countries that need them the most. The of the Ministry of Finance and Public Credit in Mexico
sophisticated risk modelling tools required for
Mexico has developed a comprehensive financial protection strategy
DRFI analysis are generally unavailable for low-
relying on risk retention and transfer mechanisms, including reserve funds,
income countries and even for middle-income indemnity-based reinsurance, parametric insurance, and catastrophe bonds.
countries. The science required for modelling some An in-depth understanding of the risks has allowed the Mexican government
important contingent liabilities, such as those from to successfully access international reinsurance and capital markets to transfer
specific risks.
food insecurity, is still immature; even for better-
understood risks, such as earthquakes, existing risk A fundamental feature of the strategy is the R-FONDEN, a probabilistic
modelling tools are often inadequate for the needs catastrophe risk assessment platform developed to estimate the government’s
of DRFI and require substantial improvements and financial exposure. R-FONDEN offers scenario-based as well as probabilistic
analysis at national, state, and sub-state levels of four major perils
additions if they are to be used for DRFI purposes.
(earthquake, floods, tropical cyclones, and storm surge) for infrastructure in
Exposure data, for example, may rely heavily
key sectors (education, health, roads, and low-income housing).
on official census data and disregard unofficial
settlements (such as shanty towns or squatter R-FONDEN takes as input a detailed exposure database (with information
on buildings, roads, and other public assets) and produces as outputs risk
towns) that regularly suffer the most damage in a
metrics such as annual expected loss (AEL) and probable maximum loss (PML).
disaster.
This model is currently used by the Ministry of Finance, in combination with
actuarial analysis of historical loss data, to monitor the disaster risk exposure
Catastrophe risk models used in low- and medium-
on FONDEN’s portfolio and to design risk transfer strategies.
income countries are usually not tailored to provide
the type of information that is essential for DRFI Source: Text is from “Disaster Risk Assessment and Risk Financing: A
G20/OECD Methodological Framework,” http://www.oecd.org/gov/risk/
(total ground-up losses suffered by the entire built
G20disasterriskmanagement.pdf.
inventory, number of collapsed buildings, fatalities,
homeless population, impact on crops, impact on
food security, etc.). Retuning existing commercial
models can be an expensive endeavor. It is also
important to keep in mind that the exposure data
underlying risk modelling tools become obsolete
quickly; some are even born obsolete or inaccurate.
Using old census data to collect information on
exposure in fast-growing developing countries is
a risky and potentially inaccurate business, even
if data are trended. Ownership from countries is
needed to maintain these tools, update databases,
and essentially keep them alive. This ownership is
hard to establish, and significant efforts in capacity
building are often needed even where it exists.
Third, underlying disaster risk information is often
lacking in developing countries. DRFI solutions are
only as reliable as the risk assessment models that
support them, and the latter are only as good as
the data used to develop them. Data on exposure
157
158
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
may be scattered among different governmental strategies are required to address this demand,
ministries and other organizations, and may be kept help governments increase post-disaster financial
in precarious conditions (see “Exposure” in part 2 response capacity, and build domestic catastrophe
for additional discussion of these challenges). Use of insurance markets. Probabilistic risk assessment
satellite imagery is often the only way to gather up- and catastrophe risk modelling tools can empower
to-date exposure data, but the cost of acquiring such policy makers to take better-informed decisions,
images can be prohibitive for developing countries, while technical support helps countries collect the
unless organizations provide information already underlying data and build the required models. More
in their possession free of charge for development work is needed to establish the link from technical
purposes. (The U.S. State Department’s Imagery outputs to financial analysis so that nontechnical
to the Crowd initiative does just that; for more decision makers can use catastrophe risk data.
information see section 3-3.) Through simplifying complex technical data and
providing key financial figures, DRFI analytics helps
Despite best efforts, challenges and imperfections strengthen the connection of policy makers and
will remain in every exposure database and need technical experts and ensures that policy makers
to be taken into account when modelling loss have the information they need to take the best
estimates. Inflated, detrended historical loss figures decisions about financing disaster risk.
provide useful statistical information about the risk
faced and can be used to adjust outputs from the Two initiatives that exemplify how probabilistic
risk model. The collection of actual loss data should risk assessment and catastrophe risk modelling
can facilitate DRFI decision making are Mexico’s
complement efforts in collecting exposure data.
National Fund for Natural Disasters (Fondo
The way forward. Developing countries are
/// ///
Nacional de Desastres Naturales, FONDEN),
increasingly requesting advisory services to created in 1996 (box 3-7), and the comparatively new
proactively manage the fiscal costs of natural Southeast Europe and Caucasus Catastrophe Risk
disasters. New financial instruments and Insurance Facility, or SEEC CRIF (box 3-8).
CHAPTER
03
Box 03—8 Southeast Europe and Caucasus Catastrophe Risk
Insurance Facility
The Southeast Europe and Caucasus Catastrophe Risk Insurance Facility
(SEEC CRIF) project was created to respond to a growing demand from
Southeast European countries for assistance in reducing their fiscal
vulnerability to natural disasters and for greater access to high-quality and
affordable catastrophe insurance products for homeowners and small to
medium enterprises.(A) In support of these efforts, the World Bank provided
financial and technical assistance to Albania, the former Yugoslav Republic of
Macedonia, and Serbia to establish the Europa Reinsurance Facility (Europa
Re).
The main objective of Europa Re is to increase access to affordable
catastrophe insurance products for homeowners and to facilitate the
development of the catastrophe insurance market in member countries.
Specifically, Europa Re aims to increase the level of catastrophe insurance
coverage from the current 1–2 percent to 10–25 percent over the next 5 to 10
years. The design of Europa Re follows that of similar successful catastrophe
insurance programs in Turkey and Romania. The Turkish catastrophe insurance
pool, for example, currently provides coverage for over 6 million households,
while the Romanian catastrophe insurance program insures over 5 million.
Increased access to insurance products will occur through investment in key
areas. These include educating homeowners and business owners about the
exposure of their properties and businesses to natural hazards; improving and
standardizing catastrophe insurance products’ credit quality; providing support
to enable insurance companies to sell complex weather and catastrophe risk
insurance products; and helping governments and insurance regulators enact
Seismic hazard map for Albania, FYR Macedonia, and Serbia showing regulatory and policy reforms that promote the development of catastrophe
horizontal peak ground acceleration at the surface with an exceedance and weather risk markets.
probability of 10 percent in 50 years (equivalent to a mean return period of
A critical factor underpinning the success of the SEEC CRIF is access to
475 years).
high-quality and high-resolution catastrophe risk models, which have been
developed for FYR Macedonia, Serbia, and Albania by AIR Worldwide. For
example, earthquake loss estimates are now available for these countries;
they give a 1 percent exceedance probability for losses of €1.15 billion, €611
million, and €955 million for Albania, FYR Macedonia, and Serbia, respectively;
a sample seismic hazard map produced in this analysis is shown here.
(A) The program is strongly endorsed by and has received financial support
from multiple donors, including European Union, UNISDR, Swiss State
Secretariat for Economic Affairs, and Global Environment Facility.
Source: Europa Re.
Note: MRP = mean return period; EQ = earthquake.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
159
160
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-19. The Pacific Catastrophe Risk Assessment Initiative66
Olivier Mahul, Iain Shuker, Michael Bonte (World Bank)
The Pacific Islands are extremely exposed to natural was created and made available for the 15 Pacific
hazards, including volcanic eruptions, floods, Island countries: the Cook Islands, the Federated
droughts, earthquakes, tsunamis, and tropical States of Micronesia, Fiji, Kiribati, Nauru, Niue,
cyclones. With rising populations, increasing Palau, Papua New Guinea, the Marshall Islands,
urbanization, and changes in climate, the impacts Samoa, the Solomon Islands, Tonga, Tuvalu,
from these hazards are growing. Indeed, some Vanuatu, and Timor-Leste.67 This information is
Pacific Island countries (PICs) face losses that now housed in the Pacific Risk Information System
could well exceed their annual gross domestic (PacRIS) platform (hosted and managed at the
product. The September 2009 tsunami that hit SOPAC) and includes the following:
Samoa, American Samoa, and Tonga provides a
tragic reminder of the potential impacts of disasters • Database of Historical Tropical Cyclones
in the Pacific. This tsunami left 150 people dead and Earthquakes (hazard database). The
and some 5,300 people—2.5 percent of Samoa’s database is the result of an exhaustive effort to
population—homeless. It also caused extensive collect, merge, and process data from multiple
damage to Samoa’s infrastructure. The total cost of sources regarding historical Pacific earthquakes
the tsunami—restoring infrastructure, maintaining and tropical cyclones, along with the monetary
access to basic social services, providing social losses and impact on populations associated
safety nets to the affected population, and investing with these events. The historical earthquake
in DRM—is estimated to be a staggering 21 percent catalog currently includes about 115,000 events
of GDP over the next three to four years (World of magnitude 5 or greater that occurred in the
Bank 2010a). region between 1768 and 2009, while the tropical
cyclone catalog includes 2,422 events from 1948
In 2007, the World Bank established the Pacific
to 2008.
Catastrophe Risk Assessment and Financing
Initiative (PCRAFI) to develop disaster risk • Database of Accumulated Losses
assessment tools and practical technical and (consequence database). Most of the events
financial applications to reduce and mitigate the included in the hazard database did not have
vulnerability of Pacific Island countries to natural
major consequences for the human population,
disasters. This was a joint initiative of the World
infrastructure, residential buildings, or crops, but
Bank, the Secretariat of the Pacific Community
some did. A consequence database was assembled
Applied Geoscience Technology Division (SOPAC),
containing approximately 450 events from 1831
and the Asian Development Bank, with financial
to 2009 that affected at least one of the 15 PICs.
support from the government of Japan and the
This database, which is the most complete in
Global Facility for Disaster Reduction and Recovery,
existence for the Pacific region, shows that,
and technical input from Geoscience Australia, GNS
on average, these countries have collectively
Science, and AIR Worldwide.
experienced losses in the order of US$1 billion
Under the PCRAFI initiative, the largest regional per decade, rising to US$4 billion in both the
collection of geospatial information on disaster risks 1980s and the 1990s.
CHAPTER
03
• Database of Assets Exposed to Disasters with different return periods. They also were the
(exposure database). This database contains basis for maps of the geographic distribution of
components for buildings and infrastructure, hazards, assets at risk, and potential losses, which
agriculture, and population. The exposure can be used to prioritize DRM interventions. This
database was created by collecting existing analysis determined that the average annual loss
data sets, remote sensing analysis, and field caused by natural hazards across the 15 countries
surveys. Country-specific data sets were used to is about US$284 million, or 1.7 percent of regional
characterize buildings (residential, commercial, GDP. Vanuatu, Niue, and Tonga were found to
and industrial), major infrastructure (such experience the largest average annual losses,
as roads, bridges, airports, ports, and utility equivalent respectively to 6.6 percent, 5.5 percent,
assets), major crops, and population. For the and 4.4 percent of their national GDPs. The analysis
building and infrastructure data set, more than also found that in a given year, there is a 2 percent
500,000 footprints of structures were digitized
chance that the Pacific region will experience
from high-resolution satellite images. These
disaster losses in excess of US$1.3 billion from
buildings represent about 15 percent (36 percent
tropical cyclones and earthquakes.
without Papua New Guinea) of the estimated
total number of buildings in the PICs. Of these, Key outcomes of this work include the following:
about 80,000 buildings were physically checked,
photographed, and classified. An additional 3 1. A substantial investment in improving the
million primarily rural buildings were geo-located underpinning data sets that enable robust risk
and classified using remote-sensing techniques. modelling in the Pacific.
In addition to information on infrastructure and
2. Substantial efforts to ensure all data and
residential buildings, the database also includes
analytical results produced under this initiative
topological maps and information on major cash
are available to all stakeholders in the Pacific,
crops, ground cover, and population. To date, this
for DRM purposes, but also more broadly for
database is the most comprehensive of its kind
development planning.
for this part of the world.
3. Support to PICs to highlight the potential
• Database of Modeled Probabilistic Hazards
impact of disasters from a physical and financial
and Losses. The effort generated a variety
perspective, and assistance to nations to
of risk-related information, including hazard
improve their macroeconomic planning for
maps for earthquake and tropical cyclones for
different return periods, maps of average annual natural disasters.
losses, and summaries of key return-period levels
4. Establishment of a catastrophe risk pool for six
of loss for various disaggregated subnational
Pacific Island nations—the Cook Islands, the
administration units.
Marshall Islands, Samoa, the Solomon Islands,
The PCRAFI project used these data sets to develop Tonga, and Vanuatu. This pilot program tests
catastrophe risk profiles for 15 Pacific Island nations a risk transfer arrangement modelled on an
using state-of-the-art risk modelling that simulated insurance plan that uses parametric triggers,
thousands of cyclones, earthquakes, and tsunamis. such as cyclone intensity, to determine payouts,
These risk models provide a robust estimation of so disbursements are quick. This insurance
the economic losses caused by natural disasters program recently paid out US$1.27 million to
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
161
162
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Tonga following the damage from Cyclone Ian in codes that include country-specific seismic and
January 2014. 68
wind loads; these will guide building designs that
ensure adequate shelter for the population.
In the future, the data provided in PacRIS can also
support efforts aimed at the following: • Rapid disaster impact estimation. The aim
of this application is to model the expected
• Urban and development planning. Planners
losses from a catastrophic event immediately
can use the information to evaluate the impact of
after a disaster using already collected baseline
changes to land use and zoning based on natural
information on assets. Rapid assessments after a
hazard risk, to develop investment plans to
retrofit buildings for earthquakes, or determine disaster will facilitate a faster flow of funds.
the benefits of raising floor levels to avoid
• Understanding the impacts of disasters as
flooding due to tropical cyclones. The data can
the climate changes. PCRAFI and the World
also be used in cost-benefit analyses of proposed
Bank, in partnership with Geoscience Australia
disaster prevention or mitigation investments.
and the Pacific Australian Climate Change
• Improved building codes. The earthquake and Science and Adaptation Program, are undertaking
tropical cyclone hazard models provide critical analyses to understand future cyclone risk to
information for creating and revising building critical assets in the Pacific (see section 3-24).
CHAPTER
03
3-20. From Multi-Risk Assessment to Multi-Risk
Governance: Recommendations for Future Directions
Anna Scolobig (Institute for Environmental Decisions, ETH Zurich; Risk, Policy
and Vulnerability Program, International Institute for Applied Systems Analysis);
Alexander Garcia-Aristizabal (Analisi e Monitoraggio del Rischio Ambientale);
Nadejda Komendantova, Anthony Patt (Institute for Environmental Decisions,
ETH Zurich; Risk, Policy and Vulnerability Program, International Institute
for Applied Systems Analysis); Angela Di Ruocco, Paolo Gasparini (Analisi e
Monitoraggio del Rischio Ambientale); Daniel Monfort, Charlotte Vinchon, Mendy
Bengoubou-Valerius (Bureau de Recherches Géologiques et Minières); Roger
Mrzyglocki (German Committee for Disaster Reduction [DKKV]); Kevin Fleming
(Helmholtz Centre Potsdam, German Research Centre for Geosciences [GFZ],
Potsdam)
Disasters caused by natural hazards can trigger making, the interactions between science and
chains of multiple natural and man-made hazardous practice in terms of knowledge transfer, and
events over different spatial and temporal scales. more generally to the development of capacities
Multi-hazard and multi-risk assessments make it at the local level. So far, research has focused on
possible to take into account interactions between the scientific aspects of risk assessment. But the
different risks. Classes of interactions include institutional aspects, such as the issues arising
triggered events, cascade effects, and the rapid when multi-risk assessment results need to be
increase of vulnerability during successive hazards implemented within existing risk management
(see Marzocchi et al. 2012; Garcia-Aristizabal, regimes, are also important, though they have
Marzocchi, and Di Ruocco 2013). received less attention.
Recent research has greatly increased the risk The project described here focused on the
assessment community’s understanding of institutional context of disasters, which includes
interactions between risks. Several international sets
a variety of elements ranging from sociopolitical
of guidelines and other documents now advocate
to governance components. It looked at how to
adopting an all-hazard approach to risk assessments
maximize the benefits arising from, and overcome
(for example, see UNISDR [2005]; European
the barriers to, the implementation of a multi-
Commission [2010a, 2010b]; for an overview, see
hazard and multi-risk assessment approach within
Council of European Union [2009, section 2]).
current risk management regimes. Working at two
Nevertheless, barriers to the application of multi- test sites, one in Naples and one in Guadeloupe, the
risk assessment remain. The challenges for the research team engaged with local authorities and
development of multi-risk approaches are related practitioners to better understand how to effectively
not only to the applicability of results, but also implement the results of multi-risk assessment.
to the link between risk assessment and decision Among the hazards considered were earthquakes,
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
163
164
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
volcanic eruptions, landslides, floods, tsunamis, engineers, and policy makers to reduce risk and
wildfires, cyclones, and marine inundation. Beside vulnerability. Moreover, both sites have performed
the practitioners working in the two test sites, multi-risk assessments. In Naples, two scenarios of
risk and emergency managers from 11 countries risk interactions were considered for quantitative
also provided feedback. In total, more than 70 analysis: the effect (on seismic hazard and risk)
practitioners took part in the research. of seismic swarms triggered by volcanic activity,
and the cumulative effect of volcanic ash and
Research design. The project, which aimed to
seismic loads. Both cases can be combined into a
/// ///
encourage interaction between researchers and
single scenario of interactions at the hazard and
practitioners/decision makers, began with a policy/
the vulnerability level; the combination highlights
institutional analysis—that is, desk studies of legal,
the different aspects of risk amplification detected
regulatory, and policy documents—to provide
by the multi-risk analysis (Garcia-Aristizabal,
a description of the institutional and regulatory
Marzocchi, and Di Ruocco 2013). In Guadeloupe,
framework for risk governance within different
researchers conducted a scenario analysis of cascade
natural hazard contexts and countries.
effects and systemic risk. Following a deterministic
To identify the barriers to effective decision approach, the analysis considered interaction
making in the case of multiple hazards, we then between earthquake and landslide phenomena,
engaged practitioners in interviews and focus group along with its consequences on the local road
discussions. In parallel, we performed multi-risk network in Guadeloupe and the transport of injured
assessments of some specific scenarios at the two people to hospitals and clinics (Monfort and
test sites. During workshops with practitioners, Lecacheux 2013).
we presented the results and also discussed the
Results. A first (and expected) finding is that
barriers to and benefits of implementing multi-
/// ///
risk and emergency managers rarely have the
risk assessments. Table 3-9 summarizes the key
opportunity to deal with multi-risk issues, including
research phases, the methods employed, and the
triggered events, cascade effects, and the rapid
accompanying aims.
increase of vulnerability during successive hazards.
Both test sites face multiple hazards. Naples, the Moreover, multi-risk assessments for different
biggest municipality in southern Italy, has a widely scenarios are at present rarely performed by
recognized high volcanic hazard and is also exposed practitioners at either the national or local level.
to interconnected hazards such as earthquakes, A second finding is that most participants saw the
floods, landslides, and fires. The French overseas benefits of including a multi-risk approach in their
department of Guadeloupe (Département-Région everyday activities, especially in land-use planning,
d’Outre Mer), an archipelago in the Lesser Antilles, as well as in emergency management and risk
is exposed to similar hazards (though it is less mitigation.
exposed to fires) and has a high risk of cyclones and
Practitioners identified the following as among the
tropical storms; its major geological risk is from
greatest benefits of a multi-risk approach:
the active volcano of la Soufrière and the seismic
activity along the inner Caribbean arc, both of which /// 1. Multi-risk assessment improves land-use
can trigger tsunamis and landslides. planning. ///
Both Naples and Guadeloupe have plans and According to practitioners, a multi-risk approach
policies designed to protect their citizens from provides a holistic view of the risks affecting a
these risks, and both have deployed scientists, territory and is appropriate in all geographic areas
CHAPTER
03
susceptible to several types of hazards. It would be Damage to some lifelines (water, electricity) was
helpful to have clear criteria to use in determining also taken into account. The final results of the
which scenarios would be most appropriate for a scenario determined realistic times required for the
multi-risk assessment. For landslide, for example, evacuation of the injured, either considering or not
hazard and risk mapping may not address the considering the damage to the road network and the
specific effects of different possible triggering events connectivity to lifelines of the hospitals (Desramaut
(intense rainfall, earthquakes, etc.). In the case of 2013; Monfort and Lecacheux 2013).
Naples, a detailed map with the areas susceptible
to landslides is available, but it does not include /// 3. Multi-risk assessment identifies priorities
information about the possible short-term effects of for mitigation actions. ///
volcanic eruptions, even though an eruption could
The quantified comparison of risks that would allow
produce unstable ash-fall deposits (including in
a multi-risk approach was also seen as a benefit.
low-susceptibility areas) that afterward contribute
Quantified comparison is particularly useful for
to the generation of lahars (mud flows) triggered by
identifying priorities for actions—a difficult task for
rainfall events.
policy makers, who generally rely on assessments
Urban planners emphasized how a multi-risk that do not take cascade and conjoint effects into
assessment could influence decisions about building account. The quantified comparison of risks has
restrictions, which themselves influence urban and policy implications for the planning of mitigation
economic planning—for example, by permitting actions. It can show, for example, that prioritizing
or forbidding construction of new houses and/or a particular hazard may mean giving insufficient
economic activities. weight to other hazards, and that mitigation
measures against a prioritized hazard could actually
2. Multi-risk assessment enhances
increase the area’s vulnerability to a different
///
response capacity.
hazard.
