77368 Methodology Report Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index East Asia Infrastructure Sector Unit 26 September 2011 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Contents List of Tables iii List of Figures iii Acronyms and Abbreviations v Acknowledgements vi Executive Summary 1   1  Introduction 11  1.1  Project Purpose 11  1.2  Project Scope 12  1.3  Implementation Arrangements 13  2  Why a Multi-hazard City Risk Index? 14  2.1  Why Cities? 14  2.2  Why Multi-hazard? 17  2.3  Why An Index? 19  2.4  Why Another Index? 21  2.5  Where MHCRI Fits in Urban Risk Assessment 21  2.6  Methodological Options for the MHCRI 22  2.7  Information Collection for an Index-based Approach 26  3  Objectives of the MHCRI 28  3.1  Context: Natural Disasters and Climate Change 28  3.2  Users and Uses 29  3.3  Comparing Risks Between Cities 29  3.4  Comparing Risks to Cities Over Time 29  3.5  Comparing Risks Within Cities 29  3.6  Design Criteria for the MHCRI 29  3.7  What is Measured: MHCR Indices 31  4  The MHCRI Model 33  4.1  Scalar Parameters 33  4.2  How Risk is Calculated 35  4.3  Calculation Modules 35  5  Module 1: Inventory of Metropolitan Elements at Potential Risk 37  5.1  Purpose of the Module 37  5.2  People 37  5.3  Buildings 37  5.4  Infrastructure 39  5.5  Model Outputs 49  6  Module 2: Defining Hazard Indices 51  6.1  Purpose of the Module 51  6.2  Overall Approach 51  6.3  Hazard Model 52  6.4  Geophysical Hazards 53  6.5  Meteorological Hazards 53  6.6  Hydrological Hazards 54  6.7  Climate Change Effects 54  6.8  Model Outputs 54  i Methodology Report: Calculating Multi-hazard City Risk 7  Module 3: Calculating Exposure 55  7.1  Purpose of the Module 55  7.2  Calculation 55  7.3  Model Outputs 56  8  Module 4: Assessing Vulnerabilities 57  8.1  Purpose of the Module 57  8.2  The Vulnerability Model 57  8.3  Physical Susceptibility 58  8.4  Fragility 59  8.5  Resilience 60  8.6  Weighting Vulnerabilities 65  8.7  Model Outputs 68  9  Module 5: Calculating Risks 69  9.1  Indicators of Risk 69  9.2  Risk Calculation 69  9.3  Range of Calculable Risk Indices 70  9.4  Risk Reduction Options 71  10  Findings from the Three Pilot Cities 73  10.1  Metropolitan Elements 73  10.2  Natural Hazards 75  10.3  Vulnerability Indices 78  10.4  Multi-hazard City Risk Indices 80  10.5  Sub-metropolitan MHCRI 86  10.6  Exploring Risk Management Options 89  11  Further Testing and Refinement 93  11.1  Lessons Learned from the Pilot Testing 93  11.2  Improving the MHCRI Model 94  11.3  Criteria for Further Testing 95  Appendix A: Terms of Reference 97 Appendix B: Comparison of MHCRI with Other Risk Indices 103 Appendix C: Hazard Indices and Climate Change 107 Appendix D: Spreadsheet Structure of the Multihazard City Risk Model 127 Appendix E: Index Results from Pilot Cities 129 Appendix F: Project Team 135 References ii Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project List of Tables Table 1:  Three Broad Approaches to Collecting Information for the MHCRI 28  Table 2:  Minimum Data for MHCRI calculation based on Three Pilot Cities 42  Table 3:  Vulnerability Weights: 2010 66  Table 4:  Vulnerability Weights: 2030 67  Table 5:  Quantities of Metropolitan Elements 75  Table 6:  Statistics of Typhoon Annual Maximum Wind Speed (km/h), within Two Degrees of City (1970-2010) using JTWC Dataset 109  Table 7:  Manila Typhoon Return Period of Annual Maximum Wind Speeds 110  Table 8:  Saffir-Simpson Hurricane Scale as compared to H Scale 111  Table 9:  Summary of HI Typhoon Calculations for Bangkok, Manila and Ningbo 111  Table 10:  Summary statistics of maximum 3-month precipitation (mm) during monsoon 113  Table 11:  Return Periods of 3-month Monsoon Precipitation Totals for Manila and Bangkok 114  Table 12:  Summary of HI Monsoon Calculations for Bangkok and Manila based upon 3-month precipitation totals 114  Table 13:  Summary of the Climate Change Effect on HI Calculations 115  Table 14:  HI calculations for drought based upon PDSI 119  Table 15:  Storm Surge HI Calculations 122  List of Figures Figure 1:  Metropolitan Regions in Southeast Asia, 2000-2010 16  Figure 2:  Metropolitan Regions in Northeast Asia, 2000-2010 16  Figure 3:  Metropolitan Regions’ Exposure to Four Types of Natural Hazard Events, ~ 1980 – 2000 18  Figure 4:  Metropolitan Regions by Population in 2010 and Exposure to Four Types of Natural Hazard Events, ~ 1980 – 2000 18  Figure 5:  Levels of the Urban Risk Assessment (MHCRI in red) 22  Figure 6:  Elements of a Catastrophe Model 24  Figure 7:  Temporal and Spatial Scales Relevant to Preparation of the MHCRI 35  Figure 8:  Overall Process for Calculating MHCRI 36  Figure 9:  Data from Existing Building Inventories up-scaled to 500 m cells 39  Figure 10:  Data Extraction Using Areal Interpolation 39  Figure 11:  Gross Area of all Buildings: Bangkok, 2009 (m2) 50  Figure 12:  Manila – People to Flood: Total Vulnerability Indices, 2030 68  Figure 13:  Example of MHCRI Calculation (People) for a 500 m x 500 m cell with Population of 5,000 in 2010 70  Figure 14:  Overview of Pilot Cities 74  Figure 15:  Manila’s Natural Hazard Indices 76  Figure 16:  Ningbo’s Natural Hazard Indices 77  Figure 17:  Bangkok’s Natural Hazard Indices 77  Figure 18:  Manila’s Vulnerability Indices 78  Figure 19:  Ningbo’s Vulnerability Indices 79  Figure 20:  Manila Households Susceptible to 30-year Flood, 2010 79  Figure 21:  Bangkok’s Vulnerability Indices 80  Figure 22:  Bangkok’s Additional Building Area in Extended Flood Inundation Zone by 2030 (blue boxes) 80  Figure 23:  Multi-hazard City Risk Indices – People 81  Figure 24:  Multi-hazard City Risk Indices – Capital Stock 82  Figure 25:  Multi-hazard City Risk – Buildings 83  Figure 26:  Multi-hazard City Risk – Infrastructure (rv$) 83  Figure 27:  Risks to Manila’s People, 2010 (30-year return periods) 84  Figure 28:  Risks to Manila’s Capital Stock, 2010 (30-year return periods) 84  Figure 29:  Risks to Manila’s Buildings, 2010 (30-year return periods) 85  Figure 30:  Risks to Manila’s Infrastructure, 2010 (30-year return periods) 85  Figure 31:  Manila - Elements’ Share of Multi-hazard Risk to Capital Stock 86  iii Methodology Report: Calculating Multi-hazard City Risk Figure 32:  Manila - Sub-elements’ Share of Multi-hazard Risk to Capital Stock 86  Figure 33:  Manila - Building Types’ Share of Multi-hazard Risk to Buildings 86  Figure 34:  Manila - Infrastructure Types’ Share of Multi-hazard Risk to Infrastructure 86  Figure 35:  Distribution of Multi-Hazard City Risk for People among Cities and Municipalities in Metro Manila, 2010 87  Figure 36:  Barangays in the top decile of MHCRI to People, 2010 87  Figure 37:  Manila - Change in MHCRI: People, 2010-2030 88  Figure 38:  Ningbo - Change in MHCRI: Capital Stock, 2010 – 2030 88  Figure 39:  Bangkok - Change in 30-yr MHCRI for Capital Stock, 2010 – 2030 89  Figure 40:  Six Areas in Bangkok Requiring Detailed Assessment of Multi-hazard Risks 89  Figure 41:  Vulnerability Index in 2030 with 30% Reduction in Depth of 30-year Floods (Metro Manila) 90  Figure 42:  MHCRI for People in 2030 with 30% Reduction in Depth of 30-year Floods 91  Figure 43:  MHCRI for Capital Stock in 2030 with 30% Reduction in Depth of 30-year Floods 91  Figure 44:  Tropical Cyclone Tracks from the beginning of records to September 2006 108  Figure 45:  Percentiles of Typhoon Maximum Wind Speeds for Manila, Bangkok and Ningbo 109  Figure 46:  Asian Monsoon Wind Patterns 112  Figure 47:  Percentiles of 3-month Monsoon Precipitation at Manila and Bangkok 113  Figure 48:  Drought frequency (events/yr 1980-2000) 117  Figure 49:  Time series of monthly PDSI values (1970-2005) 118  Figure 50:  Histograms of monthly PDSI (1970-2005) 118  Figure 51:  Figure 8: Percentiles of the PDSI for Bangkok, Manila and Ningbo 119  Figure 52:  East Asia PDSI (2030-2039) (adapted from Dai, 2010) 120  Figure 53:  Drought in East Asia for an ensemble of models for doubled carbon dioxide 120  Figure 54:  Storm Surge Height as a Function of Hurricane Wind Speed (Franck, 2009) 121  Figure 55:  Effect of Storm Motion on Surface Wind Speeds 121  Figure 56:  Sea Level Changes Since the Last Glacial Maximum 122  Figure 57:  Projected sea-level rise for the 21st century 123  Figure 58:  Type I Gumbel distribution for the Maximum Case 125  iv Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Acronyms and Abbreviations ASTER: Advanced Spaceborne Thermal Emission and Reflection Radiometer AUSAID: Australian Agency for International Development CC: climate change DEM: digital elevation model DR: drought DRM: disaster risk management DRR: disaster risk reduction EM-DAT: International Disaster Database (Centre for Research on the Epidemiology of Disasters (CRED)) EQ: earthquake GCM: Global Circulation Model GIS geographic information system HI: Hazard Index IPCC: Intergovernmental Panel on Climate Change LIDAR: Light Detection And Ranging (remote sensing technology) LRAP Local Resilience Action Plan LS: landslide ME: Metropolitan Element (People, Capital Stock comprised of Buildings and Infrastructure) MHCRI: Multi-hazard City Risk Index (MHCRI=ME x HI x VI) MSE: Metropolitan Sub-element NAPA: National Adaptation Programmes of Action NOAA: National Oceanic and Atmospheric Administration, USA rv: replacement value SS: storm surge TY: typhoon VHR: Very High Resolution (satellite imagery) VI: Vulnerability Index v Methodology Report: Calculating Multi-hazard City Risk Acknowledgements This document is one in a set of five reports produced under the Multi-Hazard City Risk Index Prototype component of the World Bank project, Resilient Cities: Decision Support Tools for Climate and Disaster Risk Reduction in East Asian Cities (P121572). The initiative was led by Fatima Shah, Senior Urban Economist, under the management of Vijay Jagannathan, Sector Manager, and John Roome, Sector Director, at the World Bank’s East Asia Sustainable Development Department. Technical leadership was provided by Edward Leman at Chreod Ltd, with contributions and feedback from several international experts, both within and outside the World Bank – a complete list is available in Appendix G. The Team gratefully acknowledges the support and partnership from the city governments of Metro Manila, Bangkok, and Ningbo. In Manila, the team was fortunate to receive support from the Metropolitan Manila Development Authority, including constructive comments received in an interim review workshop with MMDA in July 2011, at which government representatives of Manila City, Makati City, and Taguig City, and staff from the Center of Disaster Preparedness, the Manila Municipal Disaster Risk Reduction and Management Office, EMI, and PHILVOCS participated; In Bangkok, the team was fortunate to receive support from the Bangkok Metropolitan Administration, including constructive feedback during and following a workshop with BMA in February 2011; In Ningbo, the team received support from the municipal government’s Development and Reform Commission, Construction Commission, Urban Planning Bureau, Transport Bureau, and Drainage Department. This first phase of the Multi-Hazard City Risk Index Prototype received financial support from the AUSAID- funded project, Knowledge Partnerships: Cities as Engines of Growth (P115851-TF093729); the Korean Trust Fund project Good Practices in Urban Environment Management – Cases from Korea and Beyond (P120210-TF096102); GFDRR for applications in Metro Manila (TF091752) and Ningbo (TF091752); and, under the Clean Energy Investment Fund, DFID support for application in Bangkok (TF090444). About GFDRR The Global Facility for Disaster Reduction and Recovery (GFDRR) is a partnership of 38 countries and 7 international organization committed to helping disaster-prone developing countries and regions reduce their vulnerability to natural hazards and adapt to climate change. GFDRR promotes technical and financial assistance to high-risk low- and middle-income countries based on a business model of ex-ante support to mainstream disaster risk reduction in national development strategies and investments, and ex-post disaster assistance for sustainable recovery. As part of its mandate, GFDRR promotes global knowledge and good practices, supports initiatives for enhanced global and regional cooperation, and promotes greater South-South cooperation in disaster risk reduction. www.gfdrr.org Supported by the Australian Government, AusAID Disclaimer The views expressed in this publication are those of the authors and not necessarily those of the Australian Agency for International Development (AusAID). Funded by the Korean Trust Fund The Korea-World Bank Trust Fund for Poverty Reduction and Socio- Economic Development (KTF) is a new partnership to finance programs and activities supported by the World Bank. This Trust vi Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Fund allows the recipients of the grants and the Bank to finance technical assistance work for issues of emerging importance to support development in East Asia and Pacific countries and in the region. Funded under the Trust Fund for Supporting Adaptation to Climate Change vii Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Executive Summary Project Purpose Nearly half of East Asia’s population lives in cities and the region is urbanizing so rapidly that built-up areas are projected to increase faster here than in any other region in the next 20 years. Still, more than half of slum dwellers around the world live in East Asia. These are the people most vulnerable to disaster impacts. Given that Asia accounted for more than a third of the number of all reported disasters in 2010, and that natural disasters have quadrupled in the region during the past 20 years – the fastest rate of increase of any region in the world – managing urban growth for resilience is increasingly important. And yet, many cities do not have masterplans to guide their projected growth, let alone tools for risk- sensitive land use planning. The impetus for action will come from accessible tools that allow local policymakers to first minimize risk today and then think about addressing future risk. The World Bank’s Climate Resilient Cities: A Primer on Reducing Vulnerabilities to Disasters (2009) provided basic guidance to governments on the concepts and interactions between climate change adaptation, disaster risk reduction (DRR), and urban vulnerabilities; and provides examples of what cities around the world are doing to proactively increase their resilience. Based on the Primer, technical assistance was provided to several cities in the region in developing Local Resilience Action Plans (LRAPs) - A Workbook on Planning for Urban Resilience in the Face of Disasters: Adapting Experiences from Vietnam's Cities for Other Cities (forthcoming) encapsulates this learning and will soon be released by the World Bank. In order to complement these guides, particularly on activities related to hazard mapping, identification of priority actions based on how they would impact the city's overall risk, as well as monitoring the implementation of identified actions in the LRAPs, the World Bank's East Asia Infrastructure Sector Unit has embarked on an exercise to develop a prototype Multi-Hazard City Risk Index (MHCRI). There is a continuing need for decision-support tools that address currrent multiple risks to cities, and future risks arising from climate change. A city-based index of risk from multiple hazards is a potentially useful decision-support tool that could provide a standardized metric to capture and aggregate risk at the city level – and across cities. The purpose of this project was to develop a methodology for measuring multiple risks at the city level from natural hazards with both rapid and gradual onset. The methodology was tested and refined in applications of the MHCRI model in Bangkok, Manila1, and Ningbo. Considerable work has been conducted around the world to assess risk to individual cities from individual hazards. Many cities have modeled seismic or flood risks, and advanced tools are available to support such efforts. However, assessing the probable cumulative impacts of multiple hazards has not entered the mainstream of urban management practice 1 “Manila� is used in this report to refer to Metro Manila, an administrative amalgam of 17 cities and municipalities. One of these 17 is the City of Manila which accounts for 15% of the population of Metro Manila; when referred to in the project reports, the City is identified as such. 1 Methodology Report: Calculating Multi-hazard City Risk around the world. A better understanding of the impacts of multiple hazards – including future hazards arising from climate change – can support more focused engagement and policy actions by various stakeholders, ranging from multilateral agencies to local governments. Understanding multi-hazard risk could help to prioritize interventions; it could also form the basis for exploring institutional and fiscal adjustments that improve stakeholders’ capacities to better manage overall risk at the city scale. As a decision-support tool, the MHCRI could inform the need for more detailed exploration of:  where to invest in disaster risk management (DRM) and climate change adaptation (CCA) to minimize risks;  which sectors to invest in;  where to concentrate policy changes, particularly to limit urban development in vulnerable areas; and  which policies need to be changed (added, removed, or adjusted). This methodology report describes the underlying risk components, including a description of the model's required inputs related to metropolitan elements at-risk (these include sub-classes of building, infrastructure, and population), hazard sub-indices for 13 hazard types (based on event of record, frequency, intensity, and area affected for 2- year and 30-year events), and 30 parameters of physical, socioeconomic, and institutional vulnerability. In addition to this report, three city reports present results that can be used by a variety of users from those interested in the overall city-wide risk from all natural hazards to those interested in more disaggregated information. Information is available on risk from one particular hazard to all population or physical assets (buildings and infrastructure) in the city as well as to one particular population group - e.g. low-income residents - or one particular infrastructure sector - e.g. transport - or to a specific building type or a specific geographic area of the city. The city reports point to areas that may require further detailed risk assessment or institutional and policy issues of particular concern. A technical manual is also available for other cities to use in preparing their own Index for planning purposes. A web portal is being developed that would allow planners in the pilot cities to customize reports based on the data gathered and analyzed. Methodological Considerations Unlike the insurance industry, disaster risk reduction and climate change adaptation goals are not principally about protecting against the risk of financial loss. They are primarily directed towards protecting lives and core assets required to sustain livelihoods. These are quantified in the MHCRI with common denominators that can be compared across cities. A purely deductive approach to preparing the MHCRI would require standardized historical data on each hazard for all cities that cover the same time periods, have been measured the same way, and at the same levels of accuracy and reliability. Availability of such data is highly unlikely 2 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project in many developing countries (and probably many developed countries). An inductive approach was therefore followed in the calculation of the MHCRI. Given the need for a high-level, strategic decision-support tool that can be readily applied across many cities at relatively low cost, the MHCRI is an ‘Index’ and not a monetized loss model. The focus of the MHCRI is on assessing risks to: people; residents’ dwellings (their principal assets); firm’s means of production (and places of residents’ employment); social services (especially health and education); and infrastructure services. These are encapsulated in the MHCRI in ‘Metropolitan Elements’ of: People, Buildings (residential, employment-use buildings (a proxy for means of production) and social service buildings (a proxy for social service delivery capacity)), and Infrastructure facilities. Replacement values of buildings, and infrastructure provide the basis for aggregating a city’s total capital stock at risk using a common metric. The MHCRI does not calculate potential damages or losses: replacement values are only used as a common measure to assess risks to buildings and infrastructure which can be aggregated to define risks to a city’s total capital stock. Risks to people cannot be compared with risks to buildings, infrastructure and captial stock since ‘people’ has no common metric (i.e. there are no replacement costs for people in the MHCRI model). Structure of the MHCRI Model The overall MHCRI is constructed through five modules (Fig. 1). Module # 1: Metropolitan Elements is the creation of baseline inventories, comparable among cities, of quantitative and qualitative characteristics of metropolitan elements at potential risk: residents, building structures, and infrastructure. These baselines are prepared for two timescales: 2010 (or most recent base year for which credible data are available), and 2030, based on scenario development2. They are used to identify and quantify elements that are potentially exposed to natural hazards, and to inform vulnerability assessments. The MHCRI is constructed using 500 m x 500 m polygon cells as the spatial unit for data capture and analysis using Geographic Information Systems (GIS). These 500 m cells comprise a mesh over the entire city enabling cell values to be aggregated to the city or district levels. In their disaggregated form, 500 m cells provide for a high degree of granularity in the analysis of sub-metropolitan risk. The spatial precision that is possible with the 500 m cell depends entirely on the data that is input to the cells. In the three pilot cities, data on metropolitan elements at risk were available for people, buildings, and infrastructure at a very fine spatial scale; these were upscaled to 500 m cells. However, some data are only available at the municipal or district scales; these are downscaled to the cell level. The result is that the final mapping of individual and multiple risks is only indicative of the spatial distribution of risk. Until all data are available at or below the 500 m cell 2 MHCRI is modeled to a maximum of 20 years in the future. Beyond that period, projections of the size, characteristics and spatial distribution of metropolitan development become too unreliable. 