///
Practitioners asserted that emergency management
4. Multi-risk assessment encourages risk
would greatly benefit from adopting a multi-
///
awareness and cooperation.
hazard and multi-risk approach. Civil protection
///
managers were especially interested in developing Multi-risk assessment can help to increase a
multi-hazard and multi-risk scenarios to facilitate population’s awareness of natural risks, of multi-
management of emergency situations in real time risk, and of associated cascade effects. Practitioners
(Monfort and Lecacheux 2013). In Guadeloupe, for in Guadeloupe working for municipal authorities
example, evidence suggests that failure to consider noted that while the culture of primary risks (such
cascade effects (earthquake-landslide interactions) as cyclones, earthquakes, and volcanoes) is well
and to employ a systemic approach may result in established in Guadeloupe, the culture of secondary
gross underestimation of risk. The work undertaken risks (such as tsunamis, landslides, marine and
in Guadeloupe considered the interaction between inland floods, and coastal and slope erosion) is less
earthquake and landslide phenomena and its established. Practitioners from other countries
consequences for road networks and the removal indicated that communicating the results of multi-
of injured people to medical facilities. It took into risk assessment to the general population would
account the possibility that a landslide triggered help to increase awareness of secondary risk.
by an earthquake in the northwest of Basse-Terre
might cut off a main east-west road that is critical A multi-risk approach can also enhance cooperation
for moving the injured to hospitals and clinics. and foster needed partnerships between policy
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
165
166
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
makers, private sector actors, and scientists. One freely available online. The same is not true for
key to promoting such partnerships is to establish the databases, however, although the reason for
a common understanding of what multi-risk this is simple: most practitioners do not know how
assessment is, what the preferences and needs of to use them. The issue, then, is not whether data
practitioners are, and what the implications for are available, but who uses and interprets the data
regulatory instruments (related to urban planning, and for what purpose—or more fundamentally,
for example) may be. Interviewees and workshop who is able to access and present information
participants, especially from the private sector, in a meaningful and useful manner. Scientists
cited the importance of partnerships between maintain that data collected by private actors
insurers and policy makers in using improved risk (such as private consultants or insurers) are often
information for the development of risk financing not available to them, or that these data are not
schemes that cover large losses after multi-hazard collected systematically and thus cannot be used for
catastrophic events. scientific purposes.
Barriers to multi-risk assessment in
///
Practitioners and researchers also have different
the science domain. Barriers to effectively
///
views about the preferred agenda for future
implementing multi-risk assessment are found research on multi-risk assessment. Researchers
in both the science and practice domains. In the working on the technical/scientific aspects want
science domain, a major barrier involves differences to improve knowledge of the physical processes
between the geological and meteorological sciences and models related especially to cascade effects;
and the research carried out under their auspices. harmonize terminology and databases; make
These differences extend to concept definitions, uncertainty assessment a focus; combine single-
databases, methodologies, classification of the risk analyses into integrated multi-risk analyses;
integrate the results of multi-risk assessment into
risk levels and uncertainties in the quantification
existing emergency scenarios and capture cascading
process, and more. Thus each type of risk has its
effects in probabilistic terms; and conduct multi-
own scale or unit of measure for quantifying risk or
vulnerability assessment.
damages (e.g., damage states for seismic risk and
loss ratios for floods). These differences may make Practitioners on the other hand prioritize collecting
it harder for the various risk communities to share evidence about lives and property saved using a
results and may represent a barrier to dialogue on multi- versus a single-risk approach, gaining an
multi-risk assessment. overview of multi-risk contexts at the town level,
and especially learning to use and integrate new
A barrier that is more worrying for risk managers
research results in existing emergency and urban
than for researchers is the lack of open access to
plans. Depending on the practitioners themselves
risk and hazard databases, the lack of tools for
(risk versus emergency managers, regional officers,
sharing knowledge, and the difficulties associated
insurers, etc.), the needs and expectations vary
with accessing new research results. According to
extensively.
a practitioner working for a meteorological service,
“The researchers want to keep the data because Barriers to multi-risk assessment in the
///
they want to publish.” Another practitioner stated: practice domain. Differences in the approaches,
///
“Private companies and research institutions often tools, and methodologies used for single-risk
do not make their data available . . . for the benefit assessment have resulted in a lack of integrated
of their competitiveness.” Scientists view the matter practices for multi-risk governance. Especially where
differently and maintain that research results are risks are managed by authorities acting at different
CHAPTER
03
governmental levels, cooperation among institutions which—depending upon the risk, the country, and
and personnel is a challenge. The priorities of the the availability of insurance schemes—may differ.
various agencies vary extensively, and there may Different levels of responsibility are attributed to
be insufficient financial capacity to cover them all. property owners in geological versus meteorological
In some cases a multi-risk approach is perceived
risk prevention, for example. In the case of
as competing with (rather than complementing)
earthquakes, the level of individual responsibility
single-risk approaches.
is high (given that property owners are usually
Capacities, mainly financial, but sometimes also in charge of household vulnerability reduction
technical and institutional, are especially lacking at measures). In the case of floods, public authorities
the local level, even though responsibility for DRM have responsibility for decisions about risk
often falls to local authorities or private actors.
mitigation measures such as protection works, and
The transfer of responsibility for disaster risk
the costs are covered collectively. In general, there
reduction to the local level (to the municipal level
are few options for public-private responsibility
in many European countries) has often occurred
without sufficient resources for implementing sharing, especially for households exposed to
necessary programs (UNISDR 2005b, 2013). multiple risks (and especially where insurance
Private actors, especially property owners, are schemes are not available, as is the case in some
being given increasing risk-related responsibilities, European countries).
96
167
168
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-21. Build Back Better: Where Knowledge Is Not
Enough71
Authors: Jason Brown (Australia-Indonesia Facility for Disaster Reduction);
Jonathan Griffin (Geoscience Australia)
Understanding risk and knowing how to prepare for of houses, and the potential for even larger
and mitigate the potential effects of natural disasters earthquakes within the next few decades (Sieh et al.
are critical for saving lives and reducing economic 2008) made clear that the affected population would
losses. But is knowledge enough? Between 2009 and need to start building back better to avoid a similar
2013, the Australia-Indonesia Facility for Disaster catastrophe in the future.
Reduction tested the premise that improved
A post-disaster engineering survey in October–
knowledge would result in changed risk behavior November 2009 assessed how different types of
among earthquake-affected populations. AIFDR’s building performed during the earthquake. The
work in West Sumatra found that better risk survey was followed by an 18-month province-wide
knowledge had limited impact on risk behavior, even Build Back Better campaign based on the slogan
among communities that had recently experienced Bukan Gempanya Tapi Bangunannya! (It’s Not
a traumatic earthquake event. This finding raises the Earthquakes, But the Buildings!). Finally, an
important considerations for governments, donors, evaluation was undertaken to analyze the impact
and program implementers seeking improved DRM of the campaign and specifically to learn about
outcomes, particularly in the early recovery and recovering communities’ motivations for engaging
disaster rehabilitation phases. in safer building practice. Each of these elements is
discussed below.
The magnitude 7.6 earthquake that struck West
Sumatra on September 30, 2009, claimed more than Engineering survey. Though the damage done
/// ///
1,100 lives, injured 3,000, destroyed or damaged by the 2009 earthquake was reasonably well
documented (it destroyed 119,005 houses and
over 270,000 houses, and affected more than 1.25
damaged 152,535),72 there was little documentation
million people in 13 of West Sumatra’s 19 districts.
of how many houses were undamaged and what
Water supply, electricity, and telecommunications
made those structures more resilient. Nor was the
were severed, and many government office
information on damaged structures disaggregated by
buildings collapsed, paralyzing services and making
construction type, age of construction, and ground
emergency response difficult. Damage and losses
shaking experienced.
were estimated at US$2.3 billion, with about 78
percent of all needs concentrated in the housing To fill this gap, AIFDR and the Indonesian National
sector (BNPB and Bappenas 2009). Disaster Management Agency (BNPB) supported a
comprehensive engineering survey jointly led by the
The earthquake exposed a combination of poor Bandung Institute of Technology and Geoscience
housing design, poor housing construction, and Australia, with additional expertise supplied by
weak settlement planning (BNPB and Bappenas Andalas University, Padang. This team consisted
2009). The enormity of the damage, the need for of 70 members with engineers from Indonesia,
reconstruction and repair of hundreds of thousands Australia, New Zealand, and Singapore.
Figure 03—23
Two common
housing types in
Padang: Unreinforced
masonry construction
using river stone
and mortar with no
reinforcement (left)
versus confined
masonry construction
using steel-reinforced
concrete columns in
the corners and tops
The engineering survey included a comparison campaign’s messages by radio, and an estimated 2.7
of walls (right).
of two common housing types: (a) unreinforced million people by television.
masonry—typically houses built from bricks, river Source: Australian
Evaluation. To determine how successful the
stone, or similar material, and mortar; and (b) Department of Foreign
/// ///
campaign was in reducing barriers to behavior
confined masonry—houses built from bricks and Affairs and Trade, 2009.
change, an evaluation was carried out to see
mortar with simple concrete and steel reinforcing
whether homeowners had been influenced to adopt
(figure 3-23). The results were unambiguous.
earthquake-safe building techniques. 73
Overall, unreinforced masonry houses in heavily
shaken areas were 5 times more likely to suffer The evaluation of the Build Back Better campaign
damage than confined masonry and 10 times more found that knowledge does not translate into action.
likely to collapse (Sengara et al. 2010). “The population in West Sumatra has received and
internalised general information about earthquake
Build Back Better campaign. The AIFDR and
safer construction,” the study found, but “when
/// ///
BNPB expected that the rebuilding process would
rebuilding their homes, they failed to act on this
motivate the people of West Sumatra to prepare
knowledge” (Janssen and Holden 2011, 7). More
themselves for future earthquakes. This preparation
specifically: approximately half of the families
seemed even more important because the risk of a
in West Sumatra were knowledgeable about
larger, potentially tsunamigenic earthquake in the
earthquakes, related risks, and available mitigation
same general area within the next few decades had strategies, partly as a result of the campaign;
not been diminished (Sieh et al. 2008; McCloskey et respondents found it difficult to remember exact
al. 2010). Despite the increased costs associated with technical specifications; there was a high level of
building earthquake-resistant houses (estimated indifference to, and no social or political pressure
at around 30 percent more than a typical house), it for, promoting safer building techniques for housing
was assumed that—given the impact experienced by (Janssen and Holden 2011).
the West Sumatra population and the trauma felt by
many families—residents who rebuilt their houses Perhaps the most intriguing finding of the
would be open to applying new knowledge of safe evaluation was that the earthquake itself had little
building techniques to build safer houses. impact on people’s resistance to change. Specifically,
the campaign’s key assumption, that the experience
For West Sumatrans to build back safer, individuals of the earthquake would lead the population of
needed to understand that building a safer house West Sumatra to be more willing to build back
was possible, and they needed to know how to get better, was not true. Janssen and Holden (2011)
technical assistance if they needed it. Between found that those living in the worst-affected areas
February 2010 and June 2011, the Build Back Better demonstrated possibly higher resistance to change
campaign ran public service announcements 8,192 than those in less-affected areas. The influence of
times on radio and 2,275 times on television. An the earthquake on safe building practice seemed
estimated 1 million people were exposed to the to be limited to those who had gone through
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
169
170
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
a traumatic, first-hand experience during the The Build Back Better campaign highlights
earthquake, such as being trapped or injured by two key lessons: Knowledge is important for
falling debris. reducing resistance to change and for promoting
contemplation of change to safer building
The evaluation found that reducing people’s
techniques. But it is not enough to ensure action.
resistance to change was a precondition for getting The post-campaign evaluation found several barriers
them to contemplate change, but it also found that that kept people from moving past contemplation of
actual exposure to the earthquake did not affect the change to action. These included a lack of resources
degree to which they were contemplating change. (more than half of respondents said safer building
Exposure to loss of assets or even loss of life techniques were too expensive); inadequate access
appeared to make no difference. to technical information; mistrust of construction
The researchers identified and tried to understand workers or building supply store employees, who
a dramatic gap between knowledge and practice— respondents feared were trying to mislead or cheat
that is, to understand why the information and them; and incentives and disincentives, such as a
knowledge did not translate into action. This lack of enforced building standards for local housing
conundrum was highlighted in answers to the and a lack of social and/or financial incentives.
following line of questioning: As a follow-up to the Build Back Better campaign
evaluation, a laboratory-style safe construction
1. When asked what would be the most disruptive
program showed that given the correct combination
event that could take place in a person’s life,
of timely information, technical training, community
most respondents answered “a natural disaster.”
supervision, and financial and nonfinancial
2. When asked what would be the worst possible incentives and disincentives, individual homeowners
consequence of a natural disaster, 62 percent will put knowledge into practice. It showed further
replied: “A family member getting killed.” that the timing of interventions is critical. Janssen
and Holden (2013) propose that government
3. When asked what was the main cause of people
subsidies be invested in immediate needs (including
getting killed in an earthquake, 80 percent the provision of easy-to-build, cheap, temporary
replied: “Collapsing buildings.” shelter) concurrently with livelihood support
4. When asked whether their houses were strong programs that enable communities to more quickly
enough to withstand an earthquake, 67 percent recover from the disaster event. Immediately after
said “No.” an earthquake, most people are trying to get on with
their lives with the resources available to them, and
5. When asked what could be done to make their the effect of the earthquake on reducing resistance
houses safer in the face of an earthquake, 68 to change is negligible. Once livelihoods are
percent could provide three correct building reestablished, programs to facilitate construction
techniques to improve the house. of permanent, earthquake-resistant housing may be
more effectively implemented using appropriately
6. Considering that retrofitting a house takes about
targeted incentives or disincentives.
three months, the respondents were asked what
they would do if they were certain an earthquake The AIFDR initiatives have unveiled a rich array of
would hit in six months: 68 percent said data and experience that can assist in the design
they would run away, while 1.2 percent said of both pre- and post-disaster programs into the
they would retrofit their houses to make them future. The Build Back Better experience showed
earthquake safe. that understanding and effectively communicating
CHAPTER
03
risk information and risk reduction strategies disaster risk reduction to the extent we would
is necessary but does not on its own lead to intuitively assume, interventions may be more
behavioral changes. Interventions must consider, successful after livelihoods and a sense of normalcy
and experiment with, incentives and disincentives have been reestablished. Identifying barriers
for acting on risk knowledge. Because communities to action within the local context is crucial to
recovering from a major disaster may not prioritize achieving change.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
171
172
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
3-22. InaSAFE: Preparing Communities to Be a Step
Ahead74
National Disaster Management Agency, Indonesia (BNPB); 75 World Bank;
Global Facility for Disaster Reduction and Recovery; Australia Government76
Indonesia is one of the most disaster-prone and The subnational focus of InaSAFE is intended to
populous countries in the world. Its disaster improve the capacity of local governments and
managers and local government planners recognize communities to make more informed disaster
the importance of investing in preparedness, but preparedness decisions. The InaSAFE tool is linked
have faced many obstacles to accessing and using to Indonesia’s disaster preparedness standards, and
up-to-date and accurate data from hazard and risk as part of its analysis it suggests various actions
assessments. Unfortunately, there is a tendency for local governments to consider in response to
for technical studies that analyze risk to end up a hazard scenario. So far AIFDR’s core partners
on a shelf or archived on a hard drive. InaSAFE have trained more than 150 Indonesian disaster
(originally the Indonesian Scenario Assessment managers across six provinces to use InaSAFE,
for Emergencies), an open source disaster impact and have provided the necessary skills for disaster
managers to collect their own hazard and exposure
modelling tool, was launched in 2012 to help
information through links with science and
overcome obstacles to understanding and using
mapping agencies and the use of crowdsourcing
impact information. Developed by Australia and
techniques. Furthermore, complementary
Indonesia in collaboration with the World Bank
programs in partnership with the Humanitarian
and GFDRR,77 InaSAFE enables communities,
OpenStreetMap Team (HOT) have promoted
local governments, and disaster managers to
the use of OpenStreetMap participatory mapping
use realistic natural hazard scenarios for floods,
technology to supplement government baseline data
earthquakes, volcanoes, and tsunamis to underpin
and prepare key inputs for InaSAFE, leading to over
emergency planning, disaster preparedness, and
1.4 million buildings being mapped throughout high-
response activities.
risk areas in Indonesia.
To date, InaSAFE has been used to develop
How InaSAFE works. InaSAFE is usable by anyone
/// ///
disaster impact scenarios for national government
experienced in disaster management and possessing
disaster exercises in Indonesia, including the 2014 basic computer skills. Users answer a series of
International Mentawai Megathrust Tsunami questions posed by the tool about a potential
Exercise. It has been implemented in Jakarta, East disaster scenario; the tool then combines hazard
Java, and South Sulawesi to develop realistic flood models or footprints with exposure information
scenarios for contingency planning. During 2014, the to produce impact analysis—specifically, reports
Australia-Indonesia Facility for Disaster Reduction estimating the potential damage caused to people
and Indonesia’s National Disaster Management and facilities, maps of affected areas, and lists of
Agency (BNPB) will focus on helping district recommended actions to assist disaster managers in
disaster management facilitators and universities to decision making. InaSAFE is capable of integrating a
develop the necessary skills to use, and train others wide range of data sets developed by various groups
to use, the InaSAFE methodology. (scientists and engineers; international, national,
and local institutions; NGOS and communities). buildings in Indonesia. Since 2011, this program Figure 03—24
Table 3-10 lists the currently available hazard inputs has successfully mapped more than 1.4 million InaSAFE can be
for InaSAFE 2.0, the version released in February structures, and OSM now forms a key part of used to improve
2014; table 3-11 lists the currently available exposure the ongoing capture of local knowledge. These understanding of the
data; and table 3-12 lists sample impact functions. valuable data sources are critical elements of the impact of disaster
InaSAFE engagement, where new analyses can events, such as floods
InaSAFE’s openness, scalability, and adaptability
be dynamically run whenever the information in Jakarta.
make it an especially valuable tool for users seeking
is updated.
information about hazards and their impact. Source: World Bank.
A variety of other characteristics contribute to • Focus on social vulnerability. InaSAFE has
its utility: been designed to take into account gender and
age as part of the impact analysis for vulnerable
• Integration of latest science with local
groups. For example, the impact analysis results
knowledge. To ensure disaster managers have
specify steps that must be taken to meet the
access to the best information to support their
needs of pregnant or lactating women (such as
decisions, AIFDR is working through Geoscience
providing additional rice) and of infants and the
Australia in partnership with the Indonesian
elderly (such as providing extra blankets).
Geological Agency (Badan Geologi), Indonesian
Agency for Meteorology, Climatology and • Demand-driven development. InaSAFE
Geophysics (BMKG), Indonesian Institute of started through a partnership with BNPB and
Science, and Bandung Institute of Technology to was intended to address the needs of subnational
improve the scientific knowledge about hazards disaster management agencies conducting
in Indonesia and to supply up-to-date hazard emergency contingency planning. Disaster
information to subnational disaster management managers and scientists are still working
agencies. In addition to using population data collaboratively to develop InaSAFE, with the
and demographic information from the national majority of requests for new development coming
census, AIFDR piloted a participatory mapping from Indonesian national government officials
program through a grant to HOT to map and provincial disaster managers, who continue
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
173
174
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Table 03—9 HAZARD TYPE MODEL HAZARD FOOTPRINTS
Hazard Data Accepted Ground shaking (Modified Mercalli
Earthquake
in InaSAFE 2.0 Intensity)
Tsunami Maximum inundation depth (meters)
Volcanic eruption–ash fall Ash load (kg2/m2) Hazard zones
Flood Maximum inundation depth (meters) Flood-prone areas
Tropical cyclone, storm surge Wind speeds, inundation depth (meters)
Table 03—10 EXPOSURE TYPE SUB-TYPE
Exposure Data Population Density (people/units2)
Accepted in InaSAFE Buildings Schools, hospitals, public buildings
2.0 Other structures Bridges, telecommunications, etc.
Roads Major, minor
Land use Agriculture, industrial
Table 03—11 EVENT OUTPUT
Sample Impact Number of fatalities and displaced persons; number of
Earthquake
Functions buildings affected
Number of people affected; number of people to be
Tsunami
evacuated
Volcanic ash fall Number of buildings affected
Number of people affected; number of people needing
Flood
evacuation; number of buildings closed and/or damaged
Number of fatalities and displaced persons; number of
Earthquake
buildings affected
Number of people affected; number of people to be
Tsunami
evacuated
Volcanic ash fall Number of buildings affected
Number of people affected; number of people needing
Flood
evacuation; number of buildings closed and/or damaged
CHAPTER
03
to request (and receive training in) the use of automatically pushed to a BNPB server, where an
InaSAFE. This training increases the capacity InaSAFE impact assessment is produced within
of local governments and communities to use minutes to inform rapid disaster response. The
scientific and local knowledge to inform disaster results are also shared with the public on the
preparedness decisions. BNPB website (http://bnpb.go.id).
• Client focus. Since its beta release at the • Tested in multiple contexts. InaSAFE has
Understanding Risk Forum in Cape Town in July been used to produce impact assessments for
2012, InaSAFE has been downloaded over 1,000 earthquakes in Yogyakarta, for a tsunami in
times. Since InaSAFE is an open source tool, the Padang, and for community-level flood scenarios.
InaSAFE user community is helping national Most recently, during the Jakarta floods of
governments to tailor the software to members’ 2012–2013 and 2013–2014, reports of flooding
needs. from village heads were joined with the subvillage
boundaries captured through participatory
• Effectiveness across DRM decision making.
mapping. This flood footprint was used by the
From its beginnings as a tool to aid in preparing
Jakarta disaster management agency and the vice
for disasters, InaSAFE has been used effectively
governor of Jakarta to illustrate the change in
to visualize critical infrastructure (such as
flooding over time.
schools, hospitals, or roads) in flood-prone areas
across Jakarta. As InaSAFE develops in response • Award-winning software development.
to client requirements, its relevance to all InaSAFE was called one of the top 10 “open
parts of the DRM cycle increases. In the future, source rookies of the year in 2012”—alongside
InaSAFE could support risk-based land-use software developed by Microsoft, Yahoo!, and
planning, determine priorities for infrastructure Twitter.78 This recognition not only affirms
retrofitting, generate real-time impact forecasts the technical merits of the software and its
for a variety of hazards, and contribute to post- commitment to open source philosophies, but
disaster needs assessments or pre/post damage also highlights the exemplary multi-institutional
and loss assessments. collaborative development of InaSAFE.
• User contributions. As part of the InaSAFE • Dynamic and inclusive software
approach to developing contingency planning and development. In February 2014, InaSAFE 2.0
preparedness scenarios, OSM tools are used to was released with new features that had been
capture high-resolution baseline geographic data requested by disaster managers, including road
on critical infrastructure. In Jakarta in 2012, in exposure data, additional map customization,
partnership with AIFDR, HOT, World Bank, and and InaSAFE reporting. This version marks the
UN Office for the Coordination of Humanitarian first release with contributions from developers
Affairs, the provincial disaster management focused on applications outside of Indonesia,
agency (BPBD-DKI Jakarta) pioneered a data such as the addition of new population impacts
collection program to map over 6,000 critical from the Philippines by partners at Environment
infrastructure locations and 2,668 subvillage Science for Social Change Inc.
boundaries within OSM.
InaSAFE global. Preparing for a disaster requires
/// ///
• Real-time analysis. Through its collaboration people from various sectors and backgrounds to
with AIFDR and BNPB, BMKG produces ground- work together and share their experience, expertise,
shaking maps following an earthquake. These are and resources. Using InaSAFE to develop a scenario
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
175
176
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
requires the same spirit of cooperation and same and disaster contingency plans. Three LGUs were
sharing of expertise and data. The more sharing assisted with the mapping of critical public buildings
of data and knowledge there is by communities, using OSM and with analysis of flood impacts
scientists, and governments, the more realistic and using InaSAFE. This initiative has also supported
useful the InaSAFE scenario will be. customization of InaSAFE based on localized needs,
including functionality for analysis of detailed
It is in this spirit that further application of the population data and the integration of InaSAFE with
platform in other countries and regions is being the web-based tools of the Philippines Department
planned as part of the GFDRR–World Bank Open of Science and Technology’s Project NOAH
Data for Resilience Initiative (for more information, (Nationwide Operational Assessment of Hazards).79
see section 3-1). InaSAFE has shown itself to be an
efficient and credible way to save agencies time and In Sri Lanka, significant investment by the
resources in developing risk assessment information government in OSM is being capitalized through
and hazard impact modelling tools. Hence a number InaSAFE and QGIS training (for more detail, see
of governments in other countries have expressed section 3-2). This work has demonstrated the power
interest in using, improving, and refining the of InaSAFE to dynamically pull data from OSM and
InaSAFE tool. the Sri Lanka GeoNode for analysis. In particular,
it has triggered significant interest in InaSAFE as
In the Philippines, a partnership between the a fundamental tool for disaster management in Sri
World Bank and Local Government Units (LGUs) Lanka and has led to widespread interest in the open
focused on the preparation of risk-sensitive land-use source QGIS software, both of which will continue
plans, structural audits of public infrastructure, to be supported in years to come.
Figure 03—25
QGIS2.0 with the
InaSAFE 2.0 dock
showing a map and
indicative results for
an assessment of the
impact of flooding on
roads in Jakarta.
CHAPTER
03
3-23. Global River Flood Risk Assessments
Philip J. Ward (Institute for Environmental Studies and Amsterdam Global
Change Institute, VU University Amsterdam); contributing authors Brenden
Jongman, Jeroen C. J. H. Aerts (Institute for Environmental Studies and
Amsterdam Global Change Institute, VU University Amsterdam); Arno Bouwman
(PBL Netherlands Environmental Assessment Agency); Rens van Beek, Marc
F. P. Bierkens (Department of Physical Geography, Utrecht University); Willem
Ligtvoet (PBL Netherlands Environmental Assessment Agency); Hessel C.
Winsemius (Deltares)
The economic losses associated with flooding are Brakenridge 2013). However, Gall et al. (2011) also
huge. Reported flood losses (adjusted for inflation) found evidence for non-exposure-driven increases in
have increased globally from US$7 billion per year disaster losses in the United States over the period
during the 1980s, to US$24 billion per year in the 1960–2009, pointing to changes in hazard frequency/
period 2001–2011.80 In response, the scientific intensity as possible drivers of risk.
community has developed a range of models for
assessing flood hazard, flood exposure, and flood Several global flood risk assessment models have
risk at the global scale.81 These are being used to been developed in the last decade. Initially, these
assess and map the current risk faced by countries models provided estimates of risk under current
and societies. Increasingly, they are also being used conditions (i.e., they did not account for changes in
to assess future risk, under scenarios of climate climate and/or socioeconomic development).
change and/or socioeconomic development.
The earliest of these was the “hot spots” project of
The growing number of global-scale flood risk the World Bank, which sought to provide “a spatially
models being used for an increasing range of uniform first-order, global disaster risk assessment,”
applications is mirrored by the growth of events including the risk of flooding (Dilley et al. 2005a).
and networks specifically focusing on global-scale Maps were developed showing risk severity at a
floods and global-scale flood risk assessment. For spatial resolution of about 2.5’ x 2.5’ (about 5km x
example, the Global Flood Working Group82 has 5km at the equator), categorized into deciles. The
been established by the Joint Research Centre of maps were based on a georeferenced data set of
the European Commission and the Dartmouth past extreme flood events between 1985 and 2003
Flood Observatory. from the Dartmouth Flood Observatory, combined
with gridded population maps. The flood extent
A large number of studies have attempted to assess data were based on regions affected by floods, not
trends in past (flood) risk, based on reported necessarily on actual flooded areas. Nevertheless,
losses in global loss databases, such as the EM- the project was successful in identifying global
DAT database83 and MunichRe’s NATHAN and disaster risk hot spots, and since then improved
NatCatService databases (e.g., Barredo 2009; flood risk maps have been developed for the
Bouwer 2011; Neumayer and Barthel 2011). GAR2009 (UNISDR 2009), which based flood
These studies have found that reported losses extent data on the modelling approach of Herold
have increased over the last half century, mainly and Mouton (2011) and produced global hazard
because of increased exposure, such as population maps for a limited number of flood return periods.
growth and the location of assets in flood-prone These data were combined with high-resolution
regions (IPCC 2012; Kundzewicz, Pińskwar, and maps of population and economic assets, as well
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
177
178
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
as indicators of vulnerability, to develop maps of urban area). A description of the model framework
current flood risk at a spatial resolution of 1km x (Winsemius et al. 2013) included a case study
1km. Pappenberger et al. (2012) have developed a application for Bangladesh (figure 3-26), in which
model cascade for producing flood hazard maps changes in annual expected damage were projected
showing flooded fraction at a 1km x 1km resolution between 2010 and 2050. These preliminary results
(resampled from a more coarse 25km x 25km grid). showed that over that period, risk was projected
The cascade can be used to develop flood hazard
to increase by a factor of 21–40. Both climate
maps for different return periods but has not yet
change and socioeconomic development were
been used to assess risk.
found to contribute importantly to this increase
As part of recent efforts to project changes in risk in risk, although the individual contribution of
in the future under scenarios of climate change socioeconomic development is greater than that
and socioeconomic development, Jongman, Ward, of climate change. The model was then further
and Aerts (2012) assessed and quantified changes developed and applied at the global scale (Ward et
in population and assets exposed to 100-year flood al. 2013b). GLOFRIS is currently being used within
events between 1970 and 2050. Combining the flood and outside the scientific community to assess
hazard maps developed for the GAR with projections changes in flood risk at the global scale under a wide
of changes in population and GDP, they found that range of climate and socioeconomic scenarios.
Figure 03—26 socioeconomic development alone is projected to
Observed flood drive an increase in the global economic exposure to Also in 2013, Hirabayashi et al. (2013) developed a
extents in flooding between 2010 and 2050 by a factor of 3. global inundation model, and combined this with
Bangladesh during high-resolution population data, to assess and map
In 2013 and 2014, three new global flood risk
July and August the number of people exposed to 10- and 100-year
assessment models were presented. The first of
2004: Dartmouth flood events at a spatial resolution of 15’ x 15’ (about
these was GLOFRIS (GLObal Flood Risk with Image
Flood Observatory Scenarios) (Ward et al. 2013b; Winsemius et al. 30km x 30km at the equator). They then used this
database versus 2013). GLOFRIS estimates flood risk at a spatial model to quantify the change in the number of
GLOFRIS model. scale of 30” x 30” (about 1km x 1km at the equator), people affected by 10- and 100-year floods between
whereby risk is expressed as several indicators the periods 1970–2000 and 2070–2100. The study
Source: Winsemius et al. (annual exposed population, annual exposed GDP; used discharge data from 11 global climate models
2013. annual expected urban damage, and annual affected and for four different scenarios of climate change.
Dart m out h m om ent of flooding GLOFRIS m om ent of flooding
Aug. 10 2004
27°N 27°N Aug. 5 2004
July 30 2004
hi
hi
s
s
26°N 26°N
Bra hm ap ut ra Bra hm ap ut ra
Ko
Ko
July 28 2004
Gan ge s Gan ge s
25°N 25°N July 25 2004
am una
am una
24°N 24°N
July 22 2004
a
a
hn
hn
July 20 2004
eg
eg
J
J
M
M
23°N 23°N
July 15 2004
22°N 22°N July 10 2004
km km
0 125 250 0 125 250
No flooding
85°E 86°E 87°E 88°E 89°E 90°E 91°E 92°E 93°E 85°E 86°E 87°E 88°E 89°E 90°E 91°E 92°E 93°E
50-year flood depth [m]
Figure 03—27
10.0
Map of modelled
inundation extent
5.0 and depth in Nigeria
using GLOFRIS. Maps
3.0 of this type can be
used to assess which
11°N 2.0 areas are exposed to
flooding.
1.5
1.0
7°N
0.5
0.3
0.1
3°N
1°E 5°E 9°E 13°E
0.0
More recently, Arnell and Lloyd-Hughes (2014) used World Bank Africa Disaster Risk Management team
a simpler method to assess changes in flood risk carried out the National Flood Risk Management
between 1960–1990 and two future time periods Implementation Plan for Nigeria.
(2050s and 2080s), using results from 19 global
climate models, four climate scenarios, and five At the time, little information was available for
scenarios of socioeconomic development. This assessing the level of flood risk in Nigeria. At the
study found that under a “middle-of-the-road” request of the GFDRR and World Bank’s Africa
socioeconomic scenario, climate change by 2050 team, GLOFRIS was used to carry out a rapid
would lead to an increased exposure to river flood assessment of flood risk per state in Nigeria. Maps
risk for between 100 million and 580 million people,
were produced showing the expected extent of
depending on the climate change scenario.
flooding for different return periods (figure 3-27),
Using the results of global-scale river flood
///
as well as the annual affected population per state
risk assessments in practice. The results of
///
(figure 3-28). The model and its results were “a
global-scale river flood risk assessment have been great first step in providing a national map showing
applied in practice. Several examples are described vulnerability to floods for Nigeria, where previously,
below. no such methodologies were in place.” However,
an assessment of the number of people affected by
State-level flood risk in Nigeria. In 2012, floods
in Ibadan, Nigeria, killed hundreds of people, different inundation depths was found to be critical,
displaced over 1 million people, and destroyed crops. as the difference between 10cm and 1m of flood
A post-disaster needs assessment carried out by the inundation is clearly significant. Since GLOFRIS
GFDRR urgently recommended strengthening the had been developed in a flexible manner, it was easy
country’s resilience to flooding, and in response the to integrate this request into the model structure,
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
179
180
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
and tailor the output to the needs of the model’s a return period of 1,000 years or less, are shown
end-users. in figure 3-29. In all regions, the urban population
living in flood-prone areas is projected to grow
Present and future urban flood risk. In 2014, rapidly between 2010 and 2050, while in almost all
UN-HABITAT will publish the fourth edition of its regions the rural population living in flood-prone
report on urban water and sanitation. This is the areas is projected to decline. An exception is Sub-
first edition to project conditions into the future Saharan Africa, where the rural population living in
and to treat flood risk. GLOFRIS is being used to flood-prone areas is projected to continue growing
project present and future flood risk in the world’s after 2030.
cities (PBL 2014), based on the scenario study for
the Organisation for Economic Co-operation and GLOFRIS has also been used to assess the increase
Development Environmental Outlook to 2050 in annual exposed GDP between 2010 and 2050,
(OECD 2012). as well as to give a preliminary assessment of how
much the overall risk could be reduced by improving
GLOFRIS has been used to project changes in flood protection standards. Figure 3-30 shows the
annual exposed population and annual exposed GDP annual exposed GDP in urban and rural areas for
to flooding, aggregated to the World Bank regions. 2010 and 2050, assuming different flood protection
Projections of the number of people living in flood- standards. The figure suggests that in all regions, the
prone areas, defined as areas exposed to floods with risk is projected to increase substantially between
Figure 03—28
Maps of Nigeria
showing the modelled
results of the number
of people affected
per state (expressed
as a percentage of
the total population
per state) for floods
of different severities.
Maps of this type
can be used for
identifying risk hot
spots.
Figure 03—29
People living in
flood-prone areas
in different regions,
2010–2050.
Source: PBL 2014.
PBL NL
Note: Flood-prone areas
are defined as areas with a
probability of a flood once
in 1,000 years or less. Note
different scales on y-axes.
PBL NL
2010 and 2050, and also that better protection also be provided. These new Aqueduct map layers
standards could significantly reduce flood risk. will help identify where new flood risks will emerge
and how severe they will be, what their potential
The World Resources Institute’s Aqueduct causes are, and how best to adapt to, mitigate, or
Water Risk Atlas. This atlas (available at http:// prevent them.
aqueduct.wri.org/) offers a suite of interactive maps
that help people better understand where and Changes in future flood risk due to inter-
how water risks and opportunities are emerging annual variability. Knowledge is this area is less
worldwide. Most of the current available map layers well developed. GLOFRIS is currently being used to
focus on water resource availability and droughts. determine whether flood risk might be increased or
Aqueduct will be extended to include global-scale reduced as a result of naturally occurring variations
flood risk maps based on GLOFRIS. The maps will in the climate system, like the El Niño Southern
show the current level of river flood risk, per sub- Oscillation (ENSO), and if so, how this information
catchment, across the globe, expressed in indicators might be used by the (re)insurance industry.
such as the annual affected number of people and Research is beginning to show that flood hazard and
level of economic risk. Future scenarios of risk will risk are indeed strongly correlated to ENSO at the
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
181
182
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
global scale (Ward et al. 2010, 2013a, 2014); the Risk when these limitations are recognized and
Prediction Initiative, based at the Bermuda Institute communicated, and the models are used to answer
of Ocean Science, is facilitating the translation appropriate questions. But because of the models’
of this research into usable results for insurance limitations, their results should not be used in
and reinsurance companies. For example, claims all situations.
may increase (or decrease) in particular ENSO
phases, affecting the amount of financial resources The matter of spatial resolution is very important.
necessary for covering eventual losses. Although many global hydrological models run
with grid cells of approximately 50km x 50km, for
Limitations of global-scale river flood risk
modelling impacts a higher resolution is preferable,
///
assessments, and how they should not
be used in practice. Global-scale flood risk
///
since the impacts of flooding are dependent on
assessment models are coarse by their very nature, physical and socioeconomic processes at a much
and represent both physical and socioeconomic finer scale. Hence, flood risk research should aim
processes in simplified ways. This is not a problem to simulate floods at a higher resolution than the
Figure 03—30
Annual exposed GDP
to flooding in 2010
and 2050, under
different assumptions
of flood protection
PBL NL
standards.
Source: PBL 2014.
Note: Y-axes use different
scales.
PBL NL
CHAPTER
03
native 50km x 50km grid size of global hydrological A final limitation of the global modelling approaches
models. described here is that they do not capture pluvial
floods or local-scale flash flood events. While flash
Geographical scale is also an issue. Although a floods cause many human fatalities in some parts
1km x 1km grid may be appropriate for calculation of the world (Gaume et al. 2009), their local-scale
purposes, the actual model outcomes at this character makes it challenging to simulate their
resolution are subject to huge uncertainties. probability and extent at the global scale.
Presenting results for a given grid cell is not
encouraged, since it may give a false sense of safety, Main research needs for the coming 5–10
///
or indeed of risk. Moreover, global models are not years. Increases in available computational power
///
intended to give assessments of risk at this high are allowing global hydrological models to adopt
resolution, but rather to indicate risk, and relative finer spatial resolutions, a development that will
changes in risk, across larger regions, such as create new scopes for application and raise new
continents, countries, river basins, and states. A research questions.
high-resolution detailed flood risk map for a city,
district, street, or building requires a more detailed To date, the accurate representation of vulnerability
modelling approach, as well as more detailed local has been one of the largest obstacles in large-scale
knowledge and interaction with local stakeholders. flood risk assessment. Large-scale risk studies either
have not incorporated the vulnerability of exposed
To date, global-scale river flood risk models have people and assets (Hirabayashi et al. 2013; Jongman,
generally assessed flood risk under the assumption Ward, and Aerts 2012; Nicholls et al. 2008), or have
that no flood protection measures are in place (an done so in a highly stylized manner (e.g., Feyen et
issue addressed in the case study in section 3-8). In al. 2012; Ward et al. 2013b; Rojas, Feyen, and Watkiss
reality, many regions are protected by infrastructural 2013). Anecdotal evidence from studies at more local
measures up to a certain design standard. Ward et to regional scales suggests that societies become less
al. (2013b) assessed the sensitivity of global flood vulnerable over time. An improved understanding
risk modelling results to this assumption. Under the of temporal changes in vulnerability, and their
assumption of no flood protection measures, they influence on risk, is a research priority.
simulated annual expected urban damage of about
US$800 billion (PPP) per year. Another priority is improving the representation of
exposure in global flood risk models. While high-
However, assuming protection standards of 5 and resolution and high-quality gridded data sets of
100 years globally, this estimate fell dramatically, current population, GDP, and land use are available,
by 41 percent and 95 percent, respectively. Clearly, and provide useful proxies for representing current
then, existing flood protection standards should be exposure, high-resolution projections for population
included in global flood risk assessment models. and GDP are only beginning to become available;
and land-use projections at the required resolution
It is possible to incorporate flood protection are still scarce. Recently, a first global forecast
standards in flood risk assessments to assess the model of urban development was presented that
impacts of different strategies to reduce risk; simulates urban expansion at a horizontal resolution
see for example Jongman et al. (2014) for such of 1km x 1km resolution, based on empirically
an assessment on continental scale. But such derived patterns (Seto, Güneralp, and Hutyra 2012).
assessments should be used only for assessing the Once available publicly, such high-resolution data
large-scale effects of strategies, and not the detailed could provide important new information in global
effects of individual measures. For example, the flood risk studies.
global model could be used to assess how much
a country could reduce its risk by increasing the The need for a coherent database of current flood
protection standard of its dikes and levees. But protection standards is becoming more and more
it should not be used to dimension individual important. Preliminary efforts to include flood
dike sections. protection standards in large-scale flood risk
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
183
184
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
assessments have been presented (Hallegatte et al. change the duration, depth, and flow velocities
2013, Ward et al. 2013b; Jongman et al. 2014) using attained during inundation events. This fact has
simplified assumptions and scenarios. These studies severe implications for the resulting hazard, and
show that the flood protection standards assumed its simulation requires an improved representation
in the modelling process have a huge effect on the
of the relevant processes in hydrological models.
overall modelled risks. This finding illustrates the
In addition, new research suggests that natural
potential benefits of adaptation, but also shows
that uncertainty in flood protection standards can ecosystems should be incorporated as important
strongly affect model outcomes. In particular, flood means of protection against floods, for both river
protection measures will modify the magnitude and flooding (Stürck, Poortinga, and Verburg 2013) and
frequency along the drainage network and locally coastal flooding (Arkema et al. 2013).
Figure 03—31
Historical tropical
cyclone tracks for
the period 1981–2000
(top) and tropical-
cyclone-like vortices
extracted from a 20-
year simulation using
a general circulation
model (bottom).
Source:
Geoscience Australia.
Note: TC = tropical cyclone.