3 Methodology Report: Calculating Multi-hazard City Risk level, the risk mapping produced from the MHCRI model should be viewed as a way to identify areas that require more detailed risk assessment using more rigorous methods, and not as a planning tool. The outputs from Module # 1 are quantities, replacement values (for capital stock) and locations of metropolitan elements at potential risk. Based on testing in the three pilot cities, a minimum data set on Metropolitan Elements has been established for the MHCRI. Module # 2: Hazard Indices is the calculation of indices for three major types of natural hazards: geophysical, meteorological, and hydrological. The Hazard Index (HI) is comprised of three parts: 1) a spatial scale to account for some hazards, such as landslides and storm surge, which only affect part of a metropolitan area; 2) a magnitude parameter that selects the size of hazard being analyzed; and 3) a frequency parameter that describes how often the hazard occurs for the selected magnitude. These three components are multiplied together to form HI. They are calculated for 30-year events and for 2-year events, the latter to enable assessment of risks from lower-magnitude but frequent risks facing cities3. The possible outputs from Module # 2 are Hazard Indices for: earthquakes, tsunamis, volcanic eruptions, landslides, sudden subsidence4, typhoons, severe thunderstorms, tornados, monsoons, extreme temperatures, drought, wildfires, and storm surges5. Hazard Indices are calculated only for those hazards facing the particular city. Module # 3: Exposure Indices is the calculation of current and future (2030) probable exposure of metropolitan elements to probable hazards. This is an automatic calculation in the Model: the quantity of whichever element is being assessed (e.g. people, derived from Module # 1) is multiplied by each individual Hazard Index calculated in Module # 2. Exposure values remain disaggregated by individual hazard at this stage in the model since vulnerabilities (calculated in Module # 4) are hazard- specific and heavily influence actual risk. Module # 4: Vulnerability Indices consists of assessing vulnerabilities according to impacts from exposure to probable natural hazards. Vulnerabilities are at the core of city risk. Vulnerabilities – not natural hazards – cause disasters. If physical susceptibilities are reduced, fragilities minimized, and resilience improved, impacts of hazardous events will be far less than if vulnerabilities are left unchecked. Importantly, while governments cannot influence the frequency and magnitude of natural hazards, they can do much to reduce vulnerabilities. The MHCRI assesses vulnerability through four parameters of physical susceptibility, eight parameters of socio-economic fragility, and eighteen parameters measuring resilience of institutions and risk management processes. Recognizing that vulnerability is specific to types of hazards, 3 Risks from hazards with 2-year return periods are reported separately from 30-year return periods as Appendices in the City Profile Reports. 4 Gradual subsidence is taken into account in the calculation of flood maps the values from which are entered into the MHCRI Model as heavily-weighted Vulnerability Indicators. 5 Sea level rise is not treated as a separate ‘hazard’ but as a major determining factor of future floods; these have been taken into account in the flood mapping for Manila and Bangkok. We cannot confirm if SLR was included in the partial Ningbo flood map. 4 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project separate Vulnerability Indices are calculated for each hazard for each 500 m cell. They can be averaged to the municipal level. In Module # 4, users can also clearly identify the quantity and location of metropolitan elements located in areas susceptible to individual hazards. This enables, for example, the identification of number of households living in areas susceptible to 30-year floods, and the quantities and replacement values of capital stock in these areas. Module # 5: Risks Calculation brings all of the preceding steps together to compile the MHCRI. Risk indices – either multi-hazard or for individual hazards – are calculated in Module # 5 for five risk indicators: people at risk, gross floor area of buildings at risk, replacement values of buildings at risk, replacement values of infrastructure at risk, replacement values of total capital stock. Risk values for buildings and infrastructure are summed to calculate risks to total capital stock. Outputs are calculable at both the Metropolitan-wide and sub-metropolitan scales, depending on user interests. Sub-metropolitan outputs can be mapped on GIS using a 500 m x 500 m cell as the basic spatial unit of analysis, subject to the caveats on data precision noted earlier. Based on experience with the availability of data in the three pilot cities, the minimum data required to calculate MHCRI have been established. This baseline provides for analysis of risk to people and and an extensive array of capital stock, comprised of buildings and infrastructure. Figure 1: Structure of the MHCRI Model What the MHCRI Does Not (Yet) Do At this initial stage in development of the MHCRI, it is important to understand what the model does not yet do. It does not incorporate loss or damage functions since the costs of doing so in multiple cities with varying types of buildings and infrastructure and hazard profiles are far 5 Methodology Report: Calculating Multi-hazard City Risk beyond what is envisioned for the MHCRI. So far, the model only assesses hazard events with 30-year and 2-year return periods. The model only assesses single events, not coupled events such as earthquakes and resulting fires. Only natural hazards are modeled; anthropogenic hazards that accelerate or amplify the effects of natural hazards are not assessed (e.g. poor maintenance of upstream water diversion infrastructure). Work on the next phase of the MHCRI will focus on extending the model’s functionality to address most, if not all, of these issues. Application in Bangkok, Manila, and Ningbo The MHCRI has been tested in the three pilot cities of Metro Manila, Bangkok, and Ningbo. Although all three are coastal cities, there are marked differences among them in: population size; capital stock; in the range, frequency and intensity of natural hazards; and in vulnerabilities. Ningbo faces four natural hazards, three of which could be calculated in the pilot study6. Bangkok and Manila face five hazards. There are major differences in Hazard Indices (HI) between the three cities with Manila generally having far larger values. Manila’s 30-yr HI for typhoons is 2.4 times Ningbo’s and 9 times Bangkok’s; for monsoons, Manila’s HI is 1.4 times Ningbo’s. For drought, Manila’s HI is 1.4 times Ningbo’s and the same as Bangkok’s. For 2-year return periods, Manila’s typhoon index is twice that of Ningbo (typhoon events in Bangkok are rare, well beyond 2 years). However, Manila’s 2-year monsoon index is only slightly higher than Bangkok’s (1.06 times). Manila is considerably more vulnerable to flooding from typhoons and monsoons than the other cities. For flood, the People Vulnerability Index in Manila (28.5) is 1.9 times Bangkok’s (15.1) and 3 times Ningbo’s (9.2). This is due to the large and much more densely inhabited area at flood risk in Manila, the city’s far larger number of poor residents, and the location of their residences in areas susceptible to flooding. Bangkok’s buildings are also quite vulnerable to floods, and will become more vulnerable by 2030 due to an increase in the area susceptible to flood caused by subsidence and climate change effects on precipitation and sea level rise. Risks were calculated for four risk conditions: current risk from 30-year events; current risk from 2-year events; risk from 30-year events in 2030; and risk from 2-year events in 2030. Multi-hazard City Risk Indices were calculated for People (Fig. 2), Capital Stock (Fig. 3), and its sub-elements of Buildings and Infrastructure. Given Manila’s larger population (ME), higher Hazard Indices (HI) and higher Vulnerability Indices (VI), it is not surprising that its current 30-yr MHCRI for People is 9 times that of Bangkok’s. MHCRI to capital stock in Manila is 9.5 times Bangkok’s and 12 times higher than Ningbo’s. Manila 6 The Government of China considers historical data on climate and flooding to be State secrets. Only typhoon, storm surge and drought risk were analyzed in Ningbo, using global and regional data sets. The Ningbo MHCRI is therefore a partial index that makes comparisons with Manila and Bangkok tenuous. 6 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project is clearly a very high risk city. Risk values for each 500 m cell were aggregated to the District scale to inform on their relative levels of risk. For example, in Manila local government units’ share of Manila’s MHCRI were calculated, enabling governments to better understand the extent of risk within their jurisdictions. Barangays at highest risk were also identified. 6.3 Bangkok 5.9 1.6 Ningbo* 1.0 70.3 Metro Manila 55.2 0 10 20 30 40 50 60 70 80 * partial risk analysis due to unavailability Multi-hazard City Risk Index (People) of meteorological data; see Ningbo City Profile Report 2030, 30 yr 2010, 30 yr Figure 2: MHCRI for People, 2010 and 2030 32.2 Bangkok 18.2 25.3 Ningbo* 14.1 246.9 Metro Manila 172.4 0 50 100 150 200 250 300 Multi-hazard City Risk Index: Capital Stock * partial risk analysis due to unavailability of meteorological data; see Ningbo City Profile Report 2030, 30 yr 2010, 30 yr Figure 3: MHCRI for Capital Stock, 2010 and 2030 7 Methodology Report: Calculating Multi-hazard City Risk With the granularity of analysis possible with the 500 m cell, detailed mapping of MHCRI was conducted which allows for the identification of specific areas at highest level of risk, and for the identification of areas in which risk will increase the most between 2010 and 2030 (Figs. 4, 5). The resulting maps are meant as a guide for local governments to identify areas requiring much more detailed risk analysis than is possible with the MHCRI; the maps are not meant to be used as an investment planning tool on their own. Figure 4: Six areas requiring more detailed Multi-hazard Risk Assessment in Bangkok Figure 5: Barangays in Metro Manila With 500 m Cells in the top Decile of 8 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project MHCRI for People, 2010 Next Steps Key lessons learned from the pilot city applications are outlined in Chapter 11. While the minimum required data is sufficient for calculating MHCRI values that are comparable between cities, greater detail provides local governments with information that enables them to dig deeper into sector and sub-sector risk. Now that the parameters for the MHCRI model have been determined, future applications can be tied to local governments’ willingness to provide specific and recent data on metropolitan elements, meteorological trends and conditions, and physical susceptibility mapping (e.g. flood mapping). Testing in the pilot cities has shown where improvements to the MHCRI model are needed: 1) integrating a loss function so that 30-year and 2- year risk indices can be compared; 2) alternately, defining critical thresholds beyond which a hazard event causes injuries or losses of life and to property that are considered significant to local communities; 3) enabling rapid sensitivity analysis in the model to assess implications of changes to quantities of metropolitan elements and vulnerability parameters on multi-hazard city risk; 4) incorporating longer return periods; and 5) exploring risk coupling. These should be addressed in the next phase of methodology development. Pilot testing should be extended to other cities to further explore the MHCRI model’s limits and utility, and to further refine the model. Selection of future pilot cities can be guided by three decision criteria:  expanding the representativeness of cities in terms of the size and characteristics of metropolitan elements (e.g. population size, scope and quality of infrastructure);  expanding the range of natural hazards that cities face;  expanding the range of vulnerability conditions in cities. The third criterion – vulnerability differences – would suggest exploring cities likely to have high or low vulnerabilities. This selection filter would yield the greatest benefits for model development as it would serve to further test the vulnerability parameters and measurement approaches. Cities with low vulnerabilities in EAP could be Singapore, Hongkong and Seoul. Beyond EAP, there could be merit in selecting more advanced cities in the US (e.g. New York, Los Angeles) and Europe (e.g. London, Amsterdam). Cities with higher vulnerabilities in EAP could be selected cities in Indonesia (that also face a broader range of hazards, including volcanoes), the Philippines (e.g. Davao), Cambodia, and Vietnam. Beyond EAP, cities in MENA (e.g. Cairo, Alexandria, Sana’a) could be considered as well as in South Asia (e.g. Chittagong, Dhaka, and a range of cities in India). The MHCRI is in its infancy and it is expected that, in collaboration with city governments and other partners, the methodology will be further refined in subsequent phases of engagement so that the underlying sources of risk can be better understood by planners and can inform appropriate actions to mitigate these risks. 9 Methodology Report: Calculating Multi-hazard City Risk 10 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 1 Introduction 1.1 Project Purpose Nearly half of East Asia’s population lives in cities and the region is urbanizing so rapidly that built-up areas are projected to increase faster here than in any other region in the next 20 years. Still, more than half of slum dwellers around the world live in East Asia. These are the people most vulnerable to disaster impacts. Given that Asia accounted for more than a third of the number of all reported disasters in 2010, and that natural disasters have quadrupled in the region during the past 20 years – the fastest rate of increase of any region in the world – managing urban growth for resilience is increasingly important. And yet, many cities do not have masterplans to guide their projected growth, let alone tools for risk- sensitive land use planning. The impetus for action will come from accessible tools that allow local policymakers to first minimize risk today and then think about addressing future risk. Despite the many international and national initiatives addressing disaster risk, there is a continuing need for decision-support tools that address currrent multiple risks to cities, and future risks arising from climate change. A city-based index of risk from multiple hazards (Multi- Hazard City Risk Index, MHCRI) is a potentially useful decision-support tool that could provide a standardized metric to capture and aggregate risk at the city level – and across cities. The MHCRI could establish a baseline of multiple risks and provide one standard for the monitoring of a city’s performance in managing risks over time and relative to other cities. The MHCRI could be a tool for identifying sectors and locations requiring more intensive analysis. The purpose of this current project was to develop a methodology for measuring multiple risks at the city level from natural hazards with both rapid and gradual onset. This multi-hazard initiative is being conducted in the context of an integrated effort by the East Asia Region of the World Bank to improve cities’ resilience to natural disasters and the expected impacts of climate change (Box 1). The Multi-hazard City Risk Index (MHCRI) is the first effort by the World Bank to integrate Disaster Risk Reduction (DRR) and Climate Change (CC) parameters into an overall index reflecting the full gamut of natural hazards facing cities over time. While indices have been developed at the country and city scales for individual hazards, a multi-hazard index has not been done elsewhere at the city scale due to the methodological challenges of: 1) incorporating temporal differences between disaster- causing hazards – which manifest rapidly – and CC-related hazards which have a gradual onset; 2) estimating how CC will affect natural hazards over time; and 3) focusing at the city scale where long-term hazards can only be modeled with regional climate models that are still rather imprecise. 11 Methodology Report: Calculating Multi-hazard City Risk Box 1: City Resilience Initiatives at the World Bank The World Bank has produced several tools at the intersection of disaster risk management, climate change, and planning in urban areas. One such tool is the Climate Resilient Cities: A Primer on Reducing Vulnerabilities to Disasters (February 2009), which provides guidance to local governments in East Asia on the concepts of climate change and disaster risk reduction; how climate change consequences contribute to urban vulnerabilities; and what is being done by city governments around the world to actively engage in capacity building and capital investment programs for building resilience. Following the launch of the Primer, six city governments – in Vietnam (Hanoi, Can Tho, and Dong Hoi), Indonesia (Yogyakarta), Philippines (Iloilo), and China (Ningbo) – have received technical assistance from the World Bank to apply the broad concepts from the Primer to their local contexts and produce Local Resilience Action Plans (LRAPs). Experience with the pilot program highlights the need for such localized tools and strategies for managing climate and disaster risks. At present, much of the disaster mitigation and climate adaptation planning and allocations take place at the national level (e.g. based on SNAPs, NAPAs, and national policies), while much of the action is inherently local. To help fill this gap, the ongoing technical assistance and city partnerships that result in LRAPs is intended to provide a tool for investment planning at the urban level. The LRAP reflects a risk assessment based on current hazards the city faces and potential impacts of future climate changes in the context of urban expansion; an institutional and policy analysis to determine whether plans are in place to mitigate some of these risks; various options to mitigate risks that remain unaddressed; and results in a set of specific prioritized structural investments and nonstructural measures, with rough cost approximations and timelines, that the city would like to undertake to increase its resilience. The World Bank is now preparing a Workbook on Developing LRAPs that captures the methodology and experiences from the initial pilot cities so that other cities can undertake the process on their own. In addition to the MHCRI, the Bank has also begun to explore approaches for managing integrated disaster and climate risk in cities with a spatial focus. In 2010, it outlined an initial framework and approach for assessing urban risks in Understanding Urban Risk: An Approach for Assessing Disaster and Climate Risk in Cities. 1.2 Project Scope The World Bank’s Climate Resilient Cities: A Primer on Reducing Vulnerabilities to Disasters (2009) provided basic guidance to governments on the concepts and interactions between climate change adaptation, disaster risk reduction (DRR), and urban vulnerabilities; and provides examples of what cities around the world are doing to proactively increase their resilience. Based on the Primer, technical assistance was provided to several cities in the region in developing Local Resilience Action Plans (LRAPs) - A Workbook on Planning for Urban Resilience in the Face of Disasters: Adapting Experiences from Vietnam's Cities for Other Cities (forthcoming) encapsulates this learning and will be released by the World Bank in a few months. In order to complement these guides, particularly on activities related to hazard mapping, identification of priority actions based on how they would impact the city's overall risk, as well as monitoring the implementation of identified actions in the LARPs, the World Bank's East Asia Infrastructure Sector Unit has embarked on an exercise to develop a prototype Multi- 12 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Hazard City Risk Index (MHCRI). The objective in developing a prototype Multi-Hazard City Risk Index (MHCRI) is to provide a standardized and low-cost metric to capture and aggregate risk from multiple hazards at the city level, and help establish a baseline to measure performance over time and relative to peers. In its first phase of development, the MHCRI methodology was piloted in three cities - Metro Manila, Bangkok, and Ningbo, with the objective of investigating whether it could provide a useful communications tool to policymakers and affected communities on the impacts extreme weather, seismic and climate change events have on their lives and livelihood. It is envisaged that this information will generate collective action to mitigate the identified risks through policy innovations, improved physical planning and appropriate investments. This methodology report describes the underlying risk components, including a description of the model's required inputs related to metropolitan elements at-risk (these include sub-classes of building, infrastructure, and population), hazard sub-indices for 13 hazard types (based on event of record, frequency, intensity, and area affected for 2- year and 30-year events), and 30 parameters of physical, socioeconomic, and institutional vulnerability. In addition to this report, three city reports present results that can be used by a variety of users from those interested in the overall city-wide risk from all natural hazards to those interested in more disaggregated information. Information is available on risk from one particular hazard to all population or physical assets (buildings and infrastructure) in the city as well as to one particular population group - e.g. low-income residents - or one particular infrastructure sector - e.g. transport - or to specific building type or a specific geographic area of the city. The city reports point to areas that may require further detailed risk assessment or institutional and policy issues of particular concern. A technical manual is also available for other cities to use in preparing their own Index for planning purposes. A web portal is being developed that would allow planners in the pilot cities to customize reports based on the data gathered and analyzed. The MHCRI is in its infancy and it is expected that, in collaboration with city governments and other partners, the methodology will be further refined in subsequent phases of engagement so that the underlying sources of risk can be better understood by governments and can inform appropriate actions to mitigate these risks. 