CHAPTER
03
3-24. Delivering Risk Information for a Future Climate in
the Pacific85
W. C. Arthur, H. M. Woolf (Geoscience Australia); P. Dailey (AIR Worldwide)
Tropical cyclones are the most common disaster such as the state of Florida, a modest 1 percent
in the Pacific, and among the most destructive. In increase in wind speeds can result in a 5 percent to
December 2012, Cyclone Evan caused over US$200 10 percent increase in loss to residential property.
million in damage in Samoa, nearly 30 percent Quantifying the change impact thus supports
of Samoan GDP. Niue suffered losses of US$85 evidence-based decision making on adaptation to
million following Cyclone Heta in 2004—over future climate risk.
five times its GDP. As recently as January 2014,
The quantitative, locally specific information needed
Cyclone Ian caused significant damage throughout
to guide adaptation decisions at the national or
Tonga, resulting in the first payout of the Pacific
community level can best be generated by adopting
Catastrophe Risk Insurance Pilot system operated
a multidisciplinary approach. Climate model
by the World Bank (see section 3-19 above for more
simulations alone are insufficient, since they deal
information).86
with extreme events that are by their nature rare
According to the Intergovernmental Panel on and unlikely to be generated in a limited set of
Climate Change (IPCC), intense tropical cyclone general circulation model (GCM) runs. Moreover,
activity in the Pacific basin will likely increase features having the greatest impact are highly
in the future (IPCC 2013). But such general localized and hence impossible to resolve in a global
statements about global tropical cyclone activity model. The analysis described here joined climate
provide little guidance on how impacts may change GCMs forced by emission scenarios to catastrophe
locally or even regionally, and thus do little to help modelling methods—a hybrid approach that drew
communities and nations prepare appropriate on the respective strengths of climate science and
adaptation measures. risk management.
The study described here87 assesses climate change Using catastrophe models, it is possible to estimate
in terms of impact on the human population and the financial impacts caused by tropical cyclones
its assets, expressed in terms of financial loss. An at a local scale. Catastrophic risk models do not
impact focus is relevant to adaptation because have the computational overhead of a GCM, and
changes in hazard do not necessarily result in a so can be run in a probabilistic framework using a
proportional change in impact. This is because catalog of events (built from statistics about past
impacts are driven by exposure and vulnerability cyclones, including intensity, frequency, and tracks)
as well as by hazard. For example, a small shift in that represents the likely actual distribution of
hazard in a densely populated area may have more loss-causing cyclones. By analyzing the projections
significant consequences than a bigger change in an from GCMs, it is possible to determine how the
unpopulated area. Analogously, a dense population distributions of loss-causing cyclones may change;
that has a low vulnerability to a particular hazard and by adjusting the catastrophe model’s hazard
might not need to adapt significantly to a change in catalog to be consistent with the GCM projections,
hazard. Even in regions with high tropical cyclone it is possible in turn to produce objective projections
risk and correspondingly stringent building codes, of hazard, damage, and loss.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
185
186
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
The project described here analyzed current and Adaptation Planning (PACCSAP) program.89
and future cyclone hazard and risk for 15 Pacific Figure 3-31 shows sample track data from GCMs and
Island countries involved with PCRAFI (whose the comparison to historical tropical cyclones.
aims are described in section 3-19). It combined
The analysis focused on the RCP8.5 scenario (the
data produced through PCRAFI with information
on tropical cyclone activity in the Pacific region most extreme Representative Concentration
extracted from model runs produced for the IPCC Pathway, or RCP, projection), under which annual
Fifth Assessment Report. mean global temperature anomalies reach +4°C by
2100 (IPCC 2013). However, the approach described
Approach. Over 20 modelling groups have
/// ///
here is applicable to any scenario where climate
conducted modelling experiments that contribute model data are available. Two time periods were
toward the fifth phase of the Coupled Model analyzed: 1981–2000, representing current climate
Intercomparison Project (CMIP5), based on conditions, and 2081–2100, representing future
the latest emission scenarios used in the Fifth climate conditions under this scenario.
Assessment Report of the IPCC (Taylor, Stouffer,
and Meehl 2012). With the goal of identifying and The climate-conditioned catalogs were validated
tracking tropical-cyclone-like vortices (TCLVs), 88 by a cross-discipline group of scientists within and
five models from the CMIP5 collection were outside the project teams at Geoscience Australia
analyzed by Australia’s Commonwealth Scientific and AIR Worldwide. Statistical and physical checks
and Industrial Research Organisation (CSIRO) as assured that the distribution of storm track,
part of the Pacific-Australia Climate Change Science intensity, evolution, wind speed, storm surge, and
Table 03—12 CURRENT FUTURE RELATIVE
Changes in Key FIELD DOMAIN CHANGE
CLIMATE CLIMATE CHANGE (%)
Tropical Cyclone– Annual frequency NH 16.1 17.9 1.81 11.2
related Parameters (tropical cyclones/
SH 11.6 11.3 -0.34 -2.9
for the Five- year)
member Ensemble NH 14 13.4 -0.64 -4.6
Genesis latitude (°N)
SH -13.8 -13.2 0.53 -3.9
Source: Arthur and
NH 159.7 170.4 10.77 6.7
Woolf 2013. Genesis longitude (°E)
SH 157.3 160.4 3.12 2
Note: Bold, italicized values Mean latitude of NH 18.5 18.1 -0.37 -2
indicate that change in the maximum sustained
SH -18.6 -19 -0.34 1.8
ensemble mean is greater wind (°N)
than the inter-model Mean latitude of NH 18.9 18.7 -0.18 -0.9
standard deviation. NH = minimum pressure (°N) SH -19 -19.1 -0.14 0.7
Northern Hemisphere; SH =
Mean minimum central NH 963.2 965.7 2.46 0.3
Southern Hemisphere.
pressure (hPa) SH 968.5 969.5 0.98 0.1
Mean maximum NH 41.2 39.4 -1.8 -4.4
sustained wind (m/s) SH 38.5 37.4 -1.1 -2.9
CMIP5 - Northern Hemisphere CMIP5 - Southern Hemisphere
0.25 0.25
Figure 03—32
Ensemble mean
0.20 0.20
proportion of
cyclones for current
PROPORTION
0.15 0.15
and future climate
in the Northern
0.10 0.10
Hemisphere (left) and
Southern Hemisphere
0.05 0.05
(right).
0.00 0.00 Source: Arthur and
TD TS TC1 TC2 TC3 TC4 TC5 TD TS TC1 TC2 TC3 TC4 TC5
Woolf 2013.
1981 - 2000 2050 2081-2100
Note: Classification is based
on the Saffir-Simpson
hurricane wind scale. Values
for 2050 were determined
using a linear interpolation
other dynamical parameters properly correlated cyclone category 5). Table 3-13 shows that mean
between the midpoint of the
in space and time with the changes informed by maximum sustained winds will decrease in both 1981–2000 and 2081–2100
the climate model projections. The experimental hemispheres, but as the changes in wind speeds at periods. TD = tropical
framework was designed to incorporate peer both ends of the distribution largely balance, the depression; TS = tropical
review at all stages of the project and to include mean intensity does not change significantly in storm; TC = tropical cyclone.
vetting of the results. This approach has been used either hemisphere.
successfully to model hazard and loss for future
The interaction of changes in frequency and
climate conditions in other studies, such as Dailey et
intensity distributions brings about nonlinear
al. (2003) and Arthur and Woolf (2014).
changes in the corresponding hazard levels. For
Results. Table 3-13 presents the change in cyclone
/// /// example, it is possible that a reduction in frequency,
hazard for the five-model ensemble mean. The coupled with an increase in the share of intense
matrix contains current, future, change, and relative tropical cyclones, could increase the probability
change values for seven parameters that inform that the most extreme winds would occur—and as a
the resampling of the 10,000-year synthetic event result, increase the likelihood of experiencing larger
catalog. Of all the parameters, only one—genesis losses. Return period losses for current conditions
longitude in the Northern Hemisphere domain— and for the five future scenarios over the whole
shows a significant change—that is, for only this Pacific region show that for two scenarios, losses
parameter is the ensemble mean change greater than will significantly increase (figure 3-33). However,
the inter-model standard deviation. local losses may differ from the regional trends.
Figure 3-32 shows the changes in tropical cyclone The 250-year return period losses are presented
intensity distribution between current and future in figure 3-34, based on the ensemble mean for the
time periods for the mean of all climate models. current climate. Across the entire Pacific region,
There is a shift in the distribution, with fewer a 250-year return period loss is around 9 percent
midrange events (tropical cyclone categories 1–4), of GDP. However, examining individual countries
more weak events (tropical depressions and tropical produces a wide range of results. The 250-year loss
storms) and more very intense events (tropical is nearly 280 percent of GDP for Niue, is 99 percent
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
187
188
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Regional End-of-Century CMP5 Agg GroundUp EP Curves
Figure 03—33
Individual regional 2000
CURRENT CLIMATE
end-of-century 1800 FUTURE SCENARIOS
exceedance
probability curves for 1600
ensemble members 1400
LOSS (MILLION US$)
(blue) compared
to the current 1200
climate exceedance 1000
probability curve
(green). 800
Source: Air Worldwide. 600
Note: Each curve represents 400
the loss across all asset
200
types arising from tropical
cyclone impacts. 0
0 25 50 75 100 125 150 175 200 225 250
RETURN PERIOD (YRS)
of GDP for the Federated States of Micronesia, and the changes in loss be considered significant under
is 79 percent of GDP for the Marshall Islands. this metric. This result suggests the spectrum of
changes in tropical cyclone activity that can be
Figure 3-35 shows that 250 year return period losses
drawn from the climate model projections.
increase in most countries under future climate
conditions; however their significance depends on Discussion. The change in wind risk in the future
/// ///
the GDP. The biggest increases are seen in Vanuatu modelled climate is neither simple nor uniform
(11 percent), Niue (29 percent); and Samoa (35 across the region. Determining appropriate
percent); there is a decrease in Nauru and Kiribati. adaptation measures requires quantitative
The changes in tropical cyclone intensity or information beyond generic “up or down”
frequency are not nearly as large as these changes statements. Changing intensity and frequency can
in loss. The nonlinear nature of the vulnerability balance out in a complex interaction. This means
models leads to major increases in loss levels for the average peak intensity may remain constant
only minor increases in the hazard level. or decline, while long return period wind speeds
increase due to a rise in the relative proportion of
However, of all the projected changes in loss, only
very intense tropical cyclones.
the change in 250-year return period loss for Samoa
(total losses) could be considered statistically The analysis here has focused on regional (basin-
significant. The mean change in loss across the five wide) changes in key tropical cyclone parameters.
models exceeds the standard deviation of those However, tropical cyclone–related risk depends
changes for this location. For no other country can on changes in tropical cyclone activity at the
Figure 03—34
Ensemble mean
250-year losses
across the Pacific as
a proportion of Pacific
Island countries’ GDP
for current climate
conditions (1981–
2000).
Source: Geoscience
Australia.
Figure 03—35
Ensemble mean
change in 250-year
return period loss.
Source: Geoscience
Australia.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
189
190
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
country scale, and on actions taken at the national significant. Given that over 20 modelling groups
and community level. It is highly likely that some conducted RCP8.5 experiments, using an ensemble
countries will experience changes in tropical of only five may in itself lead to skewed results.
cyclones that are at odds with the basin-wide Careful selection of the members, based on
changes. Adaptation options need to recognize quantitative measures of performance in the region,
the localized nature of the changing hazard and would minimize the risk of biased results. More-
risk, and be tailored to suit the local capacity for reliable results are more likely to be accepted, and
implementing possible options.
hence more likely to prompt action.
The results of this study demonstrate that assessing
Assessing results from multiple climate models
the impact of climate change on hazard alone is
also encourages stakeholders to consider a range
not sufficient. The large increase in risk in many
of potential outcomes for which they could
regions, compared to the relatively small changes in
prepare adaptation options. While the ensemble
hazard, highlights the significance of exposure and
mean can provide greater confidence than any
vulnerability. The nonlinear nature of vulnerability
individual model result, using a worst-case result
means losses can increase dramatically as a
result of only small changes in hazard. This is an that provides an upper limit of the potential
important finding because it suggests that the most impacts may be desirable in some applications. This
effective way to reduce financial risk is to reduce conservative approach would be appropriate, for
vulnerability. At the country scale, little can be done example, for standards for building design, given
to minimize changes in hazard, and exposure to the expected lifetime of built assets, especially large
tropical cyclones is likely to continue to increase infrastructure (e.g., hospitals or port facilities). For
as populations grow. By improving the resilience of longer planning timelines, the expense and time
exposed assets (reducing vulnerability), risks can needed to modify the asset as projections of risk
be significantly lowered. Some examples include change make it harder to change adaptation options.
preemptive vegetation reduction to minimize At shorter timelines (e.g., annual crop planting), risk
chance of tree crops suffering damage in a tropical reduction options can be more readily evaluated,
cyclone, improved site selection for vulnerable crops making a mean estimate of risk more suitable for
and other land-use planning measures, or changes
consideration.
in and/or more stringent enforcement of local
building standards. Finally, it should be noted that this study did not
consider projections of future exposure. It is widely
Using an ensemble of climate models for this work
acknowledged that increased exposure has been the
makes it possible to understand the robustness
most significant driver of increased disaster losses
of the projected changes. Analyzing loss changes
over the past decades (Barthel and Neumayer 2012).
derived from a single climate model could be
Thus future studies of the kind described here would
misleading if it were an outlier compared to the
ensemble. A consistent trend across several models benefit from considering exposure projections,
would give end-users much greater confidence in although the complex nature of exposure modelling
the robustness of the results, even if the mean result is likely to add significantly to the uncertainty in the
is not statistically significant. As it is, our analysis results. For policy makers, decisions about climate
found several models with statistically significant change adaptation (particularly decisions related to
changes in tropical cyclone frequency, while assets with a long lifetime) may need to be made in
the ensemble mean change was not statistically the absence of unambiguous evidence.
CHAPTER
03
3-25. A Framework for Modelling Future Urban Disaster
Risk90
David Lallemant, Steven Wong, Anne Kiremidjian (Stanford University)
This case study proposes a framework to understand Intense competition for land in urban environments,
and model the drivers of new risk creation, with a driven mostly by accessibility to livelihood, means
particular focus on dynamic urban environments. that hazardous areas such as floodplains and steep
Such a framework will help policy makers to slopes will be settled.
understand and predict risk as it relates to dynamic
changes in urban environments—such as increases Cities shift the economic balance of risk mitigation,
in population, specific urban growth patterns over since expected losses are so high (Lall and
an evolving multi-hazard landscape, and evolving Deichmann 2012; World Bank 2010b). This suggests
vulnerability—and in turn help them promote a great opportunity for city officials and policy
resilient and sustainable future cities. makers to implement risk mitigation policies and
projects. Because cities are growing, officials also
By 2030, the global population will reach 9
have a unique chance to affect the distribution and
billion, of which 60 percent will reside in cities
quality of future constructions, so that all new city
(United Nations 2007). To put these numbers
growth is resilient.
in perspective, twice as many people will live in
cities in 2030 as there were total people living in To capitalize on these opportunities, policy
1970. This population shift has made cities the makers need urban risk assessment models that
major source of global risk, in large part because take projections of future risk into account.
of the increase in exposure linked to increases in Current probabilistic risk assessment models use
population in hazard-prone areas (Bilham 2009).
static—current—conditions for hazard, exposure,
Cities often emerge in locations with favorable
and vulnerability. They therefore have the effect
economic conditions (coastal zones, river crossings,
of underestimating risk, and they also constrict
fertile volcanic soils, valleys), but these often
policy makers to a hopeless catch-up mode: since
correlate with increased hazard probability (floods,
conditions are always evolving past the latest data,
hurricanes, volcanoes, earthquakes). Furthermore,
their scope of action is limited to mitigating risk
since urbanization typically has occurred during a
to existing assets, rather than proactively seeking
time frame that is very short as compared to the
to reduce future risk. The model proposed here, by
return period of damaging natural hazards, there has
been little learning from past disasters, and hazards contrast, is a dynamic urban risk analysis framework
that in the past affected villages and towns will now that accounts for time-dependent changes in
be affecting large urban agglomerations. exposure and vulnerability in order to project risk
into the future.
Evidence suggests that the risk linked with such
increases in exposure at the macro scale (increases By focusing on modelling future risk, the framework
in population in hazard-prone areas) is significantly enables the investigation of risk consequences
exacerbated by trends in distribution of this new of various policy and planning decisions. It can
urban population within the urban boundary. therefore readily inform risk-sensitive urban and
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
191
192
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
regional policy and planning to promote resilient in its present state. This approach is a significant
communities worldwide. limitation for assessing risk in rapidly changing
environments, in particular cities. The proposed
Dynamic urban risk framework. Probabilistic
approach builds on current practices by integrating
/// ///
disaster risk assessment consists of taking
urban growth models to forecast exposure. The
the convolution of the hazard, exposure, and
resulting risk assessment is more accurate and
vulnerability. Hazard refers to the potential
enables policy makers to take preventive measures
occurrence of an event that may have adverse
to reduce future risk.
impacts on vulnerable and exposed elements
(people, infrastructure, the environment, etc.). The simplest method for modelling future exposure
Exposure describes the elements that are impacted is to project exposure trends based on past data.
by the hazard due to their spatial and temporal Census data for population or building inventory
overlap. Vulnerability describes the propensity to at a minimum of two separate dates can be used
Figure 03—36 suffer adverse effects from exposure to particular to develop projections for the future. Auxiliary
The three hazard intensity. These definitions make clear that data—such as general migration rate, natural
components of the fundamental components of risk are not fixed population growth, and economic growth—can
risk and their in time, particularly in rapidly changing urban further be used to improve these projections.
time dependence. environments (see figure 3-36). Alternatively, agent-based models can be developed
and calibrated to simulate patterns of urban growth,
Source: Lallemant et al. Dynamic exposure modelling. Current risk
creating numerous alternatives of future urban form
/// ///
2014. assessment methodologies characterize exposure
(Batty 2007).
Risk Hazard Exposure Vulnerability
Time-dependent hazards Time-dependent exposure Time-dependent vulnerability
• Large earthquakes do not occur • Population growth and migration • The vulnerability of buildings
following a Poission process. are rapidly changing the global changes in time due to
The occurrence of an event is landscape of risk exposure. deteriorations, retrofits, and
dependent on the time since the • Cities in particular are sites of alterations.
last event, consistent with elastic very rapid exposure change, • Most buildings in the world’s
rebound thoery of earthquakes. often reflecting significant growing cities are note static,
• Similarly, hydrometeorological migration into and within cities. but are continually being
hazards (e.g. floods and •Urban land markets often expanding upward or outward.
hurricanes) have recently received create pressures to settle on These practices have significant
a lot attention as research is increasingly hazardous land, impact on building vulnerability.
predicting increasing rate and includding steep slope, flood- • New construction practices
intensity of extreme weather planes and reclaimed land. further result in changes
events. in vulnerability of the built
environment.
Figure 03—37
Incrementally
expanding buildings
and corresponding
changes
in vulnerability.
Regional End-of-Century CMP5 Agg GroundUp EP Curves Note: The top panel
illustrates incremental
Building
of Collapse
1.00
1.00
building construction typical
Stage
PROBABILITY OF COLLAPSE
of cities throughout the
world; the bottom panel
0.75
0.75 Base
BUILDING Case
STAGE
illustrates the increase in
Base Case
vulnerability in hypothetical
0.50
0.50 Expansion
Expansion 1 1 fragility curves as floors are
Probability
Expansion 2 added and discontinuous
0.25
0.25 Expansion
Expansion 3 2 expansions occur.
0.00
0.00 Expansion 3
0
0 1
1 2
2 3
3 4
4
Peak GROUND Acceleration
Ground
PEAK (PGA)
ACCELERATION (PGA)
Indeed, the informal sector builds an estimated
Dynamic vulnerability modelling. Current
70 percent of all urban housing in developing
/// ///
risk assessment models implicitly assume that
countries (Goethert 2010). This process starts
vulnerability is constant over time. Increase in
with a simple shelter and, given enough resources
vulnerability of structures with deterioration has
and time, transforms incrementally to multi-story
been the subject of increasing study (Frangopol,
homes and rental units. However, no robust studies
Lin, and Estes 1997; Ghosh et al. 2013; Rokneddin
have investigated the effect of these incremental
et al. 2013). Recent work by Anirudh Rao provides
expansions on vulnerability, particularly to
a time-dependent framework for modelling
seismic hazards.
structural deterioration of individual bridges
and their resulting increased seismic risk.91 The Using seismic risk as a case study, the proposed
framework proposed here builds on this research to framework defines typical stages within building
incorporate time-dependent fragility into large- evolution, along with associated earthquake fragility
scale risk assessment models, and looks at other curves reflecting the changes in vulnerability
common drivers influencing fragility. In particular induced by each building expansion (see figure 3-37).
it investigates incremental construction as a Earthquake fragility curves describe the probability
significant cause of changes in vulnerability, and also of experiencing or exceeding a particular level of
looks at the role of changing building practices and damage when subjected to a specific ground motion
structural deterioration. intensity, usually measured in terms of peak ground
motion acceleration or spectral acceleration.
In rapidly urbanizing areas, the pay-as-you-go
Alternatively, instead of linking building expansions
process of informal building construction and
to new fragility curves, these increments can be
expansion is the de facto pattern of growth.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
193
194
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
treated as additional vulnerability indicators in This simplified application of the framework uses
multivariate fragility models. very limited data and simple models. The results
themselves are therefore not aimed at accuracy
Simplified case study of Kathmandu, Nepal.
of risk forecasting but are simply intended to
/// ///
The framework described above was applied in
demonstrate the importance of including urban
order to forecast the earthquake risk of Kathmandu,
dynamics in risk assessment of cities. A discussion
Nepal. Since the main interest is to capture
is included explaining how the model could be made
changing risk driven by time-dependent exposure
more complex to better reflect the uncertainties and
and vulnerability, the study describes the risk at
real urban dynamics.
different time periods based on a single earthquake
scenario: a reproduction of the 8.1 magnitude Bihar The seismic hazard was developed by simulating
earthquake of 1934. 2,500 equally likely scenarios of the 1934 Bihar
Figure 03—38
Number of buildings
sustaining heavy
damage or collapse
from a single ground
motion field, at six
different time periods.
Note: PGA = peak ground
acceleration.
CHAPTER
03
µ=48,200 1991 Figure 03—39
2001 Full distribution of the
µ=76,200 2011 number of buildings
sustaining heavy
2015
µ=127,500 damage or collapse,
2020
for six different
FREQUENCY
µ=154,600 2025 time frames.
µ=193,600 Source: Lallemant et al.
2014.
µ=238,400
0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000
NUMBER OF BUILDINGS SUSTAINING HEAVY DAMAGE OR COLLAPSE
1934 BIHAR EARTHQUAKE SCENARIO
earthquake (spatially correlated ground motion wards, as well as the distribution/redistribution
intensity fields) using the OpenQuake analysis of vulnerable building types. However, the values
engine (GEM 2013).92 Six exposure models were predicted are an example from a single ground
used, corresponding to years 1991, 2001, 2011 (from motion simulation (shown in bottom left of figure
the ward-level census), and 2015, 2020, and 2025
3-38), and very different results would be generated
(projected based on quadratic fit to past census
from a different simulation.
data). Vulnerability curves used are those derived
from Arya (2000), who has developed many The east side of the city sustains heavier damage in
vulnerability curves for typical buildings in the area. large part as a result of higher ground motions from
For simplicity, rates and distribution of “heavy this specific simulation. In order to characterize the
damage or collapse” are used as metrics to measure full distribution of heavy building damage for the
time-varying risk. Figure 3-38 shows the distribution entire Kathmandu municipality, the process above
of the number of heavily damaged or collapsed is repeated for every ground motion field simulation
buildings for each of the exposure models, based on (n = 2,500). The total number of heavily damaged
a single ground motion field simulation. or collapsed buildings is computed for every ground
The results clearly show significant changes in motion field simulation. We can then compute the
risk driven by urban growth patterns and changes expected (mean) risk due to changing exposure and
in primary construction type. The changing risk vulnerability, as well as the full empirical probability
reflects both the high growth rates of specific distribution of damage.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
195
196
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Figure 03—40
Number of buildings sustaining heavy damage
Expected number of
NUMBER OF BUILDINGS SUSTAINING HEAVY DAMAGE
buildings sustaining 400,000
400,000
heavy damage
or collapse as a
function of time, with 300,000
300,000
confidence interval.
Source: Lallemant et al.
2014. 200,000
200,000
100,000
100,000
00
1990
1990 1995
1995 2000
2000 2005
2005 2010
2010 2015
2015 2020
2020 2025
2025 2030
2030
Year
YEAR
The results shown in figures 3-39 and 3-40 to existing buildings is a ubiquitous practice and is
demonstrate that changes in exposure and not accompanied with proper seismic strengthening.
vulnerability in Kathmandu drive a significant Conversely, models could be developed reflecting
increase in risk. The expected number of buildings potential vulnerability reduction policies, such as
sustaining heavy damage or collapsing (mean improvements in construction practices, building
values shown in figure 3-39) nearly doubles every height restrictions, or risk-sensitive zoning, among
10 years. Furthermore, the spread of the probability
others. Finally, the effects of urban dynamics on
distribution of damage also increases. This increase
exposure to secondary seismic hazards, in particular
is most likely the result of increased concentration
liquefaction and landslides, could also be modelled.
of exposure.
The proposed framework for assessing risk as it
Given additional data, this preliminary study of
changes in time includes dynamic exposure and
Kathmandu could be extended to more accurately
vulnerability models in order to forecast future
capture the urban dynamics. Instead of using the
losses. The basic framework can be applied for
constant compound growth model over entire
wards, different population growth patterns various levels of data availability and resolution.
could be explored. In addition, the model could By focusing on modelling future risk, the
directly incorporate changing vulnerability due framework enables the further investigation of risk
to incremental construction. The failure to do so consequences from various policy and planning
tends to underestimate damage, since incremental decisions. It therefore can readily serve to inform
construction typically leads to increased risk-sensitive urban and regional policy and planning
vulnerability. In Kathmandu, the addition of floors to promote resilient communities.
CHAPTER
03
198
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Endnotes the Western Visayas region of the Philippines (see Bautista
et al. 2012).
21 See the institutions’ websites at www.codeforresilience.
31 For more information, see Geoscience Australia, “New
org and http://www.rhok.org/.
Building Assessment Tool Supports Better Risk Analysis,”
22 Parts of this paper also appear in Soden, Budhathoki,
February 12, 2014, http://www.ga.gov.au/about-us/
and Palen (2014). news-media/news-2014/new-building-assessment-tool-
supports-better-risk-analysis.html.
23 In the 20th century alone, over 11,000 people lost
their lives to earthquakes in Nepal. The 1934 Bihar-Nepal 32 This paper draws in part on Petiteville, Bally, and Seguin
earthquake destroyed 20 percent of Kathmandu’s building (2012).
stock and damaged 40 percent. Geohazards International,
33 Institutions include the Arab Centre for the Studies
“Kathnandu Valley Earthquake Risk Management,” http://
of Arid Zones and Drylands, Beijing Normal University,
geohaz.org/projects/kathmandu.html.
Centro Internazionale in Monitoraggio Ambientale (CIMA)
24 Funding for this research is provided by the Global
Foundation, Geoscience Australia, Global Volcano Model,
Facility for Disaster Reduction and Recovery. For more Joint Research Centre of the European Commission,
information on VGI, see http://crowdgov.wordpress.com/. Kokusai Kogyo, the Norwegian Geotechnical Institute,
International Centre for Numerical Methods in Engineering
25 This opinion is attributed to Schulyer Erle in Tim Waters,
(CIMNE), University of Geneva, Famine Early Warning
“OpenStreetMap Project & Haiti Earthquake Case Study, Systems Network (FEWS-NET), Global Earthquake Model
slide presentation, 2010, http://pt.slideshare.net/chippy/ Foundation, the United Nations Environment Programme–
openstreetmap-case-study-haiti-crisis-response. Global Resource Information Database (UNEP-GRID), and
the World Agency for Planetary Monitoring and Earthquake
26 Tim Waters, “OpenStreetMap Project & Haiti Earthquake
Risk Reduction (WAPMEER).
Case Study, slide presentation, 2010, http://pt.slideshare.
net/chippy/openstreetmap-case-study-haiti-crisis- 34 The full technical description of the approach can be
response.
found in Herold and Rudari (2013).
27 This point was made by Jeffrey Johnson, Where 2.0
35 Two publications are planned under this effort:
conference, March 30–April 1, 2010, San Jose, CA, http://
“Probabilistic Volcanic Ash Hazard Analysis (PVAHA) I:
hot.openstreetmap.org/updates/2013-12-17_imagery_for_
Adapting a Seismologically Based Technique for Regional
haiyan.
Scale Volcanic Ash Hazard Assessment,” by A. N. Bear-
Crozier and colleagues; and “Probabilistic Volcanic Ash
28 Geoscience Australia holds a Creative Commons
Hazard Assessment (PVAHA) II: Asia-Pacific Modelling
Attribution 3.0 Australia license for the material in this
section. All terms of the license apply for reuse of text and Results,” by Victoria Miller and colleagues.
graphics. Jones, Van Putten, and Jakab publish with the
36 The International Centre for Water Hazard and Risk
permission of the CEO, Geoscience Australia.
Management (ICHARM) operates under the auspices
29 GMMA RAP was a component of BRACE (Building the
of UNESCO and the Public Works Research Institute,
Resilience and Awareness of Metro Manila Communities Japan. The authors would like to express their sincere
to Natural Disaster and Climate Change Impacts), an appreciation to the following for their valuable inputs:
Australian aid program in the Philippines initiated in 2010 Dr. Satoru Nishikawa (special representative of the
that sought to reduce the vulnerability and enhance the Secretary-General for DRR on the Post-2015 Framework
resilience of Metro Manila and selected neighboring areas for DRR and the Global Platform); Mr. Yusuke Amano
to the impacts of natural disasters and climate change. As (Water and Disaster Management Bureau, Japan); and
a component of the BRACE program, GMMA RAP is also Dr. Yuki Matsuoka (UNISDR Hyogo Office). We are also
known as the Enhancing Risk Analysis Capacities for Flood, indebted to the Philippine Atmospheric, Geophysical
Tropical Cyclone Severe Wind, and Earthquake for Greater and Astronomical Services Administration and the Asian
Metro Manila Area program. Disaster Preparedness Center for providing their data
and comments.
30 This project was the joint Geoscience Australia/
PHIVOLCS (Philippine Institute of Volcanology and 37 For the sake of brevity, the discussion here will focus on
Seismology) pilot study of earthquake risk in Iloilo City, in the risk of fatality-causing floods.
CHAPTER
03
38 Such a conceptual approach uses hazard, exposure, and
47 This module, called CRISIS, was developed at the
vulnerability indices to assign data to various categories. Engineering Institute of the National University of Mexico
For each category, a score is derived by arithmetic by M. Ordaz, A. Aguilar, and J. Arboleda.
computations, such as by using the weighted rank sum
method. A conceptual risk index is finally presented on a 0 48 World Bank, Asian Development Bank, UN System,
to 1 scale by summing the scores. “Tsunami: Impact and Recovery, Joint Needs Assessment,
2005.”
39 Only the effectiveness of the levee with respect
to overflow is considered. Breaching of levees is not 49 The outputs of this phase included a synthesis report,
considered in this analysis. This may underestimate the a report on methodologies, 10 detailed island reports,
calculated inundation extend and the water depths of the and a technical specification report on databases. All
are accessible at the Maldives Department of National
flood when levees are included in the calculation.
Planning website, http://planning.gov.mv/en/content/
40 Geoscience Australia holds a Creative Commons
view/306/93/.
Attribution 3.0 Australia license for the material in this
50 This phase produced social and economic vulnerability
section. All terms of the license apply for reuse of text and
assessment reports as follows: a synthesis report, a
graphics. Jones, Griffin, Robinson, and Cummins publish
methodological description, and 10 detailed island reports.
with the permission of the CEO, Geoscience Australia.
All are accessible at the Maldives Department of National
The authors gratefully acknowledge Guy Janssen, whose Planning website, http://planning.gov.mv/en/content/
independent review of the Indonesian Earthquake Hazard view/306/93/.
Project identified and articulated many of the factors for
51 The cost-benefit report is accessible at http://www.
success discussed in this paper.
preventionweb.net/english/professional/publications/v.
41 For more information, see http://aid.dfat.gov.au/
php?id=14437.
countries/eastasia/indonesia/ and http://www.aifdr.org/.
52 These data were from the National Statistics Office 2008
The AIFDR is managed by Australian and Indonesian co- population and housing census.
directors, and AIFDR work programs and funding decisions
53 Note that for the purposes of analysis flood defenses
are jointly developed by the Australian Department of
were assumed to be not effective due to insufficient
Foreign Affairs and Trade (DFAT) and Badan Nasional
maintenance.
Penanggulangan Bencana (BNPB; Indonesian National
Agency for Disaster Management)
54 Material in this section is based on the World Bank–
commissioned Shire Integrated Flood Risk Management
42 The agencies are the BNPB; Badan Geologi (Geological
Program Final Report: Volume 1; the report was completed
Agency of Indonesia); Badan Meteorologi, Klimatologi, dan
in 2012 by Atkins.
Geofisika (Indonesian Agency for Meteorology, Climatology
and Geophysics); Lembaga Ilmu Pengetahuan Indonesia
55 For this determination, the 1-in-100-year scenario with
(Indonesian Institute of Sciences); and Institut Teknologi
climate change was used.
Bandung (Bandung Institute of Technology).
56 The figure is as of March 1, 2014.
43 GMMA RAP is also known as the Enhancing Risk Analysis
Capacities for Flood, Tropical Cyclone Severe Wind, and 57 This involved data from 1,076 existing boreholes and 48
Earthquake for Greater Metro Manila Area program. new drillings undertaken under the project.
44 Geoscience Australia, “International Work Helps Build
58 See Pektas and Gulkan (2004).
Safer Communities in the Philippines,” http://www.ga.gov.
au/about-us/news-media/news-2014/international-work- 59 ISMEP is a €1.5 billion project running from 2006 to
helps-build-safer-communities-in-the-philippines.html. 2018. It is funded by the World Bank, European Investment
Bank, European Council Development Bank, and Islamic
45 The event occurred 95km south of Aqaba.
Development Bank.
46 The software development and the risk assessment
60 ISMEP Magazine, May 2012, http://issuu.com/
exercises were undertaken by ERN-AL consortium. guvenliyasam/docs/ismep_dergi_en5/9?e=0/6534273.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
199
200
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
61 This assumed the same level of seismicity across
70 See Scolobig et al. (2013).
the country.
71 Geoscience Australia holds a Creative Commons
62 This section is drawn from the World Bank report
Attribution 3.0 Australia license for the material in this
entitled Building Morocco’s Risk Resilience: Inputs into an section. All terms of the license apply for reuse of text
Integrated Risk Management Strategy (Washington, DC: and graphics.
World Bank, 2013), which summarizes technical assistance
work performed by the World Bank and funded by GFDRR 72 Data are from the Data dan Informasi Bencana Indonesia
and the Swiss Development Cooperation in the period (Disaster Data and Information Indonesia) database, BNPB
2008–2013. (Indonesian National Disaster Management Agency), 2009,
http://dibi.bnpb.go.id.
63 EM-DAT: The OFDA/CRED International Disaster
Database, www.emdat.be, Université Catholique de 73 The evaluation’s theoretical framework was the
Louvain, Brussels (Belgium). Transtheoretical Model for Behavior Change (Prochaska,
Norcross, and DiClemente 1994). The five steps identified in
64 Abdelhamid Ben Abdelfadel and Fatima Driouech,
the framework are resistance, contemplation, preparation,
“Climate Change and Its Impacts on Water Resources in action, and maintenance.
the Maghreb Region,” http://www.arabwatercouncil.org/
administrator/Modules/Events/IWRA%20Morocco%20 74 Geoscience Australia holds a Creative Commons
Paper.pdf. Attribution 3.0 Australia license for the material in this
section. All terms of the license apply for reuse of text
65 The MnhPRA technical contractor was RMSI Ltd.
and graphics.
66 This account of PCRAFI is based on World Bank project
75 Contribution from Dr. Agus Wibowo, Head of Data
documents, including World Bank, “Pacific Catastrophic Division, Center for Data, Information and Public Relations,
Risk Assessment and Financing Initiative: Better Risk
Badan Nasional Penanggulangan Bencana (BNPB).
Information for Smarter Investments—Catastrophic Risk
Assessment Methodology,” Washington, DC, 2013, https:// 76 Contribution from Geoscience Australia staff members
www.gfdrr.org/sites/gfdrr.org/files/publication/PCRAFI_ Kristy van Putten, Charlotte Morgan, and David Robinson.
Catastrophe_Risk_Assessment_Methodology.pdf.
77 The Australian government agencies involved in
67 Timor-Leste is technically not in the Pacific but was
developing InaSAFE include the Department of Foreign
included in the PCRAFI program. Affairs and Trade–Development Corporation and
Geoscience Australia through the Australia-Indonesia
68 World Bank, “Tonga to Receive US$1.27 Million Payout for
Facility for Disaster Reduction. The World Bank’s
Cyclone Response,” press release, http://www.worldbank.
participation was supported by AusAid’s East Asia and
org/en/news/press-release/2014/01/23/tonga-to-receive-
Pacific Infrastructure for Growth Trust Fund.
payout-for-cyclone-response.
78 The “rookies” were chosen by Black Duck, a software
69 This case study presents the results of interdisciplinary
and consulting company. See Klint Finley, “Microsoft, Yahoo
research undertaken within the framework of the MATRIX
Among Open Source ‘Rookies of the Year,’” Wired, http://
(New Multi-HAzard and MulTi-RIsK Assessment MethodS
www.wired.com/wiredenterprise/2013/01/open-source-
for Europe) project. The research was supported by the
rookies-of-year/.
European Community’s Seventh Framework Programme
through the grant to the budget of the MATRIX project
79 For more information, see the project website at http://
(New methodologies for multihazard and multi-risk
noah.dost.gov.ph/.
assessment methods for Europe [FP7/2007-2013]) under
grant agreement no. 265138. The paper reflects the 80 Kundzewicz et al. (2014), based on MunichRe’s
authors’ views and not those of the European Community. NatCatSERVICE data.
Neither the European Community nor any member of the
MATRIX Consortium is liable for any use of the information 81 For more details, see Pappenberger et al. (2012);
in this paper. We wish to thank all who offered professional Jongman, Ward, and Aerts (2012); Dilley et al. (2005a);
advice and collaboration. We are especially grateful to UNISDR (2009b); Hirabayashi et al. (2013); Ward et al.
the practitioners who discussed with us the challenges of (2013b); Winsemius et al. (2013); Arnell and Lloyd-Hughes
multi-risk assessment. (2014).
CHAPTER
03
82 See the Global Flood Working Group portal at http://
portal.gdacs.org/Expert-working-groups/Global-Flood-
References
Working-Group.
Abbs, D. 2012. “The Impact of Climate Change on the
Climatology of Tropical Cyclones in the Australian
83 EM-DAT: The OFDA/CRED International Disaster
Region.” CSIRO Climate Adaptation Flagship Working
Database, www.emdat.be, Université catholique de
Paper No. 11, CSIRO, Canberra.
Louvain, Brussels, Belgium.
Abbs, D. J., S. Ayral, E. Campbell, J. L. McGregor, K. C.
84 The quotation is from D. Wielinga, senior disaster risk
Nguyen, M. Palmer, A. S. Rafter, I. G. Watterson, and
management specialist, World Bank Africa Region; see
GFDRR, “GFDRR Connects Science with Policy to Help B. C. Bates. 2006. Projections of Extreme Rainfall and
Address Flood Risk in Nigeria,” https://www.gfdrr.org/ Cyclones: Final Report to the Australian Greenhouse
node/27850. Office. Canberra: CSIRO.
85 Geoscience Australia holds a Creative Commons
ADRC (Asian Disaster Reduction Center). 2006. “Report
Attribution 3.0 Australia license for the material in this on Survey on Tsunami Awareness in Indonesia: Banda
section. All terms of the license apply for reuse of text Aceh and Aceh Besar Area of Aceh Province.” ADRC.
and graphics. http://www.adrc.asia/publications/Indonesia_Survey/
Banda%20Aceh/en/index.html.
86 World Bank, “Tonga to Receive US$1.27 Million Payout for
Cyclone Response,” press release, http://www.worldbank. Ahmad Al Waked. 2011. “Evolution of the Impact of an
org/en/news/press-release/2014/01/23/tonga-to-receive- Earthquake on Aqaba and Jordan Economy and Public
payout-for-cyclone-response. Finances.” In Disaster Risk Assessment for Aqaba. UNDP
and ASEZA. http://www.preventionweb.net/files/31205_
87 Analysis benefited from funding provided under a
aqabasraeisei1.pdf.
grant from the Global Facility for Disaster Reduction
and Recovery. Amarasinghe, S. 2007. “Identifying Vulnerability Using
Semi-structured Interviews.” In Rapid Vulnerability
88 The identification and tracking algorithm used was
Assessment in Sri Lanka: Post-Tsunami Study of Two
based on the works of Nguyen and Walsh (2001), Walsh
Cities: Galle and Batticaloa. SOURCE Publication No. 7,
and Syktus (2003), and Abbs et al. (2006), and applies
47–53. http://ihdp.unu.edu/file/get/3992.pdf.
eight criteria to identify a tropical cyclone. Further details
of the method can be found in Abbs (2012). Annaka, T., K. Satake, T. Sakakiyama, K. Yanagisawa, and
N. Shuto. 2007. “Logic-tree Approach for Probabilistic
89 The five models were ACCESS 1.0, Can ESM, CSIRO
Tsunami Hazard Analysis and Its Applications to the
Mk3.6.0, IPSL CM5A, and NorESM-1M. More information is
Japanese Coasts.” Pure and Applied Geophysics 164:
available about the PACCSAP program on the Australian
577–92.
Department of the Environment website, http://www.
climatechange.gov.au/climate-change/grants/pacific-
Aon Benfield. 2012. Annual Global Climate and
australia-climate-change-science-and-adaptation-
Catastrophe Report: Impact Forecasting, 2011.
planning-program.
http://thoughtleadership.aonbenfield.com/
Documents/20120110_if_annual_global_climate_cat_
90 This case study draws on D. Lallemant, S. Wong, K.
Morales, and A. Kiremidjian, “A Framework and Case Study report.pdf.
for Dynamic Urban Risk Assessment” (paper presented at
Arkema, K. K., G. Guannel, G. Verutes, S. A. Wood, A. Guerry,
the 10th National Conference in Earthquake Engineering,
M. Ruckelshaus, P. Kareiva, M. Lacayo, and J. M. Silver.
Earthquake Engineering Research Institute, Anchorage, AK,
2013. “Coastal Habitats Shield People and Property from
July 2014).