1.3 Implementation The assignment was conducted by a multi-disciplinary team from Chreod Arrangements Ltd. (Toronto and Shanghai) in association with SOGREAH Consultants7 (France and Shanghai) and national consultants in the Philippines and Thailand. The initiative was led by Fatima Shah, Senior Urban Economist, under the management of Vijay Jagannathan, Sector Manager, and John Roome, Sector Director, at the World Bank’s East Asia Sustainable Development Department. The Task Team included Henrike Brecht (East Asia Infrastructure Sector Unit), Yuri Dikhanov (Development Economics 7 SOGREAH merged with COTEBA in late 2010 to form the ARTELIA Group. 13 Methodology Report: Calculating Multi-hazard City Risk Data Group), Federica Ranghieri (World Bank Institute Urban Unit), Peter Jipp (Southeast Asia Sustainable Development Unit), Chris Pablo (Philippines Sustainable Development Unit), and Cathy Vidar (Philippines Sustainable Development Unit). The full Project Team is outlined in Appendix F. 2 Why a Multi-hazard City Risk Index? 2.1 Why Cities? Why should multihazard risks be assessed for cities? The world’s population became predominantly urban in 2008/09 with over 50% living in cities. Urban areas account for at least 70% of global GDP; they are also the source of 80% of all GHG emissions worldwide. Larger cities – metropolitan regions with populations over 1 million – are growing most rapidly in developing countries, especially in East Asia. Because of their inherent agglomeration economies, metropolitan regions are generally the engines of national economic growth and are magnets for migrants from rural areas and smaller cities and towns seeking jobs, greater income security and more sustainable futures for their families8. The population of metropolitan regions worldwide grew 21.5% from 2000 to 20109. They now account for 23.3 % of global population overall (and therefore for 47% of the global urban population). In East Asia, there are now 103 metropolitan regions holding just over 600 million residents; in 2000, there were 89 metropolitan regions with a total population of 530 million (Figs. 1, 2). Northeast Asia holds the bulk of the region’s metropolitan population, largely in China but also in Japan and the Republic of Korea. Metropolitan regional growth was twice the rate of total population growth in Northeast Asia from 2000 to 2010, and 2.1 times the total population growth rate in Southeast Asia. East Asia’s population is clearly gravitating to metropolitan regions. All of these metropolitan regions are exposed to natural hazards, and many regularly experience disasters which are becoming more frequent: East Asia has experienced a quadrupling of the number of annual natural disasters over the last 20 years10. Since current and reliable information is often not available in peri-urban areas beyond municipal boundaries, the initial applications of the MHCRI will focus on core municipalities. However, over time, it should be extended to assess risks across entire metropolitan regions. Cities in East Asia are among the most vulnerable in the world to the physical, social, and economic impacts of disasters. They absorb 2 million new urban residents every month and are projected to triple their built-up areas in the coming two decades – exposure is increasing and will translate into heavy loss of life and property unless proactive measures are mainstreamed into urban planning processes. These losses are particularly high in densely populated peri-urban and informal 8 see World Bank (2009) 9 Chreod Global Metropolitan Region Database, 2010 10 Shah (2010) 14 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project settlements, whose residents live on marginal lands in poorly constructed shelters and lack the financial resources to cope with the loss of property. 15 Methodology Report: Calculating Multi-hazard City Risk Figure 1: Metropolitan Regions in Southeast Asia, 2000-2010 Figure 2: Metropolitan Regions in Northeast Asia, 2000-2010 Source: Chreod Ltd. (2010) Source: Chreod Ltd. (2010) Note: vertical axis shows total population by size class of metropolitan region; horizontal axis shows number of metropolitan regions by size class 16 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 2.2 Why Multi-hazard? Cities’ exposure to the kinds and effects of natural hazards obviously vary according to location-specific geographical and geological conditions. While a city may only face a single type of hazard, such as earthquakes, the intense damage and loss from a single event can be catastrophic (e.g. Aceh). Other cities may be exposed to a larger number of hazards, but events are infrequent and effects are minimal (e.g. Singapore). Still others may be exposed to a large number of hazards, with greater frequency and intensity (e.g. Manila). While a city’s exposure to coastal floods might be minimal today, sea-level rise (SLR) arising from climate change might dramatically increase risk (e.g. Dhaka and Chittagong). Impacts on mortality, morbidity and property loss may be greater in a city experiencing frequent and multiple disasters with relatively low impacts but cumulative damages than in a city experiencing a single, significant disaster with a 500-year return period. Considerable work has been conducted around the world to assess risk to individual cities from individual hazards. Many cities have modeled seismic or flood risks, and advanced tools are available to support such efforts. However, assessing the probable cumulative impacts of multiple hazards has not entered the mainstream of urban management practice around the world. A better understanding of the impacts of multiple hazards – including future hazards arising from climate change – can support more focused engagement and policy actions by various stakeholders, ranging from multilateral agencies to local governments. Understanding multi-hazard risk could help to prioritize interventions; it could also form the basis for major institutional and fiscal adjustments that improve stakeholders’ capacities to better manage overall risk at the city scale11. Earlier work on hazards affecting cities provides an indication of the potential value of assessing multi-hazards12. A very rough approximation of cities’ past exposure to multiple hazards can be made using data on past events. Recent historical data on the frequency and intensity of floods, wind damage and storm surge from typhoons, landslides caused by heavy rains, and drought were obtained and analyzed by Columbia University’s Center for Hazards and Risk Research for the World Bank, in preparation of the “Natural Disaster Hotspots: A Global Risk Analysis� project which ended in 200513. Events were classified in deciles of frequency and intensity14. All of these data sets were geo-referenced and therefore can be spatially correlated with the location of metropolitan regions. We took the top four deciles for each event and attached decile values for each event type to each of the 444 metropolitan regions in the world using GIS15. The unweighted sum of these four sets of deciles creates an initial Composite Event Occurrence Index which is mapped globally in Figure 3 and graphed on Figure 4. 11 For example, categorical grants from senior governments to be applied to disaster risk reduction and climate change adaptation. 12 Leman (2009) 13 Dilley et al (2005) 17 Methodology Report: Calculating Multi-hazard City Risk Figure 3: Metropolitan Regions’ Exposure to Four Types of Natural Hazard Events, ~ 1980 – 2000 Source: Chreod analysis of World Bank Hot Spots Database 40,000 37,500 Tokyo 35,000 32,500 30,000 Population (thousands), 2010 27,500 25,000 Seoul 22,500 Shanghai 20,000 New York 17,500 Wuhan Delhi Hongkong-Shenzhen Kolkota 15,000 Guangzhou-Foshan 12,500 Manila Hangzhou Shantou 10,000 Dongguan Xiamen 7,500 Bangkok Ningbo Chittagong 5,000 2,500 New Orleans Davao 0 0 5 10 15 20 25 30 Severe Climate Events Occurence Index Figure 4: Metropolitan Regions by Population in 2010 and Exposure to Four Types of Natural Hazard Events, ~ 1980 – 2000 Source: Chreod analysis of World Bank Hot Spots Database 14 This was done using: 1) data on more than 1,600 storm tracks from 1980–2000, assembled by the UNEP/GRID-Geneva Project of Risk Evaluation; 2) for drought, monthly average precipitation data from 1980–2000 to assess the degree of precipitation deficit using the Weighted Anomaly of Standardized Precipitation process developed by the International Research Institute for Climate and Society at Columbia University; 3) for floods, data on extreme flood events from 1985–2003, compiled and geo-referenced by the Dartmouth Flood Observatory; and 4) for landslides, a global landslide hazard map, at a very small spatial scale, prepared by the Norwegian Geological Institute and UNEP Grid-Geneva for the Bank’s Natural Disaster Hotspots project. 15 Chreod Global Metropolitan Region Database, 2010 18 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project This work by the World Bank was not meant to be a MHCRI. The results are not part of the MHCRI model developed for this project since the Hotspots analysis does not enumerate elements of cities at risk, vulnerability, or the probability of all types of hazards (e.g. geophysical hazards are not included). It does, however, illustrate two important points: 1) that aggregation of multiple hazards can provide useful insights into the differences in potential risks facing cities (at least based on historical events); and 2) that metropolitan regions in Asia have historically been exposed to at least four natural hazards16 to a far higher degree than elsewhere in the world. 2.3 Why An Index? If sufficient resources of time, financing, technical expertise, and reliable data were available in every city, the most accurate approach to estimating multi-hazard city risk would be through a globally- standardized actuarial assessment of probable losses using catastrophe modeling with detailed fragility curves for all types of assets for all types of natural hazards. The results would be a clear and quantified estimation of probable losses for any given series of events. The resources required to conduct such an assessment of all assets for a large city would be enormous. Ensuring comparability of cities with such an approach would be challenging, especially given the divergence in currency and accuracy of information on urban capital stock both in developing and developed countries. Insurers, re-insurers and risk management consultants have invested millions of dollars and decades of research to compile and maintain catastrophe (CAT) models around the world. These CAT models are very sensitive to initial assumptions17. Even small errors in measuring wind speed, for example, can result in variations of damage estimates of up to 100%18. Uncertainties in city scale damage estimates can have confidence intervals as much as 1100% of Probable Maximum Loss19. Unless there are very accurate scientific estimates of hazard functions, inventories of building stock, and type and quality of construction, damage outputs from these models must be considered a rough estimate. Since MHCRI is meant to be a high-level, strategic decision-support tool that can be applied relatively rapidly and at far lower cost, an alternative approach is required. City risk is multi-dimensional: it includes multiple ‘elements’ at risk (people, buildings,and infrastructure), multiple hazards (geophysical, meteorological, hydrological), multiple temporal scales (the present, and future points in time), and multiple types of vulnerabilities (that vary by type of hazard). Integrating (or ‘mashing up’20) these multiple dimensions in a rational and transparent way through a single index – or a small number of indices – could form one basis for 16 floods, wind damage and storm surge from typhoons, landslides caused by heavy rains, and drought 17 Simpson, 2011 18 Air Worldwide, 2010 19 GC Briefing, 2011 20 Ravallion (2010) 19 Methodology Report: Calculating Multi-hazard City Risk supporting policy decisions that reduce city risk to natural hazards. In contrast to the actuarial approach which leads to hard numbers (number of persons at risk, monetary values of property at risk), an ‘Index’ approach is an abstraction of indicators and is meant to approximate relative significance. An index has meaning in two contexts: when comparing an entity’s index values over time (showing trends); or in comparison with other entities for which values have been calculated in the same way at the same point in time. On its own, without a spatial or temporal comparator, an index value has no utility. The value of an index increases proportionally with the number of temporal or spatial comparators. Three cities now have index values (Bangkok, Manila, Ningbo21) at two points in time: 2010 and 2030. This is admittedly a small sample set, but hopefully sufficient to attract interest in the MHCRI from more cities. Numerous indices have been created within the disaster risk management (DRM) and climate change adaptation (CCA) communities to summarize and abstract multiple factors associated with vulnerability and risk. The inherent values and limitations of risk indices are summarized in Box 2. Box 2: Characteristics of Risk Indices Indicators and indexes both seek to represent a complex reality or abstract concept with summary values. They are useful tools for showing decision- makers where they are, which way they are going and how far they are from where they want to be. A good indicator or index should alert decision-makers to a problem before it gets too difficult to fix. The distinction between an indicator and an index can be fuzzy. In this Report, an index is used specifically to refer to a numerical summary value providing information on the relative status of a unit of interest...The character of an index comes from the particular elements and values chosen as important for measurement, the subjects and units (individuals, countries, etc) of analysis, the methodology used to generate the index from input data and the specific data sources used. Indexing approaches can be characterised as inductive or deductive. Inductive approaches model risk through weighting and combining different hazard, vulnerability and risk reduction variables. Deductive approaches are based on the modelling of historical patterns of materialised risk...Inductive approaches are challenged by the absence of a universally accepted procedure for assigning values and weights to different inputs. Deductive approaches find it difficult to accurately reflect risk when disasters occur infrequently or where historical data is not available. The two approaches can support one another, for example with deductive indexes being used to validate results from inductive models. Once indicators and indexes have been developed it is important to evaluate their use. The utility of an indicator or index is not only a product of its internal logic and robustness, but also of its ability to communicate a message. The message given should be simple and intuitive. This reduces the scope for contrary interpretations of index values from contrasting political, lay or expert judgements. For constructing disaster risk and risk management indicators and indexes, different definitions of acceptable levels of risk (those adopted in building codes for example) may be used but must be made explicit. 21 A complete calculation of risks for Ningbo was not possible since meteorological and flood data required to compile Hazard Indices are considered State secrets in China. Once local authorities provide the required model inputs (not the original data), calculation of the full Ningbo MHCRI will become possible. 20 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Summing and weighting procedures invariably contain a degree of subjectivity. These and the identification of cut-off points for judging acceptability must all be carefully disclosed. Neither indicators nor indexes are politically or culturally neutral artefacts... Even a combination of indicators and indexes should be seen as only one methodology among a broader array of tools for making visible disaster risk and risk management performance. Individual indexes cannot capture all the relevant pressures and processes acting to produce disaster risk as it is experienced in the world. As with any indicator, reality is reduced down to those elements that have been judged to be the most essential in describing or understanding the phenomenon, in this case disaster risk and its management. Source: Pelling M. (2004) Given the need for a high-level, strategic decision-support tool that can be readily applied across many cities at a relatively low cost22, the MHCRI is an ‘Index’ and does not attempt to be a monetized loss model. The MHCRI is a ‘mashup index’ since it integrates multiple metropolitan elements, multiple hazards, and multiple vulnerabilities for each hazard into a single index. Ravallion23 raises four important questions that need to be addressed in the design of ‘mashup indices’: 1) what is really being measured?; 2) what trade-offs are embedded in the index?; 3) how robust are rankings given the uncertainties about data and weights?; and 4) how is the index useful for development policy? We address each question in subsequent chapters to test the integrity of the MHCRI model. 2.4 Why Another Index? Many risk indices – both for individual and multiple hazards – have been developed by various agencies and researchers over the last twenty years. Most measure risk at the country level, although several have been done for cities. To our knowledge, none integrate natural disasters and climate change adaptation at the city scale. Appendix B reviews several of these indices and highlights both differences with the MHCRI and approaches that have become mainstream in the literature, and that have been incorporated into the MHCRI. 2.5 Where MHCRI Fits in The MHCRI is a strategy-level index, situated on a continuum of Urban Risk Assessment precision in urban risk assessment. The World Bank’s Urban Development and Local Government Unit has recently proposed a three- level framework for assessing urban risk based on the cost and complexity of data and analysis requirements. The MHCRI functions at the second level, but provides the basis for a city to eventually conduct the third level of assessment using more precise and extensive information and modeling of multi-hazard risk (Fig. 5). 22 See TORs in Appendix A. 23 Ravallion (2010) 21 Methodology Report: Calculating Multi-hazard City Risk Figure 5: Levels of the Urban Risk Assessment (MHCRI in red) Source: Dickson et al (2010) MHCRI is not meant to replace Local Resilience Action Plans (LRAPs) which identify specific priority needs for short, medium, and long-term project investment and adaptation activities with specific costs, timelines, and responsible actors. 2.6 Methodological Options There are three general approaches to estimating city risks. for the MHCRI Methodology 1: An Actuarial Approach The first methodology is robust and objective but also more limited. It involves using empirical data and hazard models to represent hazard frequency and intensity mathematically, and to then combine that information with surveys of vulnerable conditions, which would include building stock data, infrastructure type and amount, as well as their locations relative to hazard. Using functions that relate hazard intensity to damage (called damage ratio curves, loss functions, or fragility curves), Risk can be estimated by integrating the product of these curves over the range of hazard. The insurance industry’s catastrophe (CAT) models follow this approach as does CAPRA24. The basic structure of the CAT models is shown in Figure 6. Final outputs typically include an exceedance probability curve of financial loss. The main sources of uncertainty relate to a lack of empirical data on hazard and vulnerability (particularly of building construction), and limited scientific knowledge of hazards and engineering. Detailed inventories of building types are an essential input to the model. There is, however, danger in viewing CAT models as highly deterministic. As a prominent CAT modeler notes25: “There is a tendency for users of catastrophe models to ask for—and expect—“the number.