Sea-level Rise and Storms.” Nature Climate Change 3:
91 Rao’s Ph.D. thesis, entitled “Structural Deterioration and
913–18. doi:10.1038/nclimate1944.
Time-Dependent Seismic Risk Analysis,” is being completed
Ambraseys, Nicholas. 2009. Earthquakes in the
at the Blume Earthquake Center, Stanford University.
Mediterranean and Middle East: A Multidisciplinary Study
92 OpenQuake 2013 release, Global Earthquake Model,
of Seismicity up to 1900. Cambridge and New York:
http://www.globalquakemodel.org/openquake/. Cambridge University Press.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
201
202
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Arnell, N. W., and B. Lloyd-Hughes. 2014. “The Global- Bautista, M. L. P., B. C. Bautista, I. C. Narag, A. G. Lanuza, J.
scale Impacts of Climate Change on Water Resources B. Deocampo, K. L. Papiona, R. A. Atando, R. U. Solidum,
and Flooding under New Climate and Socio-economic T. A. Allen, M. Jakab, H. Ryu, M. Edwards, K. Nadimpalli,
Scenarios.” Climatic Change 122: 127–40. doi:10.1007/ M. Leonard, and M. A. Dunford. 2012. “Strengthening
s10584-013-0948-4. Natural Hazard Risk Assessment Capacity in the
Philippines: An Earthquake Impact Pilot Study for Iloilo
Arnold, Margaret, Maxx Dilley, Uwe Deichmann, Robert City, Western Visayas.” Record 2012/070, Geoscience
S. Chen, and Arthur L. Lerner-Lam. 2005. Natural Australia, Canberra. http://www.ga.gov.au/metadata-
Disaster Hotspots: A Global Risk Analysis. Washington, gateway/metadata/record/gcat_74132.
DC: International Bank for Reconstruction and
Development/World Bank and Columbia University. Bear-Crozier, A. N., N. Kartadinata, A. Heriwaseso, and
O. Nielsen. 2012. “Development of Python-Fall3D: A
Arthur, W. C., and H. M. Woolf. 2013. “Assessment of Tropical Modified Procedure for Modelling Volcanic Ash Dispersal
Cyclone Risk in the Pacific Region: Analysis of Changes in the Asia-Pacific.” Natural Hazards 64 (1): 821–38.
in Key Tropical Cyclone Parameters.” Record 2013/23,
Geoscience Australia, Canberra. Bear-Crozier, A., and A. Simpson. 2011. “Development and
Implementation of Volcanic Ash Modelling in West Java:
Arya, A. S. 2000. “Non-engineered Construction in Completion Report for AusAID.” Geoscience Australia
Developing Countries—An Approach toward Earthquake Professional Opinion 2011/04, Canberra.
Risk Prediction.” Bulletin of the New Zealand Society for
Earthquake Engineering 33 (3): 187–208. Beaulieu, A., D. Begin, and D. Genest. 2010. “Community
Mapping and Government Mapping: Potential
ASEP (Association of Structural Engineers of the Collaboration?” Symposium of ISPRS Commission I,
Philippines). 2010. National Structural Code of the Calgary, Canada, June 16–18. http://www.isprs.org/
Philippines, Volume 1: Buildings, Towers, and Other proceedings/xxxviii/part1/10/10_01_Paper_163.pdf.
Vertical Structures. ASEP: Quezon City, Philippines.
Bilham, R. 2009. “The Seismic Future of Cities.” Bulletin of
Atkins. 2012. Shire Integrated Flood Risk Management Earthquake Engineering 7 (4): 839–87.
Program Final Report: Volume 1.
BNPB (National Disaster Management Agency) and
Badilla, R. A., R. M. Barde, G. Davies, A. C. Duran, J. C. Bappenas (National Development Planning Agency),
Felizardo, E. C. Hernandez, M. G. Ordonez, and R. S. with provincial and district/city governments of West
Umali. 2014. “Enhancing Risk Analysis Capacities for Sumatra and Jambi. 2009. “West Sumatra and Jambi
Flood, Tropical Cyclone Severe Wind, and Earthquake for Natural Disasters: Damage, Loss and Preliminary Needs
the Greater Metro Manila Area. Component 3: Flood Risk Assessment.” https://www.gfdrr.org/sites/gfdrr.org/
Analysis.” CSCAND Technical Report for DFAT. files/documents/GFDRR_Indonesia_DLNA.2009.EN_.
pdf.
Barredo, J. I. 2009. “Normalised Flood Losses in Europe:
1970–2006.” Natural Hazards and Earth System Sciences Bouwer, L. M. 2011. “Have Disaster Losses Increased
9: 97-104. doi:10.5194/nhess-9-97-2009. Due to Anthropogenic Climate Change?” Bulletin
of the American Meteorological Society 92: 39-46.
Barthel, F., and E. Neumayer. 2012. “A Trend Analysis of doi:10.1175/2010BAMS3092.1.
Normalized Insured Damage from Natural Disasters.”
Climatic Change 113: 215–37. Burbidge, D., P. Cummins, R. Mleczko, and H. Thio. 2008. “A
Probabilistic Tsunami Hazard Assessment for Western
Batty, M. 2007. Cities and Complexity: Understanding Australia.” Pure and Applied Geophysics 165: 2059–88.
Cities with Cellular Automata, Agent-Based Models, and
Fractals. Cambridge, MA: MIT Press. Burbidge, D., P. R. Cummins, R. Mleczko, and H. K. Thio.
2009. “A Probabilistic Tsunami Hazard Assessment
Bautista, M. L. P., B. Bautista, I. C. Narag, A. D. Aquino, for Western Australia.” In Tsunami Science Four Years
K. Papiona, A. L. Delos Santos, J. Nadua, J. P. Deximo, after the 2004 Indian Ocean Tsunami, edited by Phil R.
M. Jakab, and M. A. Dunford. 2014. “Enhancing Risk Cummins, Laura S. L. Kong, and Kenji Satake, 2059–88.
Analysis Capacities for Flood, Tropical Cyclone Severe Basel, Switzerland: Birkhäuser.
Wind, and Earthquake for the Greater Metro Manila Area
Component 2—Exposure Information Development.” Chapman, K. 2012. “Community Mapping for Exposure in
CSCAND Technical Report for DFAT. Indonesia.” Humanitarian OpenStreetMap Team. http://
CHAPTER
03
hot.openstreetmap.org/sites/default/files/CM4E-Indo- Desramaut, N. 2013. “Functional Vulnerability: Report on
en.pdf. the Functional Vulnerability Assessment of a System
Prone to Multiple Hazards.” Technical Report D4.3,
Chapman, K., A. Wibowo, and Nurwadjedi. 2013. “Filling MATRIX project.
the Data Gap with Participatory Mapping for Effective
Disaster Preparedness.” Paper presented at the Disaster Dilley, M., R. S. Chen, U. Deichmann, A. Lerner-Lam, M.
Risk Management in East Asia and the Pacific Distance Arnold, J. Agwe, P. Buys, O. Kjekstad, B. Lyon, and G.
Learning Seminar Series, Jakarta, June 18. http://www. Yetman. 2005a. Natural Disaster Hotspots. A Global
Risk Analysis. Washington, DC: International Bank for
jointokyo.org/files/cms/news/pdf/(Final)_Session_2_
Reconstruction and Development/World Bank and
Summary.pdf.
Columbia University.
CIMNE (Centro Internacional de Métodos Numéricos
———. 2005b. Natural Disaster Hotspots: A Global
en Ingeniería) et al. 2013. “Probabilistic Modeling of
Risk Analysis. Synthesis Report. Washington,
Natural Risks at the Global Level: Global Risk Model.”
DC: International Bank for Reconstruction and
Background Paper prepared for the 2013 Global
Development/World Bank.
Assessment Report on Disaster Risk Reduction, UNISDR,
Geneva. www.preventionweb.net/gar. Erdik, M. 2013. “Earthquake Risk in Turkey.” Science 341:
724–25.
Cornell, C. A. 1968. “Engineering Seismic Risk Analysis.”
Bulletin of the Seismological Society of America 58: Erian, Wadid, Bassem Katlan, Bassem Ouldbedy, Haider
1583–1606. Awad, Ebrahim Zaghtity, and Sanaa Ibrahim. 2012.
“Agriculture Drought in Africa and Mediterranean.”
Council of European Union. 2009. “Council Conclusions on Background paper prepared for the 2013 Global
a Community Framework on Disaster Prevention within Assessment Report on Disaster Risk Reduction, UNISDR,
the EU.” Minutes of the 2979th Justice and Home Affairs Geneva. www.preventionweb.net/gar.
Council Meeting, Brussels, November 30.
European Commission. 2010a. “The EU Internal Security
Cummins, P., and M. Leonard, M. 2004. “Small Threat but Strategy in Action: Five Steps towards a More Secure
Warning Sounded for Tsunami Research.” AusGeo News Europe.” Communication from the Commission to the
75: 4–7. September. European Parliament and the Council, COM(2010) 673
final, European Commission, Brussels.
Cummins, P. R., D. R. Burbidge, R. Mleczko, D. H.
Natawidjaja, and H. Latief. 2009. Probabilistic ———. 2010b. “Risk Assessment and Mapping Guidelines
Assessment of Tsunami Hazard in the Indian Ocean. for Disaster Management.” Commission Staff Working
Canberra: Geoscience Australia. Paper, SEC(2010) 1626 final, European Commission,
Brussels.
Cyranoski, D. 2011. “Japan Faces Up to Failure of Its
FEMA (Federal Emergency Management Agency). 2003.
Earthquake Preparations.” Nature 471: 556–57.
Multi-hazard Loss Estimation Methodology. Earthquake
Dailey, P., M. Huddleston, S. Brown, and D. Fasking. 2009. Model. HAZUS-MH 2.1. Technical Manual. Washington,
“The Financial Risks of Climate Change.” ABI Research DC: FEMA.
Paper 19, Association of British Insurerers. http://static.
Feyen, L., R. Dankers, K. Bódis, P. Salamon, and J. I. Barredo.
weadapt.org/knowledge-base/files/1040/504a19b1e3d0
2012. “Fluvial Flood Risk in Europe in Present and Future
efinancial-risks-of-climate-change-pdf.pdf.
Climates.” Climatic Change 112: 47–62. doi:10.1007/
s10584-011-0339-7.
Daniell, J. E., B. Khazai, F. Wenzel, and A. Vervaeck. 2011.
“The CATDAT Damaging Earthquakes Database.” Natural Frangopol, D. M., K.-Y. Lin, and A. C. Estes. 1997. “Reliability
Hazards and Earth Systems Sciences 11: 2235–51. of Reinforced Concrete Girders under Corrosion Attack.”
doi:10.5194/nhess-11-2235-2011. Journal of Structural Engineering 123 (3): 286–97.
De Bono, A. 2013. “The Global Exposure Database for GAR Fraser, A., and D. Vincent Lima. 2012. “Survey Results
2013.” Background paper prepared for the 2013 Global Report: Regional Technical Assistance Initiative on
Assessment Report on Disaster Risk Reduction, UNISDR, Climate Adaptation Planning in LAC Cities.” Latin
Geneva. www.preventionweb.net/gar. America and Caribbean Regional Urban, Water,
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
203
204
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
and Disaster Risk Management Unit, World Bank, Government of Malawi. 2010. “National Disaster Risk
Washington, DC. Reduction Framework 2010–2015.”
Gadjah Mada University and HOT (Humanitarian ———. 2011. “National Disaster Risk Management Policy.”
OpenStreetMap Team). 2012. “Evaluation of Department of Disaster Management Affairs.
OpenStreetMap Data in Indonesia: Final Report.”
Government of Morocco. 2012. “Evaluating Direct and
http://openstreetmap.or.id/docs/Final_Report-OSM_
Indirect Economic Impacts of Natural Disasters: The
Evaluation_in_Indonesia_2012.pdf.
Development of an Input-Output and a CGE Models
Gall, M., K. A. Borden, C. T. Emrich, and S. L. Cutter. 2011. for Morocco.”
“The Unsustainable Trend of Natural Hazard Losses in
Hallegatte, S., C. Green, R. J. Nicholls, and J. Corfee-
the United States.” Sustainability 3: 2157–81. doi:10.3390/
Morlot. 2013. “Future Flood Losses in Major Coastal
su3112157.
Cities.” Nature Climate Change 3: 802–6. doi:10.1038/
nclimate1979.
Garcia-Aristizabal, A., W. Marzocchi, and A. Di Ruocco. 2013.
“Probabilistic Framework for Multi-hazard Assessment.” Herold, C., and F. Mouton. 2011. “Global Flood Hazard
Technical Report D3.4, MATRIX project. Mapping Using Statistical Peak Flow Estimates.”
Hydrology and Earth System Sciences Discussions 8:
Gaume, E., V. Bain, P. Bernardara, O. Newinger, M. Barbuc,
305–63. doi:10.5194/hessd-8-305-2011.
A. Bateman, L. Blaškovičová, G. Blöschl, M. Borga, A.
Dumitrescu, I. Daliakopoulos, J. Garcia, A. Irimescu, S. Herold, C., and R. Rudari. 2013. “Improvement of the Global
Kohnova, A. Koutroulis, L. Marchi, S. Matreata, V. Medina, Flood Model for the GAR 2013 and 2015.” Background
E. Preciso, D. Sempere-Torres, G. Stancalie, J. Szolgay, I. paper prepared for the 2013 Global Assessment Report
Tsanis, D. Velasco, and A. Viglione. 2009. “A Compilation on Disaster Risk Reduction, UNISDR, Geneva. www.
of Data on European Flash Floods.” Journal of Hydrology preventionweb.net/gar.
367 (1): 70–78. doi:10.1016/j.jhydrol.2008.12.028.
Hirabayashi, Y., M. Roobavannan, K. Sujan, K. Lisako, Y. Dai,
Geller, R. J. 2011. “Shake Up Time for Japanese Seismology.” W. Satoshi, K. Hyungjun, and K. Shinjiro. 2013. “Global
Nature 472: 407–9. Flood Risk under Climate Change.” Nature Climate
Change 3: 816–21. doi:10.1038/nclimate1911.
GFDRR (Global Facility for Disaster Reduction and
Recovery). 2014. Open Data for Resilience Initiative Field ICHARM (International Centre for Water Hazard and Risk
Guide. Washington, DC: World Bank. Management). 2013. Technical Assistance for Supporting
Investments in Water-related Disaster Management:
Ghosh, Jayadipta, Keivan Rokneddin, Jamie E. Padgett, Main Volume. Manila: Asian Development Bank. http://
and Leonardo Dueñas–Osorio. 2013. “Seismic Reliability www.adb.org/projects/documents/supporting-
Assessment of Aging Highway Bridge Networks with investments-water-related-disaster-management-tacr.
Field Instrumentation Data and Correlated Failures. I:
IPCC (Intergovernmental Panel on Climate Change). 2012.
Methodology.” Earthquake Spectra. doi:http://dx.doi.
Managing the Risks of Extreme Events and Disasters to
org/10.1193/040512eqs155m.
Advance Climate Change Adaptation. A special report
of Working Groups I and II of the Intergovernmental
Goethert, R. 2010. “Incremental Housing.” Monday
Panel on Climate Change. Cambridge and New York:
Developments. September. http://monthlydevelopments.
Cambridge University Press.
org/sites/monthlydevelopments/files/MD_Sept_10_
small.pdf.
———. 2013. “Summary for Policymakers.” In Climate Change
2013: The Physical Science Basis. Contribution of
Gonzalez, F., E. Geist, B. Jaffe, U. Kanoglu, H. Mofjeld, C.
Working Group I to the Fifth Assessment Report of the
Synolakis, V. Titov, D. Arcas, D. Bellomo, D. Carlton, T.
Intergovernmental Panel on Climate Change, edited by
Horning, J. Johnson, J. Newman, T. Parsons, R. Peters,
T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen,
C. Peterson, G. Priest, A. Venturato, J. Weber, F. Wong,
J. Boschung, A. Nauels, Y. Xia, V. Bex, and P. M. Midgley.
and A. Yalciner. 2009. “Probabilistic Tsunami Hazard
Cambridge: Cambridge University Press.
Assessment at Seaside, Oregon, for Near- and Far-field
Seismic Sources.” Journal of Geophysical Research 114: Jankaew, K., B. F. Atwater, Y. Sawai, M. Choowong, T.
C11023. Charoentitirat, M. E. Martin, and A. Prendergast. 2008.
CHAPTER
03
“Medieval Forewarning of the 2004 Indian Ocean Kagan, Y. Y., and D. D. Jackson. 2013. “Tohoku Earthquake:
Tsunami in Thailand.” Nature 455: 1228–31. A Surprise?” Bulletin of the Seismological Society of
America 103 (2B): 1181–94. doi:10.1785/0120120110.
Janssen, G., and D. L. Holden. 2011. External Independent
Evaluation of AIFDR ‘Build Back Better’ Campaign: Final Kundzewicz, Z. W., S. Kanae, S. I. Seneviratne, J. Handmer,
Evaluation Report. Jakarta: Australia-Indonesia Facility N. Nicholls, P. Peduzzi, R. Mechler, L. M. Bouwer, N.
for Disaster Reduction. Arnell, K. Mach, R. Muir-Wood, G. R. Brakenridge,
W. Kron, G. Benito, Y. Honda, K. Takahashi, and B.
———. 2013. Rumah Aman Gempa Andalan Masyarakat Sherstyukov. 2014. “Flood Risk and Climate Change:
(RAGAM): Independent Completion Report. Jakarta: Global and Regional Perspectives.” Hydrological
Australia-Indonesia Facility for Disaster Reduction. Sciences Journal 59 (1). doi:10.1080/02626667.2013.857
411.
Jayanthi, H., and G. J. Husak. 2012. “A Probabilistic Approach
to Assess Agricultural Drought Risk.” Background Kundzewicz, Z. W., I. Pińskwar, and G. R. Brakenridge.
paper prepared for the 2013 Global Assessment Report 2013. “Large Floods in Europe, 1985–2009.” Hydrological
Sciences Journal 58: 1–7. doi:10.1080/02626667.2012.74
on Disaster Risk Reduction, UNISDR, Geneva. www.
5082.
preventionweb.net/gar.
Kwak, Y., K. Takeuchi, K. Fukami, and J. Magome. 2012. “A
Jayasinghem, T., and J. Birkmann. 2007. “Vulnerability
New Approach to Flood Risk Assessment in Asia-Pacific
Comparison between Galle and Batticaloa based on
Region Based on MRI-AGCM Outputs.” Hydrological
the Household Survey using Questionnaires.” In Rapid
Research Letters 6: 70–75.
Vulnerability Assessment in Sri Lanka: Post-Tsunami
Study of Two Cities: Galle and Batticaloa. SOURCE Lall, S. V., and U. Deichmann. 2009. “Density and Disasters:
Publication No. 7, 39–47. Economics of Urban Hazard Risk.” Policy Research
Working Paper 5161, World Bank, Washington, DC.
Jenkins, S., C. Magill, K. McAneney, and R. Blong. 2012a.
“Regional Ash Fall Hazard I: A Probabilistic Assessment ———. 2012. “Density and Disasters: Economics of Urban
Methodology.” Bulletin of Volcanology 74: 1699–1712. Hazard Risk.” World Bank Research Observer 27 (1):
74–105.
Jenkins, S., K. McAneney, C. Magill, and R. Blong. 2012b.
“Regional Ash Fall Hazard II: Asia-Pacific Modelling Lallemant, D., S. Wong, K. Morales, and A. Kiremidjian.
Results and Implications.” Bulletin of Volcanology 74: 2014. “A Framework and Case Study for Dynamic Urban
1713–27. Risk Assessment.” Paper presented at the 10th National
Conference on Earthquake Engineering, Anchorage, AK,
JICA (Japan International Cooperation Agency). 2011. “The July.
Study on Integrated Water Resources Management
Lansang, M., and R. Dennis. 2004. “Building Capacity in
for Poverty Alleviation and Economic Development in
Health Research in the Developing World.” Bulletin of the
the Pampanga River Basin.” National Water Resources
World Health Organisation 82: 764–70.
Board, Philippines.
Løvholt, F., D. Kühn, H. Bungum, C. B. Harbitz, and S.
———. 2013. “Project for the Comprehensive Flood
Glimsdal. 2012. “Historical Tsunamis and Present
Management Plan for the Chao Phraya River Basin.”
Tsunami Hazard in Eastern Indonesia and the
Thailand.
Philippines.” Journal of Geophysical Research–Solid
Earth. doi:10.1029/2012JB009425.
Jongman B., S. Hochrainer-Stigler, L. Feyen, J. C. J. H. Aerts,
R. Mechler, W. J. W. Botzen, L. M. Bouwer, G. Pflug,
Marzocchi, W., A. Garcia-Aristizabal, P. Gasparini, M. L.
R. Rojas, and P. J. Ward. 2014. “Increasing Stress on Mastellone, and A. Di Ruocco. 2012. “Basic Principles of
Disaster Risk Finance Due to Large Floods.” Nature Multi-risk Assessment: A Case Study in Italy.” Natural
Climate Change. http://dx.doi.org/10.1038/nclimate2124. Hazards 62 (2): 551–73.
Jongman, B., P. J. Ward, and J. C. J. H. Aerts. 2012. “Global McCloskey, J., D. Lange, F. Tilmann, S. S. Nalbant, A.
Exposure to River and Coastal Flooding—Long Term F. Bell, D. H. Natawidjaja, and A. Rietbrock. 2010.
Trends and Changes.” Global Environmental Change 22: “The September 2009 Padang Earthquake.” Nature
823–35. doi:10.1016/j.gloenvcha.2012.07.004. Geoscience 3 (2): 70–71.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
205
206
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
McKee, T. B., N. J. Doesken, and J. Kleist. 1993. “The Ordaz, M., F. Martinelli, A. Aguilar, J. Arboleda, C. Meletti, and
Relationship of Drought Frequency and Duration of V. D’Amico. “2012 CRISIS 2012, Program for Computing
Time Scales.” Paper presented at Eighth Conference on Seismic Hazard.” Instituto de Ingeniería, Universidad
Applied Climatology, American Meteorological Society, Nacional Autónoma de México.
Anaheim CA, January 17–23.
ORNL (Oak Ridge National Laboratory). 2007. “LandScanTM
MLIT (Ministry of Land, Infrastructure, Transport Global Population Distribution Data (Raster dataset).”
and Tourism). 2006. Basic Plan for the Tone River Oak Ridge National Laboratory, U.S. Department of
Improvement. Tokyo: MLIT. Energy. www.ornl.gov/sci/landscan/index.shtml.
Monfort, D., and S. Lecacheux. 2013. “West Indies Test Site.” Pacheco, B. M., J. Y. Hernandez Jr., E. A. J. Tingatinga, P. P.
Technical Report D7.4, MATRIX project. M. Castro, F. J. Germar, U. P. Ignacio, M. C. L. Pascua,
L. R. E. Tan, I. B. O. Villalba, D. H. M. Aquino, R. E. U.
MunichRe. 2013. “Significant Natural Catastrophes Longalong, R. N. Macuha, W. L. Mata, R. M. Suiza, and M.
1980–2012, 10 Deadliest Worldwide Events.” Geo Risks A. H. Zarco. 2013. “Development of Vulnerability Curves
Research, NatCatSERVICE. https://www.munichre. of Key Building Types in the Greater Metro Manila Area,
com/site/corporate/get/documents_E-1233315815/ Philippines.” Institute of Civil Engineering, University of
mr/assetpool.shared/Documents/0_Corporate%20 the Philippines Diliman, Quezon City, Philippines.
Website/_NatCatService/Focus_Analyses/1980-2012-
Pacific Consultants International, OYO Cooperation, JICA,
geophysical-events-worldwide-en.pdf.
and IMM. 2002. “The Study on a Disaster Prevention
Nadim, F., and T. Glade. 2006. “On Tsunami Risk / Mitigation Basic Plan in Istanbul including Seismic
Assessment for the West Coast of Thailand.” ECI Microzonation in the Republic of Turkey.” http://www.ibb.
Symposium Series 7, Engineering Conferences gov.tr/tr-TR/SubSites/DepremSite/PublishingImages/
International, New York. http://dc.engconfintl.org/cgi/ JICA_ENG.pdf.
viewcontent.cgi?article=1000&context=geohazards.
Pappenberger, F., E. Dutra, F. Wetterhall, and H. L. Cloke.
Neumayer, E., and F. Barthel. 2011. “Normalizing Economic 2012. “Deriving Global Flood Hazard Maps of Fluvial
Loss from Natural Disasters: A Global Analysis.” Global Floods through a Physical Model Cascade.” Hydrology
Environmental Change 21 (1): 13–24. doi:10.1016/j. and Earth System Sciences 16: 4143–56. doi:10.5194/
hess-16-4143-2012.
gloenvcha.2010.10.004.
Parsons, T., and E. Geist. 2009. “Tsunami Probability in the
NGI (Norwegian Technological Institute). 2013. “Landslide
Caribbean Region.” Pure and Applied Geophysics 165:
Hazard and Risk Assessment in El Salvador.” Background
2089–2116.
paper prepared for the 2013 Global Assessment Report
on Disaster Risk Reduction, UNISDR, Geneva. www.
Parsons, Tom, Shinji Toda, Ross S. Stein, Aykut Barka,
preventionweb.net/gar.
and James H. Dieterich. 2000. “Heightened Odds of
Large Earthquakes Near Istanbul: An Interaction Based
Nguyen, K. C., and K. J. E. Walsh. 2001. “Interannual,
Probability Calculation.” Science 288: 661–65.
Decadal, and Transient Greenhouse Simulation of
Tropical Cyclone-like Vortices in a Regional Climate Pektas, Mesut, and Polat Gulkan. 2004. “A Metropolitan
Model of the South Pacific.” Journal of Climate 14: Municipality Prepares for the Worst: Istanbul Earthquake
3043–54. Master Plan.” Paper presented at 13th World Conference
on Earthquake Engineering, Vancouver, August 1–6.
Nicholls, R. J., S. Hanson, C. Herweijer, N. Patmore, S.
Hallegatte, J. Corfee-Morlot, J. Chateau, and R. Muir PBL (Netherlands Environmental Assessment Agency).
Wood. 2008. “Ranking Port Cities with High Exposure 2014. Towards a World of Cities in 2050: An Outlook on
and Vulnerability to Climate Extremes: Exposure Water-related Challenges. PBL background report for
Estimates.” OECD Working Paper No. 1, OECD Publishing, UN Habitat Global Report. The Hague: PBL Netherlands
Paris. Environmental Assessment Agency.
OECD (Organisation for Economic Co-operation and Peduzzi, P., B. Chatenoux, H. Dao, A. De Bono, U.
Development). 2012. OECD Environmental Outlook Deichmann, G. Giuliani, C. Herold, B. Kalsnes, S. Kluser,
to 2050. Paris: OECD Publishing. http://dx.doi. F. Løvholt, B. Lyon, A. Maskrey, F. Mouton, F. Nadim, and
org/10.1787/9789264122246-en. H. Smebye. 2009. “The Global Risk Analysis for the 2009
CHAPTER
03
Global Assessment Report on Disaster Risk Reduction.” Scolobig, A., C. Vichon, N. Komendantova, M. Bengoubou-
Universite de Geneve Publication. http://www-fourier. Valerius, and A. Patt. 2013. “Social and Institutional
ujf-grenoble.fr/~mouton/Publis_HDR_applis/Peduzzi- Barriers to Effective Multi-hazard and Multi-risk
The_Global_Risk_Analysis_for_the_2009_GAR-149. Decision-making Governance.” Technical Report D6.3,
pdf. MATRIX project.
Petiteville, Ivan, Philippe Bally, and Guy Seguin. 2012. SCS (Soil Conservation Service). 1986. “Urban Hydrology for
“Satellite Earth Observation for Risk Management.” In Small Watersheds.” Technical Release TR-55, USDA, Soil
The Earth Observation Handbook, edited by S. Ward. Conservation Service, Hydrology Unit.
European Space Agency. www.eohandbook.com.
Sengara, W., M. Suarjana, D. Beetham, N. Corby, M.
PHIVOLCS (Philippine Institute of Volcanology and Edwards, M. Griffith, M. Wehner, and R. Weller. 2010.
Seismology), JICA (Japan International Cooperation “The 30th September 2009 West Sumatra Earthquake:
Agency), and MMDA (Metropolitan Manila Development Padang Region Damage Survey.” Record 2010/44,
Authority). 2004. “Earthquake Impact Reduction Study Geoscience Australia, Canberra.
for Metropolitan Manila, Republic of the Philippines”
(MMEIRS). Sengara, W., M. Suarjana, M. Edwards, H. Ryu, W.
Rahmanusyairi, I. Adiputra, I. I. Wahdiny, A. Utami,
Prochaska, J. O., J. C. Norcross, and C. C. DiClemente. 1994. A. Mariany, M. A. Yulman, and B. Novianto. 2013.
Changing for Good: The Revolutionary Program that Research on Earthquake Damage Models for Buildings
Explains the Six Stages of Changes and Teaches You in Indonesia. Bandung: Research Centre of Disaster
How to Free Yourself from Bad Habits. New York: W. Mitigation, Bandung Institute of Technology.
Morrow.
Setiadi, N. J. 2014. “Assessing People´s Early Warning
RMSI. 2011. “Malawi: Economic Vulnerability and Disaster
Response Capability to Inform Urban Planning
Risk Assessment.” www.masdap.mw/documents/145/
Interventions to Reduce Vulnerability to Tsunamis: Case
download.
Study of Padang City, Indonesia.” PhD dissertation,
Rheinische Friedrich‐Wilhelms‐Universität zu Bonn.
Robinson, D., T. Dhu, and J. Schneider. 2006. “Practical
Probabilistic Seismic Risk Analysis: A Demonstration
Setiadi, N., H. Taubenböck, S. Raupp, and J. Birkmann. 2010.
of Capability.” Seismological Research Letters 77 (4):
“Integrating Socio-Economic Data in Spatial Analysis:
453–59.
An Exposure Analysis Method for Planning Urban Risk
Rojas, R., L. Feyen, and P. Watkiss. 2013. “Climate Change Mitigation.” Paper presented at REAL CORP 2010, Vienna,
and River Floods in the European Union: Socio- May 18–20. http://programm.corp.at/cdrom2010/
Economic Consequences and the Costs and Benefits of papers2010/CORP2010_80.pdf.
Adaptation.” Global Environmental Change 23: 1737–51.
Seto, K. C., B. Güneralp, and L. R. Hutyra. 2012. “Global
doi:10.1016/j.gloenvcha.2013.08.006.
Forecasts of Urban Expansion to 2030 and Direct
Rokneddin, Keivan, Jayadipta Ghosh, Leonardo Dueñas– Impacts on Biodiversity and Carbon Pools.” Proceedings
Osorio, and Jamie E. Padgett. 2013. “Seismic Reliability of the National Academy of Sciences 109: 16083–16088.
Assessment of Aging Highway Bridge Networks with doi:10.1073/pnas.1211658109.
Field Instrumentation Data and Correlated Failures.
Shela, Osborne, Gaye Thompson, Paul Jere, and George
II: Application.” Earthquake Spectra. doi:http://dx.doi.
Annandale. 2008. “Analysis of Lower Shire Floods &
org/10.1193/040612EQS160M.
a Flood Risk Reduction and Recovery Programme
Satake, K., and B. F. Atwater. 2007. “Long-Term Proposal for the Lower Shire Valley.” Government of
Perspectives on Giant Earthquakes and Tsunamis Malawi, Department of Disaster Management Affairs,
at Subduction Zones.” Annual Review of Earth and Office of the President and Cabinet.
Planetary Sciences 35: 349–74.
Sieh, K., D. H. Natawidjaja, A. J. Meltzner, C. C. Shen, H.
Schneider, A., M. A. Friedl, and D. Potere. 2009. “A New Cheng, K. Li, B. W. Suwargadi, J. Galetzka, B. Philibosian,
Map of Global Urban Extent from MODIS Satellite and R. L. Edwards. 2008. “Earthquake Supercycles
Data.” Environmental Research Letters 4 (4): 044003. Inferred from Sea-Level Changes Recorded in the Corals
doi:10.1088/1748-9326/4/4/044003. of West Sumatra.” Science 322 (5908): 1674–78.
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
207
208
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
Simpson, A. L., and T. Dhu. 2009. “Enhancing Natural Tsunami Pilot Study Working Group. 2006. “Seaside,
Hazard Risk Assessment Capacity in the CSCAND Oregon Tsunami Pilot Study—Modernization of FEMA
Agencies in the Philippines: An Options Paper.” Flood Hazard Maps.” NOAA OAR Special Report. NOAA/
Geoscience Australia Professional Opinion 2009/004, OAR/PMEL, Seattle.
Canberra.
UNDP (United Nations Development Programme) and
Soden, R., N. Budhathoki, and L. Palen. 2014. “Resilience RMSI. 2006. “Developing a Disaster Risk Profile for
Building and the Crisis Informatics Agenda: Lessons Maldives.” http://www.preventionweb.net/files/11145_
Learned from Open Cities Kathmandu.” Proceedings MaldivesDisasterRiskProfileFinalRep.pdf.
of Information Systems for Crisis Response and
UNDP (United Nations Development Programme). 2004.
Management Conference, State College, PA.
Reducing Disaster Risk: A Challenge for Development.
Sørensen, M., A. Babeyko, S. Wiemer, and G. Grünthal. 2012. New York: United Nations.
“Probabilistic Tsunami Hazard in the Mediterranean Sea.”
UN-HABITAT. 2010. “Estado de Las Ciudades de America
Journal of Geophysical Research: Solid Earth 117 (B1):
Latina y El Caribe.” Regional Office for Latin America
2156–2202.
and the Caribbean, United Nations Human Settlement
Programme, Rio de Janeiro.
Stein, S., and E. Okal. 2007. “Ultralong Period Seismic Study
of the December 2004 Indian Ocean Earthquake and
UNISDR (United Nations Office for Disaster Risk Reduction).
Implications for Regional Tectonics and the Subduction 2005a. “10 Preliminary Lessons Learned from the Indian
Process.” Bulletin of the Seismological Society of Ocean Tsunami of 26 December 2004.” http://www.
America 97 (1A): 279–95. unisdr.org/files/5605_ISDR10lessonslearned.pdf.
Stürck, J., A. Poortinga, and P. H. Verburg. 2014. “Mapping ———. 2005b. 2005 Hyogo Framework for Action 2005–2015:
Ecosystem Services: The Supply and Demand of Flood Building the Resilience of Nations and Communities to
Regulation Services in Europe.” Ecological Indicators 38: Disasters. Geneva: UNISDR. www.unisdr.org/wcdr.
198a–211. doi:10.1016/j.ecolind.2013.11.010.
———. 2009. Global Assessment Report on Disaster Risk
Takeuchi, K., T. Ao, and H. Ishidaira. 1999. “Introduction of Reduction: Risk and Poverty in a Changing Climate.
Block-wise Use of TOPMODEL and Muskingum-Cunge Geneva: UNISDR. http://www.preventionweb.net/gar.
Method for the Hydro-environmental Simulation of a
Large Ungauged Basin.” Hydrological Sciences Journal ———. 2011. Global Assessment Report on Disaster Risk
44 (4): 633–46. Reduction 2011: Revealing Risk, Redefining Development.
Geneva: UNISDR. http://www.preventionweb.net/gar.
Taubenböck, H., J. Post, A. Roth, G. Strunz, R. Kief, S.
———. 2013a. Global Assessment Report 2013 Annex 1.
Dech, and F. Ismail. 2008. “Multi-scale Assessment of
Geneva: UNISDR. http://www.preventionweb.net/gar.
Population Distribution Utilizing Remotely Sensed Data:
The Case Study of Padang, West Sumatra, Indonesia.”
———. 2013b. Global Assessment Report on Disaster Risk
Paper presented at International Conference on Tsunami
Reduction 2013: From Shared Risk to Shared Value.
Warning, Bali, Indonesia, November 12–14.
Geneva: UNISDR. http://www.preventionweb.net/gar.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl. 2012. “An United Nations. 2007. World Urbanization Prospects.
Overview of CMIP5 and the Experiment Design.” Bulletin http://www.un.org/esa/population/publications/
of the American Meteorological Society 93: 485–98. wup2007/2007WUP_Highlights_web.pdf.
Thio, H. K., P. Somerville, and J. Polet. 2010. “Probabilistic Walsh, K. J. E., and J. I. Syktus. 2003. “Simulations of
Tsunami Hazard in California.” PEER Report 2010/108, Observed Interannual Variability of Tropical Cyclone
Pacific Earthquake Engineering Research Center, Formation East of Australia.” Atmospheric Science
University of California, Berkeley. Letters 4: 28–40.
Thomas, C., and D. Burbidge. 2009. “A Probabilistic WAPMERR (World Agency for Planetary Monitoring and
Assessment of Tsunami Hazard of Southwest Pacific Earthquake Risk Reduction). 2013. “Approximate Model
Nations.” Geoscience Australia Professional Opinion for Worldwide Building Stock in Three Size Categories of
2009/02, Canberra. Settlements.” Background paper prepared for the 2013
CHAPTER
03
Global Assessment Report on Disaster Risk Reduction, Strategy Development, Final Report.” Prepared by RMSI
UNISDR, Geneva. www.preventionweb.net/gar. Ltd. for the International Bank for Reconstruction and
Development, Washington, DC.
Ward, P. J., W. Beets, L. M. Bouwer, J. C. J. H. Aerts, and
H. Renssen. 2010. “Sensitivity of River Discharge to Yamin, Luis Eduardo, Francis Ghesquiere, Omar Darío
ENSO.” Geophysical Research Letters 37: L12402. Cardona, and Mario Gustavo Ordaz. 2013. Probabilistic
doi:10.1029/2010GL043215. Modelling for Disaster Risk Management: The Case
of Bogota, Colombia (in Spanish). Washington, DC:
Ward, P. J., M. Dettinger, B. Jongman, M. Kummu, F. International Bank for Reconstruction and Development
Sperna Weiland, and H. Winsemius. 2013a. “Flood Risk and World Bank.
Assessment at the Global Scale—The Role of Climate
Variability.” Paper presented at EGU General Assembly,
Vienna, Austria, April 7–12.
Ward, P. J., S. Eisner, M. Flörke, M. D. Dettinger, and M.
Kummu. 2014. “Annual Flood Sensitivities to El Niño
Southern Oscillation at the Global Scale.” Hydrology and
Earth System Sciences 18: 47–66. doi:10.5194/hess-18-
47-2014.
Ward, P. J., B. Jongman, F. Sperna Weiland, A. Bouwman, R.
Van Beek, M. Bierkens, W. Ligtvoet, and H. Winsemius.
2013b. “Assessing Flood Risk at the Global Scale: Model
Setup, Results, and Sensitivity.” Environmental Research
Letters 8: 044019. doi:10.1088/1748-9326/8/4/044019.
Winsemius, H. C., R. Van Beek, B. Jongman, P. J. Ward,
and A. Bouwman. 2013. “A Framework for Global River
Flood Risk Assessments.” Hydrology and Earth System
Sciences 17 (5): 1871–92. doi:10.5194/hess-17-1871-2013.
Wisner, B., P. Blaikie, T. Cannon, and I. Davis. 2004. At Risk:
Natural Hazards, People’s Vulnerability and Disasters.
2nd ed. New York: Routledge.
World Bank. 2010a. Emerging Stronger from the Crisis.
Vol. 1 of World Bank East Asia and Pacific Update
2010. https://openknowledge.worldbank.org/
handle/10986/2455.
———. 2010b. Natural Hazards, UnNatural Disasters: The
Economics of Effective Prevention. Washington, DC:
World Bank.
———. 2012a. Bangladesh: Towards Accelerated, Inclusive,
and Sustainable Growth—Opportunities and Challenges.
Vol. 2. Washington, DC: Poverty Reduction and Economic
Management Sector Unit, South Asia Region, World
Bank.
———. 2012b. “Consultancy for Prioritization of High Seismic
Risk Provinces and Public Buildings in Turkey, by
Proto Engineering.”
———. 2013. “Risk Assessment Report, Morocco Natural
Hazards Probabilistic Risk Assessment and National
CASE STUDIES HIGHLIGHTING EMERGING BEST PRACTICES
209
210
CHAPTER
04
U N D E R S TA N D I N G R I S K I N A N E V O LV I N G W O R L D
RECOMMENDATIONS
T his publication has highlighted the remarkable
progress made in understanding, quantifying,
and communicating risk since 2005, when the
greater transparency and accountability in the risk
assessment process. We stress, however, that the
best outcomes are likely to be achieved when those
Hyogo Framework for Action was endorsed. The investing in risk information and those carrying
array of projects and experiences described here out the risk analysis work in concert and share a
for 40 countries demonstrates that no single common understanding of the undertaking.
approach to risk assessment is right in every
case, and that the best risk assessments are those 1. Clearly define the purpose of the
tailored to the context and identified need. At the risk assessment before analysis
same time, the recurrence of certain themes across starts.
the various projects makes it possible to start
Too many risk assessments are implemented
framing best practices and suggests some concrete
precipitously. These risk assessments—initiated
recommendations for the next 10 years of risk
without first defining a question to be answered and
assessment practice.
a specific end-user—often become scientific and
The recommendations we offer here draw on engineering exercises that upon completion must
submissions to this publication as well as on find a use case and a purpose. Properly targeted
discussions with both users and developers of assessments, on the other hand, suit their intended
risk information. For users of risk information— purpose and are not over-engineered or over-
disaster risk management (DRM) practitioners, resourced. If a community seeks to understand the
government officials, donors, and nongovernmental hazards it faces and to develop plans for evacuation,
organizations (NGOs)—our key recommendations then mapping of exposure and natural hazards is a
are designed to ensure that any investment valid approach, but a different approach would be
in risk assessment promotes more resilient needed for financial planning or retrofitting design.
development and communities. For developers of Similarly, collecting detailed site-level construction
risk information, we see an opportunity to promote information on selected buildings may be
211
212
R E C O M M E N DAT I O N S
appropriate for the design of retrofitting measures, project supplied the evidence for an earthquake
but this approach is not practical for a national-level risk management master plan and served as
risk assessment. the basis for an operational framework for
earthquake risk reduction.