� The question is asked: Given that wind speeds at location X were Y mph, what are my losses at that location? By now, 24 See Appendix C. 25 Jain (2010) 22 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project however, the reader will recognize that damage—and therefore damage functions—are influenced by both complex human actions and the almost infinite complexities of the natural phenomenon being modeled. These complexities conspire to make “the number� an unobtainable goal. Even the sophisticated wind-measuring instruments currently in place in the U.S. are associated with errors on the order of 5-10%. This can translate to errors in damage estimation on the order of 50-100%. One can therefore appreciate the uncertainty in a catastrophe model’s estimation of wind speed at a location when the historical observation data on which the model is based is itself so uncertain. Predicting human behavior in response to a catastrophe is even more uncertain.� Capturing these uncertainties in the model’s damage functions is critical—and challenging. Actual catastrophe loss data has repeatedly shown that the damage probability distribution is very complex and cannot be adequately modeled using typical parametric distributions�. He goes on to identify some common myths about damage functions:  Damage functions are completely independent of other model components (e.g., hazard module).  Damage functions for a given construction type (and peril) are very similar across countries and regions.  Engineering analysis and expertise is sufficient to develop damage functions.  Damage functions are deterministic in nature. Once the wind speed is known, for example, it is possible to obtain the degree of damage with certainty.  All structures within a modeled hazard footprint will necessarily suffer damage at damaging wind speeds or ground shaking.  Changes in building codes have an immediate effect on vulnerability. Despite these limitations of damage models, the CAT approach is more robust than other methodologies since it is less subjective, and does not depend on local cultural perspectives. However, limitations to this methodology are:  It is much more data dependent than other approaches. Very detailed surveys of building and infrastructure types are required, as well as a complete set of fragility curves, which are very sensitive to type and quality of construction;  Important social, cultural, institutional, economic and environmental variables are impossible to include; and  There are significant uncertainties related to the accuracy and quality of data (beginning with biases in measuring hazard, but also including building stock data) that can result in damage estimates having errors of 100% or more, even where data sets are comprehensive. Variations in quality of construction and deterioration due to aging can result in damages that vary by orders of magnitude, for similar type buildings. Therefore, final risk estimates may be inaccurate. 23 Methodology Report: Calculating Multi-hazard City Risk Figure 6: Elements of a Catastrophe Model Source: adapted from Grossi and Kunreuther, (2005) Methodology 2: A Case Study Approach The second approach uses case studies and scenarios26. This approach can complement the others and is usually based upon a historical or hypothetical event (e.g. Hurricane Pam exercise in New Orleans). In many ways this is the approach that provides the richest amount of information since it can include a mix of quantitative and qualitative analysis, including cultural and personal narratives. It is also the one that is most difficult to use for a quantitative comparison between cities and hazard and, by definition, is outside the scope of the MHCRI. Methodology 3: An Index Approach The third methodology, and the one this project is based upon, is to create sets of dimensionless indices – focusing on cause in terms of hazard and sectors in terms of vulnerability – that can be combined in various ways to represent hazard, vulnerability and risk. If the same methodology is applied for each hazard and city, the results are comparable27. Much like the UN’s Human Development Index that combines life expectancy, education and income into a single index, this approach has both advantages and disadvantages in that it can provide a useful relative measure over time or between entities, but can be difficult to interpret. An advantage to this approach is that it includes the ability to incorporate many relevant social variables for which quantitative data is very hard to obtain or can only be estimated, and where the importance of different factors has to be subjectively assessed. As well, it uses a relatively straightforward mathematical process. The index approach28 is useful for examining relative risk between different groups, entities and geographical areas, and to evaluate trends. 26 For example, Mitchell, (1999) 27 This approach has been used, for example, by Cutter (2003) for vulnerability, and by Carreno et al (2007) and Cardona (2005) for risk. 28 reviewed extensively in Birkman (2006) 24 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project It can also be used to create profiles (as opposed to a single composite number), which can be used to identify more or most vulnerable elements. It is an approach that does not explicitly use historical disaster damage data29 as an estimate of risk, though that information provides a useful context. Rather it focuses on characteristics that make people and cities vulnerable to harm. This methodology does have limitations:  The variables/ indicators represent approximations of vulnerability, resilience and capacity, and frequently are not independent of each other. They are often chosen for intuitive reasons under the assumption that they correlate to some aspect of vulnerability, or due to data availability. These two issues create a fundamental limitation that makes the process mathematically unclear, particularly since levels of correlations are unknown.  The assignment of weights is normally subjective and in many cases varies according to local culture.  It can be difficult to precisely define what averaged or summed numbers mean; indices have meaning in a comparative sense.  In the absence of an actuarial approach, risk cannot be robustly integrated over a range of hazard magnitudes, but can only be estimated for specified magnitude levels.  Information is lost in the averaging process. In particular, case studies have demonstrated that in many disasters the existence of a single critical variable and/or critical threshold dominates outcomes. There are many types of natural hazards (and disasters) and their characteristics vary greatly. Some hazards such as drought have a gradual onset, can last for months or years, and can be very diffuse spatially. Others such as tornadoes are the opposite. Disaster studies show that many disasters occur when some critical threshold is passed – an example is a flood stage exceeding the design standards of a dam or levee. These thresholds can also be of a social nature, as happened in the 2003 heat wave in France, where elderly people without support networks were particularly vulnerable30. Existing index-type studies are generally based upon an averaging process of many factors that are given a weight in terms of their perceived importance. The process of adding and averaging a large number of factors excludes critical thresholds as an important factor in estimating disaster risk; models of this type may therefore not be a good representation for certain types of disaster risks. There is an inherently subjective aspect to this type of risk assessment that is not avoidable. In part this is because risk is socially constructed, 29 These data suffer from various problems that can make them unreliable as an indicator of risk, including observational biases, lack of data, lack of common methodologies and sampling error. 30 Identification of critical thresholds was initially proposed as part of the MHCRI methodology. However, in consultation with the pilot cities, local stakeholders requested that it be removed due to difficulties in obtaining consensus on what is considered ‘critical’ at the local level, and what are appropriate thresholds. 25 Methodology Report: Calculating Multi-hazard City Risk but it also exists because data are sometimes insufficient, and therefore potentially important variables must either be excluded or estimated using expert judgment, for example, the weighting of vulnerability parameters31. The intent of this project is to measure disaster risk, yet this is an amorphous concept. Beginning with the very definition of disaster32, the phenomenon that we are attempting to measure is not well defined. Though there is a common understanding in the literature that it is related to a range of hazard frequencies and intensities and to a large number of variables that are related to vulnerability (many of which are cultural), there is no universally accepted or robust definition. Also, it is inevitable that many vulnerability variables will be correlated (for example, poverty, housing quality and gender). It becomes difficult, as a result, to ascribe a specific meaning to a generated index in isolation. As the calculation of risk accumulates, eventually ending in a single number for a city, information tends to be lost. This suggests that several indices – rather than a single one – might best inform policy decisions33. The greatest usefulness of this index is in its application to relative levels of risk – between hazards and between cities. The MHCRI is designed to present risk in several ways to provide users of this methodology with options in how they can use the information. These include risk by hazard and vulnerability, spatial distribution of risk, and risk as a function of a Probable Maximum Event (PME). Where possible, PME is defined as a 30-year return period event34 or a historical event of record. In addition, 2-year return periods are assessed for frequent but less intense events that cause cumulative damage. Greater levels of robustness can be sought, but only at the cost of representativeness. As Ravallion suggests, there is a trade off: by reducing variables and parameters (especially those that are more subjective or difficult to quantify) calculations become more robust but less related to disaster risk35. To develop a model that includes the many different faces of vulnerability, robustness is limited. To some degree this can be tested by applying the model to a variety of different types of cities to evaluate its ability to capture varying kinds and levels of hazard and vulnerability. We address this possibility at the end of this report. 2.7 Information Collection What degree of information precision is appropriate for MHCRI? for an Index-based In terms of information collection, there are three broad possible Approach approaches to preparing the MHCRI using an Index-based Methodology, depending on the availability of resources (Table 1). The Macro-level approach relies largely on automatic processing of readily-available earth observation data (satellite imagery, digital elevation models, land use 31 See Chapter 8 for how weights were set for the first calculations of MHCRI in the three pilot cities. 32 see for example, Perry and Qarantelli, 2005 33 MHCRI measures risk separately for people and capital stock (using replacement values of buildings and infrastructure as the first variable in the risk equation). These are not combined into a single index as they are analysing completely different phenomena that do not have a common metric (ie. replacement values for people cannot – and should not – be quantified). 34 The methodology was initially designed to measure 50-year return periods. Local governments in the pilot cities requested that 30-year return periods be used instead since flood mapping had already been prepared for this return period in Bangkok and Manila. 35 see for example Figure 2.2 in Birkmann 2006, which contrasts objective versus normative information. 26 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project interpretation at 1 km scale, population estimates at 1 km scale (e.g. LANDSCAN 2008) etc., and application of national or provincial vulnerability indicators to the city scale. The principle underlying this approach is the use of remotely-obtained geographic and socio- economic information, obviating the need for local data-gathering and analysis. While conceptually attractive in terms of efficiency and comparability, the disadvantages of the Macro-level approach are: low levels of precision and accuracy at the local scale due to generalization of conditions across large areas and large spatial units (1 km+); and lack of local engagement and, consequently, potentially low acceptance by local stakeholders. At the opposite end of the scale is the Micro-level approach which is most applicable to Local Resilience Action Plans (LRAPs) which the Bank is supporting in several cities in East Asia. LRAPs are based on detailed data and information, and extensive fieldwork, including community participation in defining local conditions, exposure, and vulnerabilities. The Micro-level approach is the most time- and resource- intensive but yields the most accurate information and very high levels of local engagement and consequent acceptance of findings. The degree of detail emerging from a Micro-level approach is not possible for the MHCRI, given that it is meant to be a rapid, broad-based assessment of city risk. The MHCRI depends on a Meso-level approach to information-gathering that captures as much remotely-obtained data as possible to define conditions and vulnerabilities at the NEIGHBOURHOOD scale in cities, and that incorporates as much available, small-scale socio-economic and geographic data that is readily available. The approach described in this report captures relevant information at the smallest possible spatial scale and is designed in an open-ended way so that it could eventually lead to applications in LRAPs by local governments. Characteristics Macro Approach Meso Approach Micro Approach Ease of compilation Relatively easy to compile; More time- and resource- LRAP-oriented; detailed relies on automatic intensive than Macro analysis of local hazards, processing of readily- approach; requires careful exposure, and vulnerable available data on a limited balance between detail and populations and firms; most number of hazards and utility time-consuming and vulnerability variables, and resource-intensive on existing earth observation approach; requires datasets considerable fieldwork and surveys Data characteristics Automatic processing of ASTER DEM, LIDAR (if Geodetic surveys, LIDAR, remote-sensed, satellite- available), Medium VHR imagery, aerial based imagery and radar Resolution satellite imagery, photography, census tract data (e.g. LANDSCAN 2008, available land use mapping socio-economic data, field SRTM DEMs, GRUMP, (existing and planned), assessments (e.g. building MODIS 1000 meter data); available infrastructure conditions, slum application of national or mapping (existing and enumeration), community provincial vulnerability data planned), city-reported participation to local scale socio-economic data Unit and scale of spatial > 1 km; regions 500 m; neighbourhoods < 3m; individual building (2- analysis and 3-D) Accuracy Very low at city scale; Medium-level accuracy Very high accuracy and generalization leads to based mostly on publicly- reliability; participation by 27 Methodology Report: Calculating Multi-hazard City Risk inaccuracies at city scale available data local stakeholders adds to which could lead to local credibility of the effort stakeholder dismissal of results; Technical Capacities heavy reliance on Varying, from highly- Varying, from highly- specialized expertise and specialized to technical skills specialized to field techniques that are difficult already within local observation and qualitative to understand at the local governments reporting by residents; level; generally, all capacities available locally or nationally Policy and Program Highly-relevant at global and Potentially valuable at local Highest at local level; results Relevance continental scales; useful at level for prioritizing deeper can lead to specific national and provincial spatial and sectoral analysis programs and investment scales, including to donors; and policy; for city-wide projects very limited value to local policy and strategic policy makers investment programming; hence, more useful to donors than the Macro approach Local Engagement Very little, if any Some, especially in Highest level of local exposure and vulnerability engagement assessments Local Acceptance Doubtful at local scale due To be seen… Potentially very high to inaccuracies that undermine credibility of findings Cost Comparably low on a unit Incremental costs based on Highest cost to produce basis extent and depth of analysis possible with available resources Table 1: Three Broad Approaches to Collecting Information for the MHCRI 3 Objectives of the MHCRI 3.1 Context: Natural The principal goals of disaster risk reduction (DRR) and climate change Disasters and Climate adaptation (CCA) – in order of priority – are to: Change 1. prevent or minimize adverse effects on residents (loss of life, injury, disease) of hazardous events (both rapid onset and gradual); 2. prevent or minimize damage to – and destruction of – residents’ dwellings (their principal assets); 3. support the continued effective functioning of the metropolis (especially social and infrastructure services and core drivers of the urban economy), including during recovery from disasters. The MHCRI should inform policy decisions of local governments, national governments, and multi- and bilateral development agencies that support these three goals at the metropolitan level. As a decision-support tool, the MHCRI should particularly inform the need for more detailed exploration of: 1. where to invest in DRR and CCA to minimize vulnerability; 2. what to invest in; 3. where to concentrate policy changes; and 28 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 4. which policies need to be changed (added, removed, or adjusted). Unlike the insurance industry, DRR and CCA goals are not principally about protecting against the risk of financial loss: they are primarily directed towards protecting lives and core assets required to sustain livelihoods. These are quantified to provide common denominators in the MHCRI that can be compared across cities. 3.2 Users and Uses The MHCRI needs to support DRR and CCA decisions on investment and policy that are made at three levels: 1. Level 1: internationally, by bilateral and multilateral agencies seeking to prioritize policy support and resource allocation among cities in different countries; 2. Level 2: nationally, by governments seeking to prioritize policy support and resource allocation among cities within the country; 3. Level 3: locally, by local government seeking to prioritize policy changes and ‘hot spots’ within the metropolis that may require more detailed risk assessment. 3.3 Comparing Risks International development partners are increasingly financing disaster Between Cities prevention and preparedness in addition to disaster relief and reconstruction. Understanding where multi-hazard risks are highest would enable agencies to explore support and engagement where they could generate the greatest benefits. The MHCRI could potentially be one tool (among many) to support the targeting of CCA funds to cities facing the greatest risks from multiple hazards exacerbated by climate change. At the national level, governments could support disaster preparedness through fiscal allocations and targeted policies. If the MHCRI has been calculated for a sufficient number of cities within a country, national efforts could focus on those facing the highest risks36. 3.4 Comparing Risks to All three Levels of users will benefit from being able to monitor changes Cities Over Time in cities’ MHCRI over time. Such monitoring will inform on the effectiveness of public investments and policies in reducing vulnerabilities. It also serves to identify new or emerging risks that need to be addressed by governments and international agencies, especially those arising from climate change. 3.5 Comparing Risks Within Since the MHCRI is calculated at the sub-municipal level, it provides for a Cities wealth of information that local governments could use to help focus further analysis of possible sectoral investments and policy adjustments. It also enables governments to identify highest-risk areas within the city. When measured over time, the MHCRI informs the effectiveness of local government efforts in minimizing vulnerabilities, and to identify areas where new risks may be emerging – and why. 3.6 Design Criteria for the A purely deductive approach to preparing the MHCRI would require 36 To enable Level 1 and 2 decision-makers to compare risks between cities, the MHCRI must consistently measure urban elements at risk using common denominators. For example, if multi-hazard risks to people are being assessed, the total population should include, if possible, un-registered migrants. If risks to buildings are being analyzed, a common metric needs to be used, e.g. gross building area. 29 Methodology Report: Calculating Multi-hazard City Risk MHCRI historical data on each hazard for all cities that cover the same time periods, have been measured the same way, and at the same levels of accuracy and reliability. Availability of such data is highly unlikely in many developing countries (and probably many developed countries). An inductive approach was therefore followed in the calculation of the MHCRI. As noted earlier by Pelling, “Inductive approaches are challenged by the absence of a universally accepted procedure for assigning values and weights to different inputs.�37 In constructing the MHCRI, subjective weighting assigns relative importance among vulnerability parameters. This is unavoidable as there are no universally- accepted norms or standards for these variables. An important issue, however, is deciding who should assign the various weightings. Some of the subjective weightings and values can be set by local experts. Many of the vulnerability parameters, however, are socio- political constructs and should ideally be valued and weighted by a representative range of local stakeholders. However, this approach raises questions on the comparability among cities: would local stakeholders weight vulnerability parameters the same in New York City as in Manila? If not, inter-city comparability is distorted. A baseline weighting of vulnerability parameters needs to be established by an expert group to ensure comparability. In the current project, these have been set by the Consultant reflecting current knowledge on metropolitan management and disasters38. Given the different uses for the three levels of agents outlined above, the question arises of who should prepare the MHCRI. As a bottom-up approach, accuracy will be highest when prepared by local governments (Level 3). Experience with the Bank-supported Global City Indicators Facility39 demonstrates that ownership of a comparative urban measurement process is highest when local governments are responsible for self-reporting. Most cities, especially in developing countries, do not have the technical resources in-house (or financial resources to contract out) for advanced risk modeling to actuarial standards. This means that the MHCRI needs to be constructed such that it is:  simple, but not simplistic;  comprehensible to end users (local policy- and decision-makers in particular);  comprehensive – but ‘sufficiently’ comprehensive – so that most (probably not all) risk factors are taken into account;  evidence-based, recognizing that evidence can be quantifiable, qualitative, or even anecdotal in the absence of accurate data;  calculable by local stakeholders (i.e. not dependent on outside, highly-technical expertise);  comparable across cities over time, requiring that key metrics are measurable (e.g. replacement values) 37 This concern is echoed in Ravallion’s third question on the validity of mashup indices. 38 Vulnerability weights and their rationale are reviewed in Chapter 8. 39 http://www.cityindicators.org/ 30 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project  replicable across cities and over time;  cost-effective such that knowledge benefits exceed the costs of collecting data and processing information; and  able to be regularly updated as information, CC models, and measurement capacities improve over time. What is at risk? Consistent with the three primary goals of DRR and CCA, the focus of the MHCRI should be on assessing risks to:  people  residents’ dwellings (their principal assets)  firm’s means of production (and places of residents’ employment)  social services (especially health and education)  infrastructure services. These are captured in the MHCRI in three ‘Metropolitan Elements’: People, Buildings (residential, employment-use buildings (used as a proxy for means of production) and social service buildings (used as a proxy for social service delivery capacity)), and Infrastructure, the last two of which constitute Capital Stock. Replacement values of buildings and infrastructure provide the basis for aggregating a city’s total capital stock at risk by providing a common metric at the start of risk calculation. What impacts need to be measured? There are three kinds of impacts of disasters and climate change:  primary impacts: direct, measurable impacts on metropolitan elements (e.g. mortality, morbidity, damage and destruction)  secondary impacts: indirect impacts such as loss of income, loss of production capacity; these are more difficult to attribute and measure, and therefore to compare among cities  tertiary impacts: urban metabolic degradation, degradation of social capital and institutional capacities, macro-economic distortions caused by relief and reconstruction; these are very difficult to attribute and measure. As a high-level, decision-support tool, MHCRI is initially designed to assess primary impacts on MEs of probable disasters and climate change effects. 3.7 What is Measured: To inform decisions at ALL three levels, the MHCRI provides comparable MHCR Indices measures of: A. METROPOLIS-WIDE RISK 1. overall metropolis-wide risk to people and capital stock from multi-hazards (MHCRI)  supporting strategic decisions at Levels 1 and 2 on where to focus efforts; 2. overall city-wide risk to people and capital stock from individual hazards  supporting prioritization among hazards by all Levels (e.g. focus on earthquakes or typhoons) 31 Methodology Report: Calculating Multi-hazard City Risk B. ELEMENT RISK (METROPOLIS-WIDE) 3. risk to each metropolitan element from multi-hazards  supporting prioritization by all Levels among elements (e.g. sectoral priority on housing vs. water supply infrastructure); most useful to Levels 1 and 2 4. risk to each element from individual probable hazards  supporting sectoral prioritization by all Levels specific to type of probable hazard (e.g. dwellings vs. power infrastructure for typhoon hazard) C. SUB-METROPOLITAN RISK (SUB-METROPOLIS SCALE) 5. risk to all people and capital stock from multi-hazards at sub- municipal scales, such as districts  spatial segmentation of risk to identify areas at highest multi-hazard risk; most useful to Level 3 6. risk to all people and capital stock from individual probable hazards at sub-municipal scales  spatial segmentation of risk to identify areas at risk from each type of hazard; most useful to Level 3 (e.g. areas at risk from storm surge) D. ELEMENT RISK (SUB-METROPOLIS SCALE) 7. risk to types of people and individual types of capital stock from all probable hazards at sub-municipal scales  supporting sectoral prioritization sub-municipally based on multi-hazards; most useful to Level 3 8. risk to types of people and individual types of capital stock from individual hazards at sub-municipal scales  spatial prioritization of sectors for each type of hazard (e.g. suburban area at highest risk of riverine flooding from typhoons). Chapter 10 reports on metropolis-wide and sub-metropolitan risks in the three pilot cities. 32 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 4 The MHCRI Model 4.1 Scalar Parameters The MHCRI is calculated across the largest territory for which a single government has policy and capital investment authority, and for which comparable data are (and will likely continue to be) available and accessible. This generally equates to a municipality, such as Metro Manila governed by the Metropolitan Manila Development Authority, Bangkok governed by the Bangkok Metropolitan Administration, and the urban districts of Ningbo governed by the Ningbo Municipal Government. Sub-municipal risks can only be assessed using a bottom-up approach that aggregates risks from the smallest spatial scale for which there is reliable and available information on each of the elements consistently over time. This bottom-up approach also serves to support the veracity of metropolis-wide risk assessment. The MHCRI analyzes metropolitan elements at potential risk at the spatial scale of a 500 m x 500 m cell40. Quantitative attributes of the elements (e.g. area, length, capacities) are upscaled from finer scales where such data are available (e.g. building inventories), or downscaled from wider areas (e.g. district surveys). Hazard, exposure, and vulnerability indices are also calculated at this scale, recognizing that effects of probable hazards – and many vulnerability parameters – usually manifest across multiple cells in the same way. Data for the MHCRI need not necessarily be collected at the 500 m cell scale. Whatever scale a dataset is based upon (e.g. a municipal sub-district population survey) is entered on the GIS at its original scale: the attributes are then distributed equally to individual 500 m cells falling within the reporting spatial unit. Similarly, if a city has already done very detailed flood or housing conditions mapping, data are entered on the GIS and smaller scale attributes are calculated upwards to the 500 m cell. Both coarse and fine spatial datasets can be consistently linked to the 500 m cell in all cities. Despite the focus on the municipal territory of each city (corresponding roughly to the ‘metropolis’), calculation of the MHCRI also requires attention to higher spatial scales. How the metropolitan region as a whole is developing will heavily influence the future urban form and structure of the metropolis, and the distribution of population within it. Development 40 Advantages of using the 500 m x 500 m polygonal cell are: 1) it facilitates simple GIS-based vector analysis; 2) attributes can be attached to the cell automatically (including from raster data) and manually when needed; 3) it is scalable, allowing for the addition of variables and the extension of coverage to cover urban expansion or wider spatial areas; and 4) the cell can be split, providing for the capture of more precise data over time (including in a LRAP). Some municipalities consider small-area data to be confidential for both privacy and security reasons: aggregating data to the 500 m cell level should obviate most of these concerns. The 500 m x 500 m cell also conforms to an important new earth observation dataset that has calculated urban land use to the same cell size. The MODIS 500-m global map of urban extent was produced by Annemarie Schneider at the University of Wisconsin-Madison, in partnership with Mark Friedl at Boston University and the MODIS Land Group. It is a consistent and seamless map of urban, built-up and settled areas for the Earth’s land surface. This work builds on previous mapping efforts using Moderate Resolution Imaging Spectroradiometer (MODIS) data at 1-km spatial resolution (Schneider et al., 2003; 2005), which was included as part of the MODIS Collection 4 (C4) Global Land Cover Product (Friedl et al., 2002). The new dataset addresses weaknesses in the first map, as well as several limitations of contemporary global urban maps (VMAP, GLC 2000, GlobCover, HYDE, IMPSA, MODIS 1 km, GRUMP, Nighttime Lights v2, LANDSCAN) by applying a methodology that relies solely on newly released Collection 5 (C5) MODIS 500-m resolution data. Specifically, a supervised decision tree classification algorithm was used to map urban areas using region-specific parameters (Schneider et al., 2009). The MHCRI 500 m x 500 m grids over each city are precisely aligned with the MODIS 500 m geometry. 33 Methodology Report: Calculating Multi-hazard City Risk trends at the metropolitan region scale need to be understood to estimate the future MHCRI. For example, Thailand’s national government has established a regional development policy for the wider Bangkok metropolitan region that directs all future growth to suburban centers in adjoining provinces; Bangkok’s population is to remain stable to 2057. The national scale also needs to be addressed in terms of DRR and CC Adaptation policies that affect the core metropolis, and the system of fiscal transfers that influence local adaptive capacities41. Broader geographic regions need to be assessed, especially in terms of natural hazards (e.g. tectonic shifts, climatic regions, and downscaled GCMs). Finally, global trends in climate change need to be taken into consideration. The MHCRI is prepared for two points in time: the present (ideally, data not older than two years); and a future decade, taking into account population growth and distribution, and effects of climate change. We recommend projecting two decades forward, i.e. to the year 2030, principally because we do not believe that metropolitan development modeling can be done rigorously beyond that date. For cities to become engaged in the MHCRI process, it must be tangible enough in terms of time for estimates to be taken seriously. The time period for future iterations of the model should therefore be at most <2 years to plus 20 years. The modeling of development scenarios for the metropolis to 2030 is conducted within the context of broad metropolitan region development trends that are modeled to 2030, especially in terms of regional growth nodes which greatly influence the role of the core metropolis for which the MHCRI is estimated. National trends in CCA and DRR policies are difficult to estimate, but need to be done – at least in a broad way – to inform vulnerability assessments for 2030 at the metropolis scale. Broader climatic and tectonic trends have been predicted in various models to 2050. Temporal scales vary according to the spatial scale being analyzed or projected (Fig. 7). 41 National and local climate change adaptation policies and fiscal allocations are included as parameters in the Vulnerability Index; see Chapter 8. 34 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Figure 7: Temporal and Spatial Scales Relevant to Preparation of the MHCRI 4.2 How Risk is Calculated The term ‘risk’ has been defined in a variety of ways. In the context of this project it is rooted in the Pressure and Release Model42. Risk is defined in this model as the product of hazard and vulnerability (R = H x V). Exposure, though not always explicitly expressed, is implicit in the equation. The risk calculation in the MHCRI extends this model to the equation: Risk Index = (Metropolitan Element) x (Hazard Index) x (Vulnerability Index) Equation (1) Except for Metropolitan Elements (which are actual quantities), each component of each index is represented by a dimensionless number to allow for the aggregation of risk from different hazards and vulnerabilities. 4.3 Calculation Modules The overall MHCRI is constructed through five modules (Fig. 8). Module # 1: Metropolitan Assessment is the creation of baselines, comparable among cities, of quantitative and qualitative characteristics of residents, building structures, and infrastructure. These baselines are prepared for two timescales: current (or most recent base year for which credible data are available), and 2030, based on scenario development. They are used to identify and define elements that are potentially exposed to natural hazards, and to inform vulnerability assessments. Replacement values are calculated for each type of building and infrastructure in local prices, converted into US$. Module # 2: Natural Hazard Assessment is the calculation of probabilities of three major types of natural hazards affecting a city: geophysical, meteorological, and hydrological for both a 2-year (frequent but less intense events) and events with a 30-year return period. These 42 Blaikie et al (1994) 35 Methodology Report: Calculating Multi-hazard City Risk are calculated both for 2010 and 2030. Module # 3: Exposure Assessment is the assessment of current and future (2030) probable exposure of metropolitan elements to probable hazards. Module # 4: Vulnerability Assessment consists of assessing existing and future vulnerabilities according to impacts from exposure to probable natural hazards; it defines physical susceptibility, fragility, and resilience of the city’s capacity to withstand, prevent, minimize, adapt and recover from a hazard event, measured through physical susceptibility, building and infrastructure conditions, social capital, economy, finance, planning capacity, regulation and enforcement, and preparedness. Module # 5: Risk Assessment brings all of the preceding steps together to compile the MHCRI through the aggregation of risk indices for People and Capital Stock calculated for each natural hazard affecting the city. Each Module is described in detail in the following pages. Figure 8: Overall Process for Calculating MHCRI 36 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 5 Module 1: Inventory of Metropolitan Elements at Potential Risk 5.1 Purpose of the Module The location and quantitative characteristics of metropolitan sub- elements are defined in this Module on all 500 m cells within the municipal boundary of each city: at the outset of the MHCRI calculation, every sub-element is assumed to be potentially exposed at some level to at least one hazard. Hence, all need to be identified and defined quantitatively. The MHCRI is designed such that attributes of metropolitan elements at risk are reported by quantity of element and replacement values for each 500 m cell level and summed to provide metropolitan totals. 5.2 People The most important element is PEOPLE – how many people live where, and where they are likely to live in the future. Future growth scenarios are determined by: assessing population and urbanization growth trends, current and planned transportation networks, and economic trends; and using GIS, analyzing georeferenced urban development master plans to understand where governments hope to direct future growth. In parallel, urban population growth estimates are prepared for the city or metropolis as a whole. Aggregate future populations are then distributed across 500 m cells, according to urban master plans or, if (as in Manila) a master plan has not been prepared, by local experts. Five types of PEOPLE are enumerated in Module 1:  Total residents  Households  Workers: total number of workers and, if data are available, number of office, retail, and industrial workers  Students: total number of students; if data are available, number in primary and secondary schools and in post-secondary institutions;  Patients: total number and, if data are available, broken down into in-patients at hospitals and out-patients at clinics. To provide the base for vulnerability analysis in Module 4, total residents are further defined in terms of potentially more vulnerable groups:  Number of female residents  Residents < 5 yrs and > 65 yrs of age  Poor Households (as defined locally in each city43). This enables MHCRI to be calculated specifically for each vulnerable group. In addition, each group is explicitly included in the vulnerability indices as proxy indicators of socio-economic fragility. 5.3 Buildings Inventories of three building uses are prepared in Module 1: residential, employment buildings, and social services facilities. Each is described in terms of the following indicators:  Gross building area (m2) 43 The MHCRI model does not attempt to define a universal measure of urban poverty. Instead, local definitions are used recognizing that they cannot be accurately compared between cities (e.g. New York and Manila). 37 Methodology Report: Calculating Multi-hazard City Risk  Ground and Underground Floor Area (m2)  Replacement Values, in current USD44  Occupants: number of residents, workers, students, and patients, depending on type of building use  Building Condition: excellent (1), average (2), and poor (3) based on local experts’ assessment unless building condition reports are available. This indicator is used as a measure of physical susceptibility in Module # 4: Vulnerability. Residential buildings are differentiated according to:  ALL RESIDENTIAL BUILDINGS o Informal Settlements o Very low-density (single-family dwellings) o Low-density (row or terrace houses) o Medium-density (5-6 storey walkup apartments; closely-spaced single family dwellings) o High-density (6-10 storey apartments) o Very high-density (> 10 storey apartments) o Mixed Use: small-scale (e.g. shophouses) o Mixed Use: large-scale (e.g. apartments over large retail facilities) Employment buildings are viewed as a proxy indicator for economic production capacities and, are differentiated as follows:  ALL EMPLOYMENT BUILDINGS o Commercial Office o Retail o Industrial Estates (ie. many individual plants) o Single, large Industries (e.g. steel plant, petrochemical complex) Social Services buildings are viewed as a proxy for education and healthcare delivery, and are differentiated as follows:  ALL SOCIAL SERVICES BLDGS o Schools o Post-secondary Facilities o Hospitals o Clinics Module # 1 is designed to enable the aggregation of all buildings uses by indicator, e.g. total replacement value of all buildings, total building area. Inventories are prepared following one or more of the following techniques:  upscaling of individual building inventories to each 500 m cell45 where such inventories exist46 (Fig. 9); 44 Replacement values are calculated by multiplying the quantity of element (e.g. m2 of building area, meters of road) by a unit cost based on local market prices. 45 If inventories are geo-referenced, even with address data, GIS is used to aggregate information on individual buildings to the 500 m cell in which they are located. 46 Such inventories were available for Bangkok and Manila. 38 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project  areal interpolation of existing land use mapping and/or classification of satellite imagery using local expert knowledge (Fig. 10).47 Figure 9: Data from Existing Building Inventories up-scaled to 500 m cells Figure 10: Data Extraction Using Areal Interpolation 5.4 Infrastructure Inventories of ten types of urban infrastructure are prepared in Module # 1: power and energy, flood control, water supply, wastewater management, solid waste management, road transport, public transport, inter-city rail transport, air transport, and water transport. 47 Geospatial tools for conducting areal interpolation are documented in the Technical Manual for the MHCRI. This approach was followed in Ningbo. 39 Methodology Report: Calculating Multi-hazard City Risk Each sub-type of infrastructure is defined by three indicators: 1) quantity (size or capacity, depending on the type of infrastructure; 2) structural condition (excellent (1), average (2), and poor (3) based on local experts’ assessment unless asset condition reports are available); and 3) replacement value. The latter indicator is summed by type of infrastructure (e.g. road transport) and all types of infrastructure. Unlike buildings, for which single indicators can be calculated across all building Tokyo Electric Power Co's Fukushima types (e.g. gross floor area), there is no quantitative metric that can apply nuclear power plant where the 2011 to all forms of infrastructure, aside from replacement values. The first earthquake caused contaminated radioactive indicator (quantity) is recorded to enable local government users to water to leak into the ocean calculate risk indices for specific types of infrastructure (e.g. kms of roads at risk). The second indicator(structural condition) is used as a proxy measure of physical susceptibility in Module # 4: Vulnerability. The third indicator (replacement value) is used to aggregate values of all types of infrastructure for multi- and individual hazard risk calculations in Module 5. Power and energy infrastructure is comprised of the following sub- elements:  Power generation plants  Transmission network  Natural gas processing plants Flood control infrastructure is comprised of:  Floodwalls  Dykes and weirs  Canals  Diversion tunnels Failed Levees, New Orleans post-Katrina  Polder areas (perimeter defences)  Pumping stations. Note that flood control infrastructure is both a sub-element at risk, and a key determinant of flood risk. The latter is addressed in Module # 4: Vulnerability. Water supply infrastructure is comprised of: Cranston Wastewater Treatment Plant, Rhode  Diversion canals Island flood, 2010  Surface water intakes  Desalinization plants  Major reservoirs  Raw water conveyors  Water treatment plants  Pumping stations  Treated water storage  Water distribution network Wastewater infrastructure is comprised of:  Conveyors  Wastewater treatment plants Typhoon Ondoy, Manila, 2009  Sludge treatment facilities  Stormwater drainage network 40 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Solid waste infrastructure is comprised of:  Solid waste transfer stations  Solid waste treatment plants Road Transport  Expressways and highways  Overpasses and viaducts  Arterial Class 1 roads  Secondary collector roads  Bridges and tunnels Public Transport  Light rail transit lines  LRT stations Collapse of a bridge span for mass transit  Subway lines trains, Great Hanshin (Kobe) earthquake  Dedicated bus rapid transit routes Inter-city Rail Transport  Passenger railway stations  Railway lines Air Transport  Local airports Derailed Shinkansen bullet train, Kobe  International airports Earthquake  Air cargo facilities Water Transport  Passenger terminals  Bulk cargo terminals  Liquid cargo terminals  Container terminals For MHCRI to work effectively, local governments must be prepared to invest the effort to enumerate as much of their capital stock as possible. Based on experience with availability of data in the three pilot cities, the minimum data required to calculate MHCRI is highlighted on the master Port-au-Prince Port Damage, 2010 Earthquake template shown on Table 2. The minimum data requirement is based on the data that were available for Bangkok; considerably more data were obtained for Ningbo and Manila. 41 42 Methodology Report: Calculating Multi-hazard City Risk Table 2: Minimum Data for MHCRI calculation based on Three Pilot Cities Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 43 Methodology Report: Calculating Multi-hazard City Risk 44 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 45 Methodology Report: Calculating Multi-hazard City Risk 46 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 47 Methodology Report: Calculating Multi-hazard City Risk 48 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 5.5 Model Outputs Unlike the Hazard, Exposure and Vulnerability parts of the overall Risk equation, Metropolitan Elements are not dimensionless. Module # 1 creates outputs that are actual quantities. The relative importance of Elements are unweighted since they are all equally at potential risk. Functionality is built into the Model for users to assign weights if they consider this necessary, most likely for strategic analysis by local governments. The core of the final Risk calculations is constructed in Module # 1 since the indicators defining elements at risk48 are established at this point in the Model. There are five Indicators which feed – individually – into the calculation of Exposure in Module # 3:  Number of PEOPLE (either total number of residents, workers, students, or patients)  Gross Floor Area of BUILDINGS (either all buildings or residential, employment, or social services buildings)  Replacement Values of ALL CAPITAL STOCK  Replacement Values of BUILDINGS  Replacement Values INFRASTRUCTURE Outputs are calculable at both the Metropolitan-wide and sub- metropolitan scales, depending on user interests. Sub-metropolitan outputs can be mapped on GIS (Fig. 11). The Model provides for user selection of Spatial Scale, Elements, Sub- Elements, and Indicators that are defined in Module # 1 as the defining parameters in the Risk equation. 48 This answers Ravallion’s first question: what is really being measured? 49 Methodology Report: Calculating Multi-hazard City Risk Figure 11: Gross Area of all Buildings: Bangkok, 2009 (m2) 50 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 6 Module 2: Defining Hazard Indices 6.1 Purpose of the Module The purpose of Module # 2 is the creation of Natural Hazard Indices defining the scale, magnitude and probability of individual hazards. The quantitative values of Metropolitan Element indicators, derived in Module # 2, are multiplied by Hazard Indices to derive Exposure Indices in Module # 3. 6.2 Overall Approach The way that hazards impact communities depends upon their characteristics which can include49 frequency, intensity, duration, area, speed of onset, spatial dispersion, temporal spacing and predictability. Each of these factors could potentially be used in the hazard index, but an argument must be made for their relevance to disaster risk as defined by the current methodology. It is important to keep in mind that excluding a factor represents just as much an assumption as including it, and requires equal justification. Factors Included:  Frequency: This factor is obviously relevant, since hazard impact is directly correlated with how often the hazard occurs.  Intensity: This factor is obviously relevant, since hazard impact increases with magnitude.  Area: Hazard impact depends directly upon the area it affects, and therefore this factor is included. Factors Potentially Useful But Not Included50:  Duration: The impact of specific hazards often depends upon their duration. For example, an earthquake that lasts 45 seconds will do much more damage than one that lasts for 15 seconds. Similarly, a drought of 4 months has a much greater impact than one that lasts 6 weeks. How durations between different hazards compare is more difficult to assess. For example, how much weight should be given to floods lasting weeks as compared to earthquakes lasting seconds? Due to the ambiguity of how to relate duration between different types of events, this is not included in HI.  Speed of Onset: Hazards that develop quickly give communities no or little time to prepare for them, and therefore impacts on people are increased. There is no general “formula� however, for how onset affects impacts – to a large degree it depends upon warning systems, knowledge of how to respond and the ability of vulnerable populations to take effective action.  Temporal Spacing: This refers to the randomness or regularity of the hazard. Earthquakes, for example are random, while snowstorms or hurricanes follow a seasonal pattern and allow communities for increased preparedness.  Predictability: Hazards that can be predicted give people an opportunity to prepare and therefore minimize impacts (by evacuating, for example). Since this is the main issue related to 49 following and expanding upon Burton et al. (1993) 50 This is another of the trade-offs raised in Raballion’s second question on mashup indices. 51 Methodology Report: Calculating Multi-hazard City Risk speed of onset and temporal spacing, it is possible to conflate those two factors into predictability. This factor has not been included explicitly, but could be made implicit in parts of the vulnerability index. Factors Obviously Excluded: Spatial Dispersion: This refers to how well bounded a hazard is in geographical space. Drought, for example, is poorly bounded while a tornado is well bounded. 6.3 Hazard Model Disaster studies show that the cumulative effect of frequent hazards can be as or more significant than rare extreme events; therefore the hazard model also includes estimates of events that occur with return periods of 2 years in addition to 30-year return periods. . Hazard Index (HI) has 5 variables:  H1: spatial scale (range 0 to 1). Where the hazard spatial scale exceeds city size, such as a typhoon, then the scale = 1. Where it is less than city size, a number less than 1 will be chosen that represents a typical ratio of (hazard scale/city scale).  H2: Probable Maximum Event (PME) (range of 0 to 10), which is the 30 year return period event or the maximum event of historical record, whichever is available or greater. Where such data is not available, expert judgment will be used to estimate a reasonable PME.  H3: 2 Year Frequent Event (range of 0 to 10), which is the hazard magnitude level at which occurs, on average, every 2 years.  H4: Annual Exceedance Probability of PME  H5: Annual Exceedance Probability of the frequent event. HI can be constructed using either an additive or multiplicative approach51.  Additive (excess) model: A model in which the combined effect of the explanatory variables is equal to the sum of their separate effects. These models are used when absolute changes in variables apply.  Multiplicative (relative) model: A model in which the joint effect of two or more causes is the product of their effects if they were acting alone. These models are used when proportional changes in variables apply. A simple test for which type of model to use52 can be made by examining the trivial case where H=0; if an a priori assumption is made that HI 0 for the trivial case, then the model is proportional and therefore multiplicative. Alternatively, it is additive. Based upon this test, a multiplicative model is used for HI. 51 Menzefricke (1979) 52 Cismaru (2006) 52 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Equations (2): HI(PME) = H1 x H2 x H4 HI(2 yr) = H1 x H3 x H5 Assumptions made in this model are that :  a single hazard magnitude can be an adequate representation of city risk, as opposed to integrating over a range of magnitudes.  some hazard characteristics, such as speed of onset, are not significant.  the ranges of the indices are sensitive enough to distinguish between important risk variations. Limitations of the model are:  Depending upon data availability, it may be difficult to estimate a 30 year return period accurately.  Some hazards (such as hurricanes and earthquakes) have standard scales that can be used for index calculation. Others such as heat waves and flood, do not, which makes it more difficult to assign a PME index. Interpretation: HI is a dimensionless number with a minimum of zero. The maximum value (equation 2) for HI(PME) is (1)x(10) x (Annual Exceedance Probability) which gives a range of 0-0.33. Similarly the range of HI(2- year) is 0-5.0. For some hazards, such as volcanoes, return periods will be much longer and this will not be the case. There is no direct physical interpretation of these numbers - larger numbers indicate greater hazards, in a relative sense. In other words, if HItyphoon=0.15, then it represents a much larger hazard than HIlandslide=0.05. – in fact 3 times the hazard, as defined by the index. This does not mean that it is 3 times the risk, since that depends upon exposure and vulnerability factors; these are calculated in subsequent modules. 6.4 Geophysical Hazards Hazard Indices are constructed for the following geophysical hazards:  Earthquakes  Tsunamis  Landslides  Sudden subsidence. Gradual subsidence is not, in itself, a hazard. It contributes to risk of flooding when correlated with water levels in rivers, water bodies, and of the sea. It should be explicitly taken into account in modeling areas susceptible to flooding in Module 4: Vulnerability53. 6.5 Meteorological Hazards Hazard Indices can be constructed for the following meteorological hazards:  Typhoons 53 Subsidence was explicitly modeled in the Bangkok and Manila flood mapping. We cannot confirm if it was modelled in Ningbo. 53 Methodology Report: Calculating Multi-hazard City Risk  Severe thunderstorms  Tornados  Monsoons  Extreme temperatures (heat waves)  Droughts  Wildfires. 6.6 Hydrological Hazards Hazard Indices are also constructed for Storm Surges arising from typhoons. Given the short time period modeled in the MHCRI, sea level rise will not be sufficent by 2030 to in itself constitute a significant hazard. It is, however, an important component of flood mapping which determines physical susceptibility 54. 6.7 Climate Change Effects The risk methodology is applied to current climate. However, the past may not be a good guide for the future for many hazards as a result of climate change, subsidence and other environmental trends. Additionally, vulnerabilities are certain to change as a result of urban growth, demographic shifts, infrastructure decay, etc. From a climate perspective, future Indices for Meteorological and Hydrological Hazards are modeled on the basis of a range (i.e. uncertainties) of possible climate futures, considering how that uncertainty may affect risk (see Appendix C). Scientific literature including the range of GCM outputs, based upon the different IPCC economic futures, are assessed for each city location in order to determine the convergence of scenarios. This is challenging, especially for precipitation, since uncertainty increases as scale decreases, which means that the greatest uncertainty exists at local scales. If sufficient information exists, future HIs are defined in a 2030-low and 2030-high range for each climate-related hazard. Where model output is explicitly used, estimation of climate change for the year 2030 is made through interpolation using GCM model output. No attempt is made to integrate the 2010 and 2030 risk calculations and tipping points are not considered. 6.8 Model Outputs Outputs of Module # 2: Natural Hazards - a Hazard Index for each hazard expressed as HI(PME) or HI(2 year). Hazard indices are not combined into a single Hazard Index since there is no utility in doing so: individual hazards interact with Vulnerability parameters in different ways (Module # 4). Outputs are calculated at the Metropolitan-wide scale. If detailed earthquake, landslide, and storm surge mapping are available, and depending on the scale of such mapping, Hazard Indices can be calculated at the sub-metropolitan level for 500 m cells This also applies to heat island effects if analysis is available. A description of how Hazard Indices are calculated is shown in Appendix C for typhoons, drought, storm surge and monsoons. 54 SLR was explicitly modelled in the flood mapping obtained for Bangkok and Manila. We cannot confirm that it was taken into account in the partial flood mapping obtained for Ningbo. 54 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 7 Module 3: Calculating Exposure 7.1 Purpose of the Module Module # 3 calculates an ‘Exposure’ Index for each Metropolitan Sub- element to each individual natural hazard. 7.2 Calculation The quantities of indicators for Sub-elements (calculated in Module #1) are multiplied separately by the Hazard Indices (calculated in Module # 2) to create the Exposure Index. This index is a dimensionless number and is calculated for each 500 m cell in the city. It is an automatic calculation in the Model: the quantitity of whichever element is being assessed (e.g. people, derived from Module # 1) is multiplied by each individual Hazard Index calculated in Module # 2. Exposure values remain disaggregated by individual hazard at this stage in the model since vulnerabilities (calculated in Module # 4) are hazard-specific55. The MHCRI does not yet incorporate loss functions. Therefore, the Exposure Indices are not measures of potential losses. They are simply a computation of the quantities of indicators for Metropolitan Sub-elements times the various Hazard Indices. Without a loss function, it is not possible to compare risk indices for 2- and 30-year return periods. The reason that loss functions are not yet incorporated into the MHCRI is that, as a high-level risk assessment tool that should be able to be applied rapidly by local governments at low cost, appropriate proxies for loss curves for the range of building and infrastructure types – and for individual hazards – have not yet been identified. Loss (or fragility) curves are available for some cities, but they are owned by specialized risk management consulting firms and insurance companies and are proprietary. They are very expensive and time-consuming to prepare and maintain. An alternative approach is to define critical thresholds beyond which a hazard event causes injuries or losses of life and to property that are considered significant to local communities. The methodological challenges to this approach are: achieving a consensus among local stakeholders on what constitutes ‘significant’ damage from each type of hazard; and standardizing critical thresholds such that they are comparable between cities. Until this methdological issue is resolved in the next phase of the MHCRI’s development, 2-year risks cannot be compared with 30-year risks. Without a loss function in the MHCRI, it is not possible to differentiate the risks from a 30-year typhoon that presumably causes serious and widespread damage from less-intense typhoons that come every 2-5 years and cause limited damage. This is why 2-year Risks are reported in the City Profile Reports in an Appendix separately from the main body of the reports which focus on 30-year risks. Without a loss function, Exposure Indices are of limited utility when 55 The average exposure of all hazards can be calculated to provide a notional idea of total exposure of a metropolitan element for a 2-year and 30-year return period. However, such averages have little utility since they do not account for physical susceptibilities and other vulnerability factors that determine residual risk. This is done in Module 4. 55 Methodology Report: Calculating Multi-hazard City Risk viewed for a city at a single point in time. They have some value when comparing changes to exposure over time as a result of changing Hazard Indices and quantities of elements, sub-elements, and components at potential risk (e.g. population, building area). Similarly, Exposure Indices can be compared between cities to identify differences in exposure to individual hazards. When loss functions are incorporated into the MHCRI, they will likely become part of this Module which will then be able to estimate potential losses from each natural hazard. 7.3 Model Outputs Outputs of Module # 3 are automatically calculated Exposure Indices at the 500 m cell scale for each Natural Hazard indexed in Module # 2. 56 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 8 Module 4: Assessing Vulnerabilities 8.1 Purpose of the Module The purpose of Module # 4 is to assess vulnerabilities at the city and 500m cell scales to exposure to natural hazards. 8.2 The Vulnerability Model There is a large literature on vulnerability, but no general agreement on exactly what the word means or how it should be calculated. Vulnerability encompasses both quantitative and qualitative factors56 and increasingly resists top down approaches of calculation57. Related notions include capacity, resilience, fragility and susceptibility that are sometimes subsumed within its definition. Assessing vulnerability using an index approach is challenging. Some aspects of vulnerability are observable and measureable (such as the amount of infrastructure in flood zones) while other aspects are less so. As well, while individual vulnerability elements can be subject to a fairly robust analysis, how different elements accumulate is a complex problem, exacerbated by their mutual dependencies (for example, gender and poverty). Some vulnerabilities, such as poverty, are considered to be generic while others are very hazard specific (for example, a house might be very vulnerable to floods but not to strong winds). Vulnerability is assessed on a hazard-by-hazard basis. Similar to the Hazard Index, the Vulnerability Index can be calculated through an additive or multiplicative function. Using the same test described above, and based upon other study approaches, we have chosen an additive approach. We define the Vulnerability Index as: Equation 4: Vulnerability Index (VI) = Susceptibility + Fragility – Resilience Susceptibility: This variable focuses on physical conditions, is hazard specific and is a measure of the amount of damage caused by the hazard (range 0 to 1). Fragility: This variable focuses on social, economic and financial parameters and is a proxy measure of the potential for damage that could be caused by the hazard (range 0 to 1) Resilience (sometimes referred to as capacity): This variable indicates the ability of the city to predict, prevent, withstand, mitigate and recover from a damaging event or to contribute to community recovery (range 0 to 1). Assumptions:  That the susceptibility, fragility and resilience parameters identified accurately represent disaster risk  That weightings can be meaningfully assigned 56 Birkmann and Wisner (2006) 57 Frerks and Bender (2004) 57 Methodology Report: Calculating Multi-hazard City Risk  That sufficient data exists to represent identified parameters  That expert opinion can meaningfully represent community susceptibility, fragility and resilience for parameters that must be subjectively assessed Limitations:  Data will not be available for all potential parameters  Some vulnerabilities will not be adequately captured  Some parameters are not independent of others, resulting in a form of double counting that is difficult to avoid  Weights are subjective and may not be robust Interpretation: VI is a relative measure of how much damage or harm may result to a community based upon several sectors, including physical susceptibility (buildings and other infrastructure), fragility (social, economic and financial factors) and resilience (planning, regulation, enforcement, DRM and climate change adaptation measures). Parameters and measures are defined below. 8.3 Physical Susceptibility All parameters are measured on a scale of 0 to 1. MHCRI assesses four Physical Susceptibility parameters.  % of building area in poor condition o this calculation comes from the building condition assessments conducted in Module # 1 at the 500 m cell scale o it is relevant to all natural hazards except droughts o sub-weighting of this parameter should be hazard-specific, ie. it is less relevant to severe thunderstorms than to earthquakes o sub-weight also depends on what indicator has been selected for the total Risk analysis, e.g. if Replacement Value of Infrastructure is selected as the Risk indicator, this parameter would be given a weight of 0.0  % of infrastructure in poor condition o the same provisos as above apply to infrastructure  Susceptibility to individual hazards o this calculation comes from overlaying mapping of susceptibility to each type of hazard over the 500 m cell grid and indexing values on a scale of 0.0 to 1.0; examples are flood maps for specific return periods, earthquake and landslide hazard maps, and mapping of heat island effects.  Extent of safe open space o experience shows that access to adequate, safe open space is critical in DRM (e.g. evacuation, collection, staging areas) o this calculation comes from the inventory in Module # 1 o values from Module #1 are translated into a ten-point scale by normalizing m2 of open space per km2 and converting to a value ranging from 0 to 1. o this parameter applies to all natural hazards except for monsoons, severe thunderstorms, and droughts. 58 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Following the additive approach described above, values for all weighted Physical Susceptibility parameters are summed to each 500 m cell in the city. This total value is the number used in the vulnerability calculation. 8.4 Fragility All parameters are measured with highest fragility scoring 1 on a scale of 0 to 1. Fragility parameters are an attempt to identify relative capacities to prepare for and withstand hazard events. Three broad parameters are considered: social conditions, the urban economy, and financial capacities. All Fragility parameters apply to all natural hazards; none are considered to be hazard-specific. MHCRI assesses eight Fragility parameters which are weighted by local stakeholders. Social Conditions58:  % of households in poverty; since there is no common metric of urban poverty that can be easily calculated in all cities, household poverty is defined locally (for example, in Ningbo ‘poor households’ are defined as those receiving minimum living allowances from the Ningbo Municipal Government). o this calculation comes from the analysis conducted in Module # 1 of PEOPLE  % of population that is female o this calculation also comes from Module # 1  % of population <5yrs and >65 years of age o this calculation also comes from Module # 1 Urban Economy:  per capita GDP o disaster experience suggests that cities with higher GDP are less fragile than poor cities o for this indicator, the city’s per capita GDP (converted to PPP) as a per cent of the city in the world with the highest per capita GDP is calculated using the most recent data from OECD59; an inverse index is then applied on to the model on a scale of 0 to 1.060  % of territorial GDP (national; provincial for large countries) o for Chinese cities (and India, Brazil, Russia, USA, other large countries) share of provincial GDP rather than national share is more comparable 58 The assumption behind these parameters is that poorer households, females, and the young and elderly are more likely to be affected negatively by hazard impacts. 59 OECD (2011). Metropolitan Regions Statistical Extract. http://stats.oecd.org/Index.aspx?DataSetCode=METRO. Oslo had the highest per capita GDP of any city in the world in 2007 (latest data) at $74,324. In comparison, Tokyo’s per capita GDP was $ 38,675 and New York’s was $ 64,410. 60 For example, if the city’s per capita GDP is 70% of the highest ranked in the world, its inverse index is 0.3. 59 Methodology Report: Calculating Multi-hazard City Risk Governance:  territorial function (national; provincial for large countries) o four-point scale of 1 (global hub city, e.g. Tokyo, Shanghai), .6 (capital city), .3 (major national economic center), and .0 (other) Finance61:  Municipal government fiscal capacities o Infrastructure capital and recurring investment as % of total expenditures (inverse value)  Risk transfer mechanisms o % of homes insured (inverse index) Values are summed for each 500 m cell. Note that, for some parameters, all cells will have the same value as information is likely to only be available at the municipal level. 8.5 Resilience As noted above, Resilience is SUBTRACTED from the sum of Susceptibility and Fragility values to create the final Vulnerability Index. Both physical susceptibility and fragility are measured through socio- eonomic and geographic indicators that can be easily measured. Resilience, however, is a function of institutional capacities of local and national governments. These capacities are rated through performance measures. The following are guidelines for rating of these measures. Resilience parameters estimate institutional capacities to prevent, mitigate, and respond to natural disasters and the gradual onset of climate change effects. Four broad capacities are assessed: planning capacities; regulation and enforcement; disaster risk management; and CC adaptation. Values for all of the Resilience parameters should be set by local expert stakeholders. Those parameters that address Hyogo Framework Indicators are indicated with the Framework Priority number and individual indicators in brackets (e.g. 1:iv for the fourth indicator under Priority Action # 1)62. 1.0 Planning63: Capacities to plan effectively are important factors in reducing risks from individual and multiple hazards. These capacities are measured through: 1.1 Quality of Information Base The currency, accuracy, scope, spatial detail, and accessibility of information are the foundations of effective planning: o 1.0: municipality has an accurate, parcel-based GIS to which 61 The assumptions behind these two parameters are that a fiscally-weak municipal government invests less in capital stock and maintenance; households with un-insured property are less likely to recover quickly from a disaster. 62 The Hyogo Framework for Action (HFA) is a 10-year plan to make the world safer from natural hazards. It was adopted by 168 Member States of the United Nations in 2005 at the World Disaster Reduction Conference, which took place just a few weeks after the Indian Ocean Tsunami. 63 The assumption behind this parameter is that governments with weak planning capacities are less able to predict, plan for, and respond to effects of hazards. 60 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project recent (maximum two years) data on all socio-economic variables, buildings, land use, infrastructure and hazards required for calculation of the MHCRI are attached64; o 0.6: municipality has at least 75% of the information required to calculate the MHCRI; o 0.3: municipality has at least 50% of the information required to calculate the MHCRI; o 0.0: no data 1.2 Urban and District Planning (4:iv) o 1.0: municipality has statutory65 land use plans, completed within the last five years, at the city and districts scales for the short (5 years), medium (10 years) and long terms (20+ years); these plans incorporate measures to manage all natural hazards facing the city (e.g. no development on floodplains); o 0.6: as above, but plans are older than five years or do not cover the entire city; o 0.3: plans do not explicitly incorporate risk management measures o 0.0: no statutory plans 1.3 Infrastructure Sector Planning (4:iv; 4vi) o 1.0: municipality has plans for all infrastructure sectors (energy + power, flood control, transport, water supply, wastewater management, solid waste management, communications), completed within the last five years, at the city and districts scales for the short (5 years), medium (10 years) and long terms (20+ years); these plans incorporate measures to prevent or mitigate effects of natural hazards facing the city’s infrastructure (e.g. siting of strategic infrastructure in areas with lowest physical susceptibility to hazards); o 0.6: as above, but older than five years or for partial sectors; o 0.3: plans do not explicitly incorporate risk management measures o 0.0: no infrastructure plans 1.4 Environmental Planning o 1.0: municipality has plans for water quality protection, air qualiity protection, and protection of natural areas, completed within the last five years, at the city and districts scales for the short (5 years), medium (10 years) and long terms (20+ years); these plans explicitly incorporate measures to prevent or mitigate effects of natural hazards facing the city’s environment (e.g. coastal erosion controls); o 0.6: as above, but older than five years or for partial sectors; o 0.3: plans do not explicitly incorporate risk management 64 The rationale is that the data required for the calculation of the MHCRI are a minimum for effective sectoral and multi-sectoral planning of a municipality. 65 ie. enacted through laws, regulations, or statutes of the municipal government. 61 Methodology Report: Calculating Multi-hazard City Risk measures o 0.0: no infrastructure plans 2. Regulation + Enforcement 2.1 Building Standards (4:iv) o 1.0: statutory building code to at least the most recent International Code Council66 standards is in place for all building construction throughout the city, and is enforced through issuance of building permits and on-site inspection of permit compliance; o 0.6: code in place but not to ICC standards; governs issuance of building permits but no on-site inspection of compliance o 0.3: code in place but not enforced consistently across the city o 0.0: no code in place 2.2 Infrastructure Standards (4:iv) o 1.0: infrastructure design and construction standards in place to at least those of EN Eurocodes67 govern design of all infrastructure throughout the city, and are enforced through issuance of construction permits and on-site inspection of compliance; o 0.6: standards in place but not to EC Eurocode standards; governs issuance of construction permits but no on-site inspection of compliance o 0.3: standards in place but not enforced consistently across the city o 0.0: no infrastructure design and construction standards in place 2.3 Environmental Standards o 1.0: statutory national or municipal standards are in place that meet most recent WHO air68 and water69 quality guidelines, and are monitored throughout the city on a daily basis; o 0.6: less stringent (e.g. national) standards in place, or monitored irregularly o 0.3: as above but not monitored o 0.0: no standards in place 2.4 Land Development Controls o 1.0: statutory instruments70 in place to control site-specific land uses and floor area ratios across the municipality consistent with medium-term urban development plans; compliance required as condition of issuance of building permit 66 http://www.iccsafe.org/AboutICC/Pages/default.aspx 67 http://eurocodes.jrc.ec.europa.eu/showpage.php?id=13B 68 http://www.who.int/phe/health_topics/outdoorair_aqg/en/ 69 http://www.who.int/water_sanitation_health/publications/2011/dwq_guidelines/en/ 70 e.g. zoning ordinance 62 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project o 0.6: as above, but not across the entire city area o 0.3: as above, but not stipulating floor area ratios o 0.0: no land development controls 2.5 Flood Plain Management Standards o 1.0: statutory instruments are in place – to at least the standards of FEMA’s Floodplain Management Ordinance71 – and being enforced consistently to control development on floodplains across the municipality o 0.6: national standards in place, enforced consistently across the municipality o 0.3: as above, but not enforced across the municipallity o 0.0: no floodplain management standards 3. Disaster Risk Management 3.1 Early Warning Systems (2:iii) o 1.0: EWS to top international standards in place and regularly maintained for all hazards potentially affecting the city72 o 0.6: EWS to top international standards in place and regularly maintained for some but not all relevant hazards o 0.3: reliant on other countries for EWS information o 0.0: no EWS 3.2 Institutional Arrangements (1:I; 1:ii; 1:iv; 5:iii) o 1.0: function-specific DRM agency in place at national level with provincial and local government counterpart agencies, all with adequate funding and reporting to executive level o 0.6: DRM responsibilities clearly allocated and codified to national, provincial and local government agencies o 0.3: either national, provincial, or local DRM agency, but not at all three levels o 0.0: no DRM institutions 3.3 DRM Processes and Procedures (1:iii; 4:I; 4:ii; 4:v) o 1.0: DRM at local level clearly covers risk reduction, relief and reconstruction o 0.6: DRM at local level focuses on relief and reconstruction o 0.3: DRM at local level focuses solely on relief o 0.0: no explicit DRM processes and procedures 3.4 Human Capital (5:i; 5:ii) o 1.0: DRM agencies fully staffed with personnel requiring formal DRM accreditation at national or provincial levels and continuing education/training to maintain that accreditation o 0.6: as above but no requirement for continuing education/training 71 http://www.access.gpo.gov/nara/cfr/waisidx_02/44cfr60_02.html 72 For references, see: http://ocha.unog.ch/drptoolkit/PEarlyWarning.html 63 Methodology Report: Calculating Multi-hazard City Risk o 0.3: as above but no accreditation o 0.0: ad hoc application of personnel from line agencies 3.5 Technological Capital (2:I; 2:ii; 3:i; 5:I; 5:iv) o 1.0: municipality has a dedicated disaster risk management information system that provides real-time, spatially-referenced information on multiple hazards and disasters across the city; o 0.6: as above, but either not real-time, or does not cover all natural hazards o 0.3: as above, but does not cover the entire city o 0.0: no DRM information system 3.6 Continuity of Operations o 1 (plan in place, updated annually, facilities ready for occupancy, communications systems functioning) o 0.5 (plan in place but either not updated annually, state of facilities and communications systems not regularly monitored) o 0.0 (no plan in place) 4. Climate Change Adaptation 4.1 National CCA Policies o 1.0: National Climate Change Adaptation Policy Framework or Action Plan has been codified in legislation o 0.6: National Climate Change Adaptation Policy Framework or Action Plan is in place but not legislated o 0.3: National Climate Change Adaptation Policy Framework or Action Plan is being prepared o 0.0: no national policies 4.2 Municipal CCA Strategy o 1.0: Municipal Climate Change Adaptation Policy Framework or Action Plan has been codified in legislation o 0.6: Municipal Climate Change Adaptation Policy Framework or Action Plan is in place but not legislated o 0.3: Municipal Climate Change Adaptation Policy Framework or Action Plan is being prepared o 0.0: no municipal policies 4.3 Financing for CCA o 1.0: long term, multi-year financing for implementation of the municipal CCA action plan has been earmarked by municipal government o 0.6: medium term (5 years) financing has been earmarked o 0.3: financing needs being assessed but not yet earmarked o 0.0: financing of CCA action plan implementation not addressed Values are summed for each 500 m cell. Note that, for some parameters, 64 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project all cells will have the same value, as information is likely to only be available at the municipal level. 8.6 Weighting Vulnerabilities An important methodological issue is how to assign weights to vulnerability parameters, or even if they should be assigned73. One argument, in the interests of improving comparability between cities, is not to assign weights or to have a universal assignment. This is counter to prevailing disaster risk theory, which emphasizes the importance of community input and values74. The price paid for improving comparability is diminishing representativeness and reduced usefulness at the city level. As raised earlier, MHCRI attempts to reach a methodological compromise by fixing baseline weights for parameters that are considered critical in terms of metropolitan and disaster risk management. Physical susceptibility parameters are assigned a fixed weight of 50%, and fragility and resilience are each weighted at 25%. Weights of sub- parameters can be set by local stakeholders. Physical susceptibility is weighted the highest since it is the principal determinant of vulnerability of people and capital stock. If a city is highly susceptible to a particular hazard and has a large stock of buildings and infrastructure in poor condition, it will be highly vulnerable to disaster risk, regardless of its socio-economic fragility and resilience. Fragility and resilience are weighted equally since they can conceivably cancel each other out if a city is very fragile but highly resilient. Vulnerabilities are specific to types of natural hazards. In addition, vulnerabiliity parameters have different relevance, depending on what is being measured (people, buildings, infrastructure, total capital stock). The relative weights and sub- weights for each type of hazard in the MHCRI model for each type of metropolitan element are shown for 2010 in Table 3 and for 2030 in Table 4. 73 Note that a decision not to assign weights effectively assigns them all an equal value of 1. 74 Heijmans (2004) 65 66 VULNERABILITY FACTORS AND WEIGHTS (%): 2010 Hazards Earthquake Tsunami Volcano Landslide Sudden Flood- Tornado Extreme Drought Wildfire Subsidence related Temperatur People (18_EQ_PEOPLE) Buildings (19_EQ_BLDGS) Infrastructure (20_EQ_INFRA) People (22_TS_PEOPLE) Buildings (23_TS_BLDGS) Infrastructure (24_TS_INFRA) People (26_VOLC_PEOPLE) Buildings (27_VOLC_BLDGS) Infrastructure (28_VOLC_INFRA) People (30_LSLIDE_PEOPLE) Buildings (31_LSLIDE_BLDGS) Infrastructure (32_LSLIDE_INFRA) People (34_SUDSUBS_PEOPLE) Buildings (35_SUDSUBS_BLDGS) Infrastructure (36_SUDSUBS_INFRA) People (38_FLOOD_PEOPLE) Buildings (39_FLOOD_BLDGS) Infrastructure (40_FLOOD_INFRA) People (42_TORNADO_PEOPLE) Buildings (43_TORNADO_BLDGS) Infrastructure (44_TORNADO_INFRA) People (46_EXTEMP_PEOPLE) Buildings (47_EXTEMP_BLDGS) Infrastructure (48_EXTEMP_INFRA) People (50_DROUGHT_PEOPLE) Buildings (51_DROUGHT_BLDGS) Infrastructure (52_DROUGHT_INFRA) People (54_WILDFIRE_PEOPLE) Buildings (55_WILDFIRE_BLDGS) Infrastructure (56_WILDFIRE_INFRA) Physical % of bldg area in poor condition 10.0 20.0 0.0 5.0 10.0 0.0 5.0 5.0 0.0 5.0 5.0 0.0 5.0 5.0 0.0 10.0 25.0 0.0 15.0 25.0 0.0 15.0 25.0 0.0 0.0 10.0 0.0 0.0 5.0 0.0 Susceptibility % of infrastructure in poor condition 10.0 0.0 20.0 5.0 0.0 10.0 5.0 5.0 10.0 5.0 5.0 10.0 5.0 5.0 10.0 5.0 0.0 25.0 5.0 0.0 25.0 5.0 0.0 25.0 10.0 0.0 10.0 10.0 5.0 10.0 susceptibility to specific hazard 25.0 25.0 25.0 35.0 35.0 35.0 35.0 40.0 40.0 35.0 35.0 40.0 35.0 40.0 40.0 30.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 50.0 50.0 50.0 40.0 40.0 40.0 extent of safe open space 5.0 5.0 5.0 5.0 5.0 5.0 5.0 0.0 0.0 5.0 5.0 0.0 5.0 0.0 0.0 5.0 0.0 0.0 5.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 5.0 0.0 0.0 Physical Susceptibility Index (PSI) 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 60.0 60.0 60.0 55.0 50.0 50.0 Fragility Social poor as % of susceptible population 4.0 4.0 4.0 1.0 1.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 8.0 4.0 1.0 8.0 4.0 4.0 8.0 4.0 4.0 females as % of susceptible population 4.0 4.0 4.0 1.0 1.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 2.5 1.0 2.5 4.0 1.0 2.5 4.0 1.0 2.5 4.0 4.0 2.5 4.0 4.0 <4 and >65 yrs as % of susceptible population 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 2.5 4.0 2.5 4.0 4.0 2.5 4.0 4.0 2.5 4.0 4.0 2.5 4.0 4.0 Urban per capita GDP 2.5 2.5 2.5 1.0 1.0 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 2.5 2.5 2.5 2.5 Economy % of territorial GDP (national; provincial for large countries) 2.5 2.5 2.5 1.0 1.0 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 2.5 2.5 2.5 2.5 Governance territorial function (national; provincial for large countries) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 0.0 0.0 0.0 0.0 0.0 0.0 2.5 2.5 2.5 1.0 1.0 1.0 2.5 2.5 1.0 2.5 0.0 0.0 1.5 1.5 1.5 Methodology Report: Calculating Multi-hazard City Risk Finance mun. infra. investment as % total expenditures for last 3 yrs 2.5 2.5 4.5 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 2.5 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 4.0 4.0 5.0 risk transfer mechanisms (% homes or infra insured) 3.0 3.0 1.0 1.0 1.0 1.0 3.0 3.0 1.0 3.0 3.0 1.0 3.0 3.0 1.0 3.0 3.0 1.0 2.0 3.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Fragility Index (FI) 25.0 25.0 25.0 16.5 16.5 16.5 27.5 27.5 16.5 25.0 25.0 14.0 25.0 25.0 14.0 25.0 24.5 16.5 22.0 26.0 15.0 25.5 24.5 14.0 25.5 22.0 22.0 23.5 22.5 23.5 Resilience Planning information base 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 urban + district planning 3.0 2.0 2.0 2.0 2.0 2.0 2.0 3.0 5.0 5.0 7.0 10.0 10.0 10.0 5.0 3.0 2.0 2.0 2.0 2.0 2.0 1.5 5.0 5.0 0.0 0.0 4.0 2.0 5.0 3.0 infrastructure sector planning 1.0 1.0 1.0 1.0 1.0 10.0 2.0 1.0 10.0 1.0 1.0 6.0 1.0 1.0 15.0 1.0 1.0 9.0 1.0 1.0 5.0 1.0 1.0 4.0 0.0 0.0 1.0 0.5 1.0 5.0 environmental planning 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 Regulation + building standards 5.0 10.0 2.0 2.0 2.0 1.0 1.0 2.0 1.0 3.0 2.0 1.0 4.0 2.0 1.0 5.0 1.0 1.0 5.0 5.5 1.0 5.0 5.5 0.0 0.0 5.5 0.0 1.0 5.0 1.0 Enforcement infrastructure standards 5.0 2.0 9.0 2.0 1.0 3.5 1.0 1.0 8.5 1.0 1.0 6.0 1.0 1.0 4.0 5.0 1.0 2.0 2.0 2.0 14.0 2.0 2.0 14.5 3.0 2.0 2.0 1.5 2.0 4.0 environmental standards 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.5 1.0 1.0 1.0 land use + intensity controls 3.0 3.0 3.0 2.5 14.0 3.0 3.0 8.0 2.0 5.0 9.0 7.0 2.0 6.0 5.0 3.0 5.0 2.0 2.0 5.0 3.5 2.5 2.5 2.0 1.0 1.0 1.0 4.0 5.0 3.0 floodplain management standards 0.0 0.0 0.0 4.0 6.5 6.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 6.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Disaster Risk early warning systems 1.0 1.0 1.0 13.0 1.0 1.0 6.5 2.5 1.0 3.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 1.0 6.5 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Management institutions 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 procedures + processes 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 human capital 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 technological capital 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 continuity of operations 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 CC national CCA policies 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Adaptation local CCA policies 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 2.0 1.0 1.0 1.0 2.0 1.0 1.0 2.0 1.0 1.0 2.0 1.0 1.0 financing for CCA 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 2.0 1.0 1.0 1.0 2.0 1.0 1.0 2.0 1.0 1.0 2.0 1.0 1.0 Resilience Index (RI) 25.0 25.0 25.0 33.5 33.5 33.5 22.5 22.5 33.5 25.0 25.0 36.0 25.0 25.0 36.0 25.0 25.5 33.5 28.0 24.0 35.0 24.5 25.5 36.0 14.5 18.0 18.0 21.5 27.5 26.5 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 low weights for CCA factors at current year; increased for 2030 Vulnerabil to reflect importance of CCA actions over next 20 years Table 3: Vulnerability Weights: 2010 VULNERABILITY FACTORS AND WEIGHTS (%): 2030 Hazards Earthquake Tsunami Volcano Landslide Sudden Flood- Tornado Extreme Drought Wildfire Subsidence related Temperatur People (18_EQ_PEOPLE) Buildings (19_EQ_BLDGS) Infrastructure (20_EQ_INFRA) People (22_TS_PEOPLE) Buildings (23_TS_BLDGS) Infrastructure (24_TS_INFRA) People (26_VOLC_PEOPLE) Buildings (27_VOLC_BLDGS) Infrastructure (28_VOLC_INFRA) People (30_LSLIDE_PEOPLE) Buildings (31_LSLIDE_BLDGS) Infrastructure (32_LSLIDE_INFRA) People (34_SUDSUBS_PEOPLE) Buildings (35_SUDSUBS_BLDGS) Infrastructure (36_SUDSUBS_INFRA) People (38_FLOOD_PEOPLE) Buildings (39_FLOOD_BLDGS) Infrastructure (40_FLOOD_INFRA) People (42_TORNADO_PEOPLE) Buildings (43_TORNADO_BLDGS) Infrastructure (44_TORNADO_INFRA) People (46_EXTEMP_PEOPLE) Buildings (47_EXTEMP_BLDGS) Infrastructure (48_EXTEMP_INFRA) People (50_DROUGHT_PEOPLE) Buildings (51_DROUGHT_BLDGS) Infrastructure (52_DROUGHT_INFRA) People (54_WILDFIRE_PEOPLE) Buildings (55_WILDFIRE_BLDGS) Infrastructure (56_WILDFIRE_INFRA) Physical % of bldg area in poor condition 10.0 20.0 0.0 5.0 10.0 0.0 5.0 5.0 0.0 5.0 5.0 0.0 5.0 5.0 0.0 10.0 25.0 0.0 15.0 25.0 0.0 15.0 25.0 0.0 0.0 10.0 0.0 0.0 5.0 0.0 Susceptibility % of infrastructure in poor condition 10.0 0.0 20.0 5.0 0.0 10.0 5.0 5.0 10.0 5.0 5.0 10.0 5.0 5.0 10.0 5.0 0.0 25.0 5.0 0.0 25.0 5.0 0.0 25.0 10.0 0.0 10.0 10.0 5.0 10.0 susceptibility to specific hazard 25.0 25.0 25.0 35.0 35.0 35.0 35.0 40.0 40.0 35.0 35.0 40.0 35.0 40.0 40.0 30.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 50.0 50.0 50.0 40.0 40.0 40.0 Resilient Cities: Multi-hazard City Risk Index Project extent of safe open space 5.0 5.0 5.0 5.0 5.0 5.0 5.0 0.0 0.0 5.0 5.0 0.0 5.0 0.0 0.0 5.0 0.0 0.0 5.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 5.0 0.0 0.0 Physical Susceptibility Index (PSI) 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 60.0 60.0 60.0 55.0 50.0 50.0 Fragility Social poor as % of susceptible population 4.0 4.0 4.0 1.0 1.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 8.0 4.0 1.0 8.0 4.0 4.0 8.0 4.0 4.0 females as % of susceptible population 4.0 4.0 4.0 1.0 1.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 4.0 1.0 4.0 2.5 1.0 2.5 4.0 1.0 2.5 4.0 1.0 2.5 4.0 4.0 2.5 4.0 4.0 <4 and >65 yrs as % of susceptible population 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 2.5 4.0 2.5 4.0 4.0 2.5 4.0 4.0 2.5 4.0 4.0 2.5 4.0 4.0 Urban per capita GDP 2.5 2.5 2.5 1.0 1.0 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 2.5 2.5 2.5 2.5 Economy % of territorial GDP (national; provincial for large countries) 2.5 2.5 2.5 1.0 1.0 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 1.0 2.5 2.5 2.5 2.5 2.5 2.5 Governance territorial function (national; provincial for large countries) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 0.0 0.0 0.0 0.0 0.0 0.0 2.5 2.5 2.5 1.0 1.0 1.0 2.5 2.5 1.0 2.5 0.0 0.0 1.5 1.5 1.5 Methodology Report: Calculating Multi-hazard City Risk Finance mun. infra. investment as % total expenditures for last 3 yrs 2.5 2.5 4.5 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 4.0 4.0 5.0 risk transfer mechanisms (% homes or infra insured) 3.0 3.0 1.0 1.0 1.0 1.0 3.0 3.0 1.0 3.0 3.0 1.0 3.0 3.0 1.0 2.0 3.0 1.0 2.0 3.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Fragility Index (FI) 25.0 25.0 25.0 16.5 16.5 16.5 27.5 27.5 16.5 25.0 25.0 14.0 25.0 25.0 14.0 26.5 24.5 16.5 22.0 26.0 15.0 25.5 24.5 14.0 25.5 22.0 22.0 23.5 22.5 23.5 Resilience Planning information base 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 urban + district planning 3.0 2.0 2.0 2.0 2.0 2.0 2.0 3.0 5.0 5.0 7.0 10.0 10.0 10.0 5.0 1.0 2.0 2.0 2.0 2.0 2.0 1.5 5.0 5.0 0.0 0.0 4.0 2.0 5.0 3.0 infrastructure sector planning 1.0 1.0 1.0 1.0 1.0 10.0 2.0 1.0 10.0 1.0 1.0 6.0 1.0 1.0 15.0 1.0 1.0 9.0 1.0 1.0 5.0 1.0 1.0 4.0 0.0 0.0 1.0 0.5 1.0 5.0 environmental planning 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 0.5 1.0 1.0 1.0 Regulation + building standards 5.0 10.0 2.0 2.0 2.0 1.0 1.0 2.0 1.0 3.0 2.0 1.0 4.0 2.0 1.0 1.0 1.0 1.0 5.0 5.5 1.0 5.0 5.5 0.0 0.0 5.5 0.0 1.0 5.0 1.0 Enforcement infrastructure standards 5.0 2.0 9.0 2.0 1.0 3.5 1.0 1.0 8.5 1.0 1.0 6.0 1.0 1.0 4.0 1.0 1.0 2.0 2.0 2.0 13.0 2.0 2.0 14.0 2.5 1.5 2.0 1.0 1.5 3.5 environmental standards 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.5 0.5 0.5 1.0 0.5 1.0 1.0 1.0 1.0 1.0 1.0 0.5 1.0 1.0 1.0 land use + intensity controls 3.0 3.0 3.0 2.5 14.0 3.0 3.0 8.0 2.0 5.0 9.0 7.0 2.0 6.0 5.0 1.0 4.0 1.0 2.0 3.0 2.0 2.0 2.0 2.0 1.0 1.0 1.0 3.0 5.0 3.0 floodplain management standards 0.0 0.0 0.0 4.0 6.5 6.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.0 5.0 4.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Disaster Risk early warning systems 1.0 1.0 1.0 13.0 1.0 1.0 6.5 2.5 1.0 3.0 0.0 0.0 0.0 0.0 0.0 5.0 1.0 1.0 6.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Management institutions 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 procedures + processes 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 human capital 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 technological capital 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 continuity of operations 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 1.0 0.0 1.0 CC national CCA policies 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 Adaptation local CCA policies 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.0 2.0 2.0 1.0 2.0 2.0 2.0 1.0 1.0 2.0 1.0 1.0 2.5 1.0 1.0 financing for CCA 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.0 2.0 4.0 1.0 2.0 2.0 2.0 1.0 1.0 2.0 1.0 1.0 2.5 1.0 1.0 Resilience Index (RI) 25.0 25.0 25.0 33.5 33.5 33.5 22.5 22.5 33.5 25.0 25.0 36.0 25.0 25.0 36.0 23.5 25.5 33.5 28.0 24.0 35.0 24.5 25.5 36.0 14.5 18.0 18.0 21.5 27.5 26.5 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 low weights for CCA factors at current year; increased for 2030 Vulnerabil Indices to reflect importance of CCA actions over next 20 years Table 4: Vulnerability Weights: 2030 67 Methodology Report: Calculating Multi-hazard City Risk 8.7 Model Outputs Model outputs are hazard-specific Vulnerability Indices calculated at metropolitan and 500 m cell scales. The latter can be mapped on GIS to identify areas with greatest vulnerability, for example to flooding in 2030 (Fig. 12). Figure 12: Manila – People to Flood: Total Vulnerability Indices, 2030 68 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 9 Module 5: Calculating Risks 9.1 Indicators of Risk Since, as noted earlier, there is no effective way to generalize all Metropolitan Elements in Module # 1, five indicators can be assessed in the Risk analysis at the highest level in Module 5:  PEOPLE (either residents, households, poor households, female population, age-vulnerable population, workers, students, or patients)  BUILDINGS by Gross Floor Area(either all buildings or residential, employment, or social services buildings)  Replacement Values of ALL CAPITAL STOCK COMBINED  Replacement Values of BUILDINGS  Replacement Values of INFRASTRUCTURE. Depending on a user’s interests, Risk indices can be conducted on all of the sub-elements in the previous paragraph – and for components of these sub-elements (e.g. expressways and highways). 9.2 Risk Calculation As outlined earlier, the MHCRI equation is: Risk Index = Metropolitan Element x Hazard Index x Vulnerability Index A Multi-hazard City Risk Index is calculated for each element and sub- element for which quantitative data are available and entered into the inventory of elements in Module # 1 (Table 1). The MHCRI is an aggregation of risks from individual hazards. The indices are calculated for people (and sub-elements, e.g. poor households) and capital stock (buildings and infrastructure and their sub-elements). For buildings and infrastructure, MHCRI is calculated using replacement values as the first element in the risk equation (Risk=MSE x HI x VI). Indices are therefore available for major elements, such as total urban capital stock, down to the level of specific types of infrastructure (e.g. collector roads). This enables the assessment of risk to be conducted at a broad strategic level and also for very specific types of assets of interest to city sector managers. The MHCRI model first calculates Individual Risks for each measure and then adds risks to define an overall MHCRI for each measure, ie. the model can report on each individual risk. Individual Risk Indices are calculated since, as seen earlier, Vulnerability Indices are hazard- specific. The People MHCRI should not be added to Capital Stock MHCRI since indices are derived from entirely different phenomena (number of people, replacement values of infrastructure and buildings). For the same reason, MHCRI values for People cannot be compared with those for Capital Stock and its elements. MHCRI values for people and gross area of buildings are divided by 1,000,000 to simplify values; values for buildings, infrastructure and therefore all capital stock are divided by 1,000,000,000 for the same reason. Except for Metropolitan Elements (which are actual quantities), each component of each index is represented by a dimensionless number to 69 Methodology Report: Calculating Multi-hazard City Risk allow for the aggregation of risk from different hazards and vulnerabilities. Risks are calculated for each hazard for each 500 m x 500 m cell in the city, and then summed to calculate the MHCRI75. An illustration of the algorithms is shown in Figure 13 for the calculation of MHCRI for the Metropolitan Element of People in a single 500 m cell with a population of 5,000 in 2010 (Ningbo). Hazard Indices are generally city-wide, although broad spatial differentiation is possible for earthquakes, landslides, and storm surge. Vulnerability Indices are calculated for each cell for the indicators described in Chapter 8. The resulting Individual Risk Indices are then summed and simplified to define the MHCRI for each 500 m cell. To provide a city-level MHCRI, the sum of MHCRI values for all 500 m cells is calculated. MODULE 1 MODULE 2 MODULE 4* MODULE 5 Metropolitan Element: Individual Risk People Hazard Indices (HI) Vulnerability Indices (VI) Indices (Ri) 5,000 x 0.0150 Drought x 3.2 DR Index = 240.00 ie. population x 0.1399 Typhoon x 9.2 Flood Vuln Index = 6,435.40 living in the 500 m x 500 m cell x 0.0313 Storm Surge x 24.2 SS Vuln Index = 3,787.30 * MODULE 3 is an automatic calculation of ME x HI cell total = 10,463 + cell totals for 7967 remaining cells = total MHCRI /1 million = MHCRI Figure 13: Example of MHCRI Calculation (People) for a 500 m x 500 m cell with Population of 5,000 in 2010 9.3 Range of Calculable The following pages illustrate the range of Risk indices that can be Risk Indices calculated. METROPOLITAN-WIDE At the scale of the entire metropolis, multi-hazard and individual hazard risks can be calculated on all of the five indicators outlined in Section 9.1. This is the highest strategic level of Risk analysis possible with the MHCRI. It can inform Level 1 and Level 2 agencies in setting priorities among cities for investment and policy support. It can also enable Level 3 governments to understand which hazards pose the greatest risk. At the scale of the entire metropolis, multi-hazard and individual hazard risks can also be conducted on individual Metropolitan elements and sub-elements . This can support decisions to focus on individual sectors, e.g. on Social Service Facilities more than Infrastructure. SUB-METROPOLITAN SCALE For analysis within cities, overall risk (to the five indicators) can be 75 The sum of Individual Risk Indices is divided by 1 million for People and Gross Building Area, and by 1 billion for infrastructure and buildings where replacement value is used as the measure of Metropolitan Element. This is done to simplify the final values. 70 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project assessed for multiple and individual hazards in specific areas. Using GIS, the areas can be user-defined or a standardized set of areas can be defined administratively (e.g. districts) and functionally (e.g. metropolitan core, inner urban area, inner suburban area, outer suburban area). This would inform on what types of areas are at highest risk from multiple and individual hazards. Also within cities, individual Metropolitan Element and Sub-elements can be analysed in terms of spatial areas. For example, risks to water supply infrastructure can be compared across administrative districts from multiple and individual hazards, informing decisions on where to focus investment. 9.4 Risk Reduction Options MHCRI is meant to be a strategic support tool to inform decisions on measures that could reduce risks. The range of options available to local governments to reduce risks are limited to measures that reduce vulnerabilities, including limiting urban development in areas susceptible to natural hazards. There is nothing that governments can do to influence the magnitude and frequency of natural hazards. They cannot reduce the number of metropolitan elements at potential risk; indeed, they add to them by constantly expanding infrastructure and social services. Governments can influence vulnerabilities by reducing physical susceptibility:  to flooding by investing in flood control infrastructure and river basin management  to a range of hazards by controlling settlement in risk-prone areas through risk-based urban planning and development control;  to landslides by controlling settlement in hazard-prone areas, and by investments in geo-infrastructure;  through investing in public open spaces that can be safe post- disaster collection and distribution areas;  through improving the structural quality of infrastructure and public buildings. Governments can also influence vulnerabilities by reducing fragilities:  by educating and informing the most vulnerable populations on DRM and CCA issues, and investing in enabling strategies that strengthen the capacities of these populations to resond to hazards;  by trying to diversify urban economies so that hazard impacts are not concentrated on single strategic sectors;  by strengthening their fiscal capacities so that they invest in needed infrastructure and its maintenance. Governments can also influence vulnerabilities by improving institutional capacities that support city resilience:  by improving their urban, infrastructure, economic, and 71 Methodology Report: Calculating Multi-hazard City Risk environmental planning capacities  by strengthening regulations (especially on building and infrastructure standards, and on floodplain management)  by strengthening enforcement of these regulations  by improving the breadth and quality of early warning systems  by improving institutional coordination in DRM  by strengthening human and technological capital in DRM  by enacting and extending local policies in CCA  and by ensuring that sufficient and predictable financing is in place to implement CCA policies. Scenarios can be constructed to reflect changes to vulnerability indices and run through the MHCRI model to assess impacts on city risk. MHCRI is not simply an analytical exercise. It can form a relatively objective basis for assessing potential impacts on risk of government actions that reduce vulnerabilities to the effects of natural hazards. 72 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 10 Findings from the Three Pilot Cities 10.1 Metropolitan Elements The MHCRI has been applied to the three pilot cities of Metro Manila, Bangkok, and Ningbo (Fig. 14). Although all three are coastal cities, there are marked differences between them in: population size; capital stock; in the range, frequency and intensity of natural hazards; and in vulnerabilities. Manila, with over 11 million residents, is one of the most exposed cities in the world to multiple natural hazards, including typhoons, monsoons, drought, earthquakes, and landslides. With its large population, economic importance (it produces 37% of the Philippines’ GDP) and political role as capital city, disasters affecting Manila can have far-reaching impacts at the national scale. Climate change is expected to substantially increase Manila’s exposure to hazards which are already felt particularly hard by Manila’s 2.8 million poor most of whom live in informal settlements scattered across the city. The significant meteorological hazards facing Manila are typhoons, drought, and monsoons with typhoons having the highest impacts on a recurring basis. Drought is becoming a significant hazard due to the impacts on water supply across the city and on provision of electricity from hydro power. Manila also faces seismic risks from earthquakes and, in a limited area, from landslides. Bangkok is the core of a large metropolitan region of at least 12 million people covering an area within a radius of approximately 75 km from the city’s center. The Bangkok Metropolitan Region (BMR) comprises the Bangkok Metropolitan Administration (BMA) and five surrounding provinces. Bangkok itself has an officially estimated registered population in 2010 of 5.7 million. Approximately 37% of Thailand’s GDP is produced in Bangkok; the city is the economic powerhouse of Thailand and its national capital. Bangkok is exposed to multiple hazards: monsoons, typhoons, storm surge, drought, earthquake and tsunami. Climate change is expected to increase the intensity and, in some cases, the frequency of most of the climate-related hazards. Ningbo, with almost 3 million residents, is a strategic hub in China’s Yangtze Delta which produces almost 25% of the country’s GDP76. It is a key port city in China, serving exports from cities and towns throughout the heavily-industrialized Yangtze Delta Region and along the Yantze River basin. In 2009, Ningbo was the 5th largest port in the world in total cargo volume, and 8th largest in container traffic77. Its capacity to manage risks from natural hazards has important national implications for external trade. Ningbo is exposed to multiple hazards: typhoons, storm surge, severe thunderstorms, monsoon, extreme temperatures, and drought, all of which will likely be exacerbated by climate change78. 76 Chreod China database; city and county-level data from provincial statistical yearbooks for Zhejiang, Jiangsu and Shanghai. 77 American Association of Port Authorities (2010) 78 Because detailed climate and flood data are considered State secrets in China, no Chinese data were available for this first calculation of the MHCRI for Ningbo. Instead, global datasets were available which enabled the assessment of current risks from typhoon, storm surge and drought. The nature of these global datasets are such that there are insufficient data to model hazards to 2030, except for typhoons. 73 Methodology Report: Calculating Multi-hazard City Risk Figure 14: Overview of Pilot Cities what this map relates to in the multi-hazard city risk equation: MHCR = ME x HI x VI Despite these serious constraints on data, the MHCRI was applied in Ningbo – albeit incompletely – to illustrate to the Ningbo Municipal Government (NBMG) the potential utlity of the model in assessing multi-hazards risks over time and at a detailed spatial scale. Agencies of NBMG have all of the data required for a complete application of the model. Recognizing that those data are State secrets, NBMG should ask these agencies to run the Hazard Index and flood calculations, and to provide the results (not the data) to the World Bank so that the MHCRI can be fully applied to Ningbo. This will