Where risk assessments have been commissioned
in response to a clear and specific request for
information, they have tended to be effective in
reducing fiscal or physical risk. Among the well- 2. Promote and enable ownership
targeted risk assessments described in part 3 of this of the risk assessment process and
publication, we note here the following: efforts to mitigate risk.
• The Pacific Catastrophic Risk Assessment
A sense of ownership is critical to ensuring that
and Risk Financing Initiative (PCRAFI)
knowledge created through a risk assessment
(section 3-19). PCRAFI was designed to inform
is promulgated and acted upon. Countries,
risk financing and insurance options, and
communities, and individuals must feel they have a
ultimately to transfer risk to the international
stake in and connection to risk information if that
financial market. Given this purpose, the analysis
information is to be used, especially by government.
had to conform to standards acceptable to
In many countries, if risk information is not seen as
the financial market. The first payout of the
authoritative—if it is not understood to originate
Pacific catastrophe risk pool in 2014 in Tonga
from government-mandated agencies—it will not be
is testament to the success of this project. An
additional benefit of the project is that the data used in decision making.
and analysis generated have been made available
Risk information can be generated anywhere. Risk
to all stakeholders to use for other purposes
assessment specialists in London, for example,
(for example, to determine how cyclone risk will
can generate risk information on flood in Pakistan.
change as climate change effects are increasingly
But extensive experience suggests that unless the
felt; see section 3-24).
Pakistan authorities have been actively engaged in
• The assessment of seismic risk to Costa the assessment process, the information produced,
Rican water and sanitation systems no matter how accurate or sophisticated, will
(section 3-12). Costa Rican water and sanitation have limited or no uptake and use. Engagement
officials seeking to ensure continuation of with official government stakeholders and local
services following an earthquake created the specialists—at the start of a risk assessment,
demand for this project. The development of the through its implementation, and finally to its
objectives, collection of data, and presentation of conclusion—is critical for the success of a DRM
results were all aimed toward a very specific goal, effort.
and as a consequence resources and ultimately
results were used efficiently. Fortunately, as many of the projects described in
part 3 make clear, the importance of ownership is
• Urban seismic risk mapping to inform increasingly being recognized:
DRM plans in Aqaba, Jordan (section 3-10).
This project was initiated to manage the urban • In Jordan, local scientific and government
development expected in response to Aqaba’s groups partnered with international and other
being declared a special economic zone. The development agencies to integrate seismic risk
CHAPTER
04
reduction considerations into Aqaba’s economic between the government of Australia and various
development (see section 3-10). scientific/technical agencies in Asia and the Pacific
(section 3-9), and between the World Bank and
• In Malawi, the government partnered with the
countries in Latin America and the Caribbean
World Bank and Global Facility for Disaster
(section 3-12)—have also been an important means
Reduction and Recovery to assess flood risk
of promoting ownership. A number of elements go
in the Shire River Valley as part of an effort to
into assuring the success of these partnerships: high
reduce entrenched poverty and make the valley a
levels of trust developed over long periods of time;
national economic hub (see section 3-14).
a focus on work that builds on existing capabilities
• In Peru, Technical Assistance Projects fostered a and interests; and the involvement of credible,
hands-on approach to generating, understanding, capable, and committed experts who understand the
managing, and using risk information, and thus partner country’s systems and cultures, including
promoted ownership of the process and the its language.
results of the assessment (see section 3-12).
The crucial role of ownership is also evident in the
3. Cultivate and promote the
increasing part played by volunteers in collecting
generation and use of open data.
fundamental data used in risk assessments (such
as through volunteered geospatial information, or All the case studies featured in this publication make
crowdsourcing). Especially relevant case studies clear that the creation and use of open data should
are described in sections 3-2, 3-3, and 3-4. This be encouraged.
shift toward community participation reflects
communities’ sense that they can contribute to A risk assessment that yields only a paper or
understanding and mitigating the risk they face. PDF report is of limited use. Its relevance and
Experience shows that governments and decision appropriateness are of short duration, and few
makers increasingly recognize the value and the decision makers are likely to engage with it. A risk
potential of this approach, but consider it critical assessment that shares with stakeholders the data
that the data are certified (for accuracy). In many it has collected and improved, on the other hand,
cases governments would also like to harness will have a much greater impact. The effort required
volunteer efforts toward particular needs—for to collect exposure information is substantial, but
example, may wish to direct volunteers toward fortunately, the data sets produced have relevance
and use for a range of DRM purposes as well as
collecting information about buildings’ attributes
for urban and local planning. If all the input data
(such as use, number of floors, vintage, and
sets and final results are made technically open,
structural materials) rather than focusing on
the broader community is able to engage through
buildings’ location and footprint. Universities
improvements in data and development of new
have shown themselves to be excellent partners
applications and information for community
in this type of volunteer data collection, and their
awareness; and the private sector is able to access
participation assists with ownership and helps to
data that can improve its resiliency. Data sharing
ensure data’s scientific validity.
can also redound to the advantage of those who
Partnerships designed to both produce risk undertook the original assessment, because it
information and build capacity—such as those allows new data to be exploited when they become
R E C O M M E N DAT I O N S
213
214
R E C O M M E N DAT I O N S
available; this means that additional or new analysis 4. Make better communication of
is less of a drain on resources and takes less time risk information an urgent priority.
than it otherwise would.
Clear communication throughout the risk
With respect to creation of new open data, our assessment process, from initiation through delivery
short experience is only beginning to speak to the of the results and the development of plans in
immense potential of structured and unstructured response, is critical for successfully mitigating
volunteered geospatial information (section 3-3), disaster risk.
better access to remote sensing data over wider
A case study featured in section 3-21—“Build Back
areas (boxes 2-2 and 2-3), better ways of exploiting
Better: Where Knowledge Is Not Enough”—is a
and integrating new exposure data sets and models
must-read for all risk assessment practitioners and
(“Exposure” in part 2), and release of technically
disaster risk managers who believe that exceptional
open data sets by governments (section 3-3), the
communication of risk information is the key to
private sector, and NGOs.
preparedness and risk reduction. A massive “Build
It is clear from case studies and research that greater Back Better” campaign led by the government of
effort is needed to open up and improve damage Indonesia in the aftermath of the 2009 Padang
and loss data collections to make them meaningful earthquake demonstrates conclusively that well-
targeted education and communication of risk
and useful for understanding and quantifying risk.
information can increase awareness of natural
An encouraging sign is a pilot being undertaken by
hazards and their potential impacts. Analysis also
the Insurance Bureau of Canada that will give cities
shows, however, that progress from increased
access to flood insurance claims data, alongside
awareness to substantive action is very difficult to
municipal infrastructure data and current and future
achieve, even in a community that has witnessed
climate data on flood93—a significant step toward
at first hand the devastation of an earthquake.
better understanding and managing urban flood risk.
The study finds overall that homeowners can be
Given the benefits it stands to gain, the global DRM motivated to put risk knowledge into practice and
community needs to be willing to press for greater build more resilient homes if they are offered the
access to fundamental data sets that quantify risk. correct combination of timely information, technical
Without access to higher-resolution digital elevation training, community supervision, and financial and
models, results for flood, tsunami, and storm nonfinancial incentives and disincentives.
surge inundation may be impossible to produce Some of the improvements that can be made
at the necessary resolution, or may be massively in communicating risk at the subnational and
inaccurate. Similar gaps in fundamental data exist city levels are evident in the InaSAFE project in
across all hazard areas, and these are hindering the Indonesia (section 3-22). Among the key partners
development of robust and accurate information. In in InaSAFE’s development were Indonesian
many cases the needed data already exist but are not authorities, who realized the need for interactive
accessible. If the DRM community comes together risk communication tools that could robustly and
and advocates for these data to become technically simply answer “what if?” questions. InaSAFE is
open, access is likely to improve and data gaps to demand driven, included user participation in its
be closed. development, uses open data and an open model,
CHAPTER
04
and offers extensive graphical displays (provided by sets, and knowledge related to seismic risk
a GIS system) and an extensive training program. (section 3-6).
Communication was frequent and wide-ranging
throughout the development of InaSAFE and • The Willis Research Network initiative
continued during the collection of data, the use of links more than 50 international research
the model, and the formation of response plans. The institutions to the expertise of the financial and
software has won awards and is being used in other insurance sector in order to support scientists’
countries, including the Philippines and Sri Lanka. quantification of natural hazard risk (box 2-12).
To build on this progress in communicating risk, • The Understanding Risk community
significant investment and innovation will be of practice, made up of more than 3,000
needed in coming years. practitioners from across all sectors in more
than 125 countries, is creating new partnerships
and catalyzing advances in understanding,
5. Foster multidisciplinary, multi-
quantifying, and communicating natural hazard
institutional, and multi-sectoral risk (box 2-11).
collaborations at all levels, from
• The Bangladesh Urban Earthquake
international to community.
Resilience Project is a platform for addressing
Risk assessment is a multidisciplinary and multi- urban risk that brings together officials in
institutional effort that requires collaborations at planning, governance, public service, and
many levels, from international, to national and construction code development as well as
subnational, down to the individual. scientists and engineers, and that fosters
Generating a usable risk assessment product consensus on how to overcome institutional,
involves consultations among technical experts, legislative, policy, and behavioral barriers to a
decision makers, and disaster managers, who must more earthquake-resilient city (box 2-13).
reach agreement on the risk assessment’s purpose
One key task of these and similar collaborations is
and process. Collaboration among technical
reaching out to communities to build consensus,
disciplines, agencies, governments, NGOs, and
raise awareness, and promote action concerning the
virtual communities, as well as informal peer-
risks they face. Greater effort is needed to provide
to-peer exchanges and engagement with local
communities, will help an effort succeed. national- and subnational-level information on risk
to community groups and NGOs working at the
This publication draws attention to a variety community level. Too often, organizations working
of collaborations that aim to build better risk within communities to increase preparedness
information:
and reduce risk lack access to this relevant risk
• The Global Earthquake Model brings together information. Significant gains could arise from
public institutions, private sector institutions merging work being produced at national or
(most notably insurance and reinsurance subnational level with communities’ understanding
agencies), NGOs, and the academic sector, all of their risks and challenges—but this opportunity
with the goal of improving access to tools, data has as yet rarely been capitalized upon.
R E C O M M E N DAT I O N S
215
216
R E C O M M E N DAT I O N S
6. Consider the broader risk to act now to avoid or mitigate the risk they will face
context. in the future. Getting ahead of risk is particularly
important in rapidly urbanizing areas or where
Rarely do countries, communities, or citizens face climate change impacts will be felt the most.
potential risks from only one hazard, or even from
natural hazards alone. Our complex environments The evolution of meteorological hazard arising from
and social structures are such that multiple or climate change will likely occur slowly. The same is
connected risks—from financial hazards, multiple true for changes in hazard due to sea-level rise (for
or cascading natural hazards, and anthropogenic example, with higher sea levels, inundation from
hazards—are the norm. A risk assessment that storm surge and tsunami events may reach further
accounts for just one hazard may struggle with inland). That said, it is possible today—with varying
relevance and will not necessarily speak to a levels of uncertainty—to estimate how climate
decision maker who is responsible for broader risk change may affect losses from meteorological
management. Moreover, failure to consider the full hazards such as cyclone; a case study described in
risk environment can result in maladaptation: heavy section 3-24 examines how tropical cyclone patterns,
concrete structures that protect against cyclone altered by climate change, affect losses in 15 Pacific
wind, for example, may be deadly in an earthquake. Island countries, assuming steady-state exposure.
Experience shows that the benefits of a multi-hazard Given the intensive data needs involved, there have
risk approach include improvements in land-use been few efforts to look at changing exposure and
planning, better response capacity, greater risk vulnerability, along with the resulting change in
awareness, and increased ability to set priorities for risk, in urban environments. While the contribution
mitigation actions. Such an approach also highlights of urbanization to greater exposure is widely
the importance of partnerships generally and of recognized, studies rarely consider how changes
multidisciplinary, multi-institutional, and multi- in urban construction practices affect building
sectoral collaborations in particular. Examples vulnerability—often for the worse. The case study
of this approach showcased in this publication of evolving seismic risk in Kathmandu offers an
include projects in Maldives (section 3-13), Morocco important example of this approach (section 3-25).
(section 3-17), and Guadeloupe and Naples (section The study shows that the incremental construction
3-20). of houses in Kathmandu, where stories are added
to buildings informally over time, has increased
Decision makers need to exercise particular caution
both exposure and vulnerability in the area. Using
where risks in food security and the agricultural
a single-scenario earthquake event, a reproduction
sector are concerned. Such risks should be
of the 8.1 magnitude Bihar earthquake of 1934, the
considered at all times alongside flood and drought
analysis finds that the potential number of buildings
analysis. Food security–related risks such as animal
sustaining heavy damage or collapse in this event
and plant pests and diseases are important for many
populations, yet they are not considered under the has increased from ~50,000 in 1990 to ~125,000 in
Hyogo Framework for Action. 2010, and that it may be as high as 240,000 by 2020
if action is not taken.
7. Keep abreast of evolving risk.
Considering global changes in hazard and exposure
Risk assessments must be dynamic because risks for flood offers some sobering statistics for the
themselves are always evolving. Assessments that future: “middle-of-the-road” socioeconomic changes
estimate evolving or future risk allow stakeholders and climate change could increase riverine flood
risk for between 100 million and 580 million people areas. Even small changes can become extremely
by 2050, depending on the climate scenario (see important during flood and storm surge events.
section 3-23). At a city level, changes in exposure
and flood hazard for Dhaka, Bangladesh, were 8. Understand, quantify, and
found likely to increase the average annual loss by communicate the uncertainties and
a factor of 20 to 40. Moreover, while both climate limitations of risk information.
change and socioeconomic development were
Once risk information is produced, its users must
found to contribute importantly to this increase in
be made aware of its limitations and uncertainties,
risk, the individual contribution of socioeconomic
which can arise from uncertainties in the exposure
development is greater than that of climate change.
data, in knowledge of the hazard, and in knowledge
Coastal regions are especially dynamic, and—in light of fragility and vulnerability functions. A failure
of future sea-level rise driven by local subsidence, to understand or consider these can lead to
the thermal expansion of the oceans, and melting flawed decision making and a potential increase
of continental ice—need special consideration. in disaster risk. A risk model can produce a very
Changes in sea level can be particularly important precise result—it may show, for example, that a
for relatively flat low-lying islands and coastlines, 1-in-100-year flood will affect 388,123 people—but
since a small change in sea level can affect huge in reality the accuracy of the model and input data
R E C O M M E N DAT I O N S
217
218
R E C O M M E N DAT I O N S
may provide only an order of magnitude estimate. assessments’ limitations. Several projects described
Similarly, sharply delineated flood zones on a hazard in this publication found that data limitations and
map do not adequately reflect the uncertainty assumptions made in the modelling process could
associated with the estimate and could lead to substantially change the end result:
decisions such as locating critical facilities just
• Multiple tsunami hazard maps were produced
outside the “flood line,” where the actual risk is the
in Padang, Indonesia, by different institutions,
same as if the facility was located inside the flood
each offering plausible information for decision
zone.
makers, and each based on different approaches,
If risk information is to be useful in making assumptions, and data (see box 3-5).
communities more resilient and better able to
manage risk, then the specialists who produce it • Depending on the choice of elevation data in
must do more to clearly and simply communicate modelling tsunami hazard, inundation levels
its uncertainties and limitations. Fortunately, some varied dramatically as a function of the digital
recent projects suggest that progress is being made elevation model used in the simulation (see box
in this regard: 2-4).
• In Kathmandu, assessment of damage to • Different seismic hazard results for ground
buildings as risk evolves over times includes a motion in Japan highlight the impact of the
range of uncertainty (section 3-25). choice of attenuation function (see box 2-10).
• In global risk models, the limitations for use These examples make clear the need for credible,
in national and subnational risk reduction are scientific, and transparent modelling of risk.
clearly articulated (sections 3-7, 3-8, and 3-23). Every risk analysis should be accompanied by
modelling metadata that articulate the data sets
• In Morocco, results of multi-hazard risk analysis and modelling parameters used so that anyone can
are communicated using a range of different recreate identical results. In other words, we need
approaches (section 3-17). to achieve an “academic level” of transparency. The
selection of modelling parameters also speaks to the
need for credible scientific and engineering inputs
9. Ensure that risk information is throughout the modelling process; in theory, anyone
credible and transparent. can run a risk model, but in reality, the absence of
necessary scientific and engineering training can
Risk information must be credible and transparent: produce results that are fundamentally inaccurate
scientifically and technically rigorous, open for and misleading.
review, and honest regarding its limitations and
uncertainties. 10. Encourage innovations in open
source software.
A risk assessment must be perceived as credible
for it to be worth acting upon. The best way to It is clear that immense progress has been made in
demonstrate credibility is to have transparent the last 5 to 10 years in creating new open source
data, models, and results open for review by hazard and risk modelling software. More than
independent, technically competent individuals. 80 open source software packages are currently
Equally important is the clear articulation of the available for flood, tsunami, cyclone (wind and
CHAPTER
04
surge), and earthquake, with at least 30 of these in and update code, and offer users easy access
wide use (see “Hazard and Risk Assessment Tools” to models.
in part 2). Moreover, significant progress has been
• Standard model outputs and data (e.g., event loss
made in improving open source geospatial tools,
such as QGIS and GeoNode, which are lowering the tables) should be made viewable at every stage
financial barriers to understanding risks at national of the analysis without significant increases in
and subnational levels. processing.
There is some tendency to assume that open source • Tools should have the capability of using custom
software may be less robust than commercial exposure data and hence of handling both static
packages, may be less user-friendly, and may not risk and dynamic risk.
offer technical support. But this assumption has
• Software should host a greater range of
little basis. Some of the most widely used packages,
vulnerability functions capable of calculating
such as InaSAFE and TCRM, provide interactive
vulnerability (susceptibility to damage or loss)
help, and others, such as the Deltares-developed
using either empirical methods (historical
packages, have impressive graphical user interfaces
trending of data) or analytical methods
that offer point-and-click capabilities. Available
(mathematical or mechanical approach).
software packages range from those that meet
These should cover both physical and social
the needs of entry-level users to those that are
vulnerability.
appropriate for advanced scientific and engineering
analysis. Some tools offer single hazard and risk • Risk should be calculable not only for a building
analysis—probabilistic and deterministic—and or building type, but also for a diverse portfolio
some, such as RiskScape and CAPRA, offer multi-
of buildings and infrastructure, or in terms of the
hazard capabilities.
total economic loss for a city or region.
Increasing the uptake of open source modelling
A great challenge for the next five years—one
tools is an important challenge that will need to met
that has arisen rapidly along with innovative
in coming years. Among specific goals in this area
software models—involves “fitness-for-purpose,”
are the following:
interoperability, transparency, and standards.
• Access to software with user-friendly interfaces, This challenge needs to be overcome in a way that
simple single-click installation, and tutorials on continues to catalyze innovation and yet also better
software use should be increased. supports risk model users. But it is an institutional
challenge, and not a technical one, and it can be met
• Licensing restrictions on how software may be
if model developers agree on minimum standards
used or altered should be easier to understand.
and build partnerships across institutions and
• Access to model source code—through wiki- hazard types.
type systems—should be increased in order to
provide improved transparency in how results
are calculated, allow for customization and Endnotes
optimization of code, enable production of better
93 Insurance Bureau of Canada, “Fighting Urban Flooding,”
code through multiple independent reviews, 2014, http://www.ibc.ca/en/Natural_Disasters/Municipal_
provide developers with an easy way to manage Risk_Assessment_Tool.asp.
R E C O M M E N DAT I O N S
219
220
Photo Credits
If not indicated otherwise, photos used in this publication
have been sourced from the following locations with
full rights:
World Bank Flickr Website
UN Flickr Website
DFAT Flickr Website
NASA Goddard MODIS Rapid Response Team
Government of Malawi
Global Earthquake Model
Joaquin Toro, World Bank Group
Francis Nkoko, World Bank Group
John Crowley, GFDRR
Emma Phillips, GFDRR
David Lallemant, Stanford University
Anne Sanquini, Stanford University
All images in this publication require permission for reuse.
Across the globe, a consensus is emerging on the central importance of risk information
in disaster risk management. When risks are quantified and the potential impacts of haz-
ards are anticipated, governments, communities, and individuals are able to make more
informed decisions.
This publication highlights some of the influential efforts—by technical specialists, insti-
tutions, and governments around the world—to create and communicate risk information
quickly and at low cost, to improve the quality and transparency of risk information, and
to enable more local engagement in the production of authoritative risk information than
ever before. Case studies spanning 40 countries and contributed by more than 50 institu-
tions showcase emerging best practices, demonstrate how risk assessments are being used
to inform disaster risk management and broader development, and highlight lessons learned
through these efforts. Taken as a group, these case studies evidence the need for contin-
ued investment in accurate and useful risk information and provide recommendations for
the future.
ABOUT GFDRR The Global Facility for Disaster Reduction and Recovery (GFDRR) helps
high-risk, low-income developing counties better understand and reduce their vulnerabilities
to natural hazards, and adapt to climate change. Working with over 300 national, community
level, and international partners GFDRR provides grant financing, on-the-ground technical
assistance helping mainstream disaster mitigation policies into country level strategies, and
thought leadership on disaster and climate resilience issues through a range of knowledge
sharing activities. GFDRR is managed by the World Bank and funded by 21 donor partners.
WWW.GFDRR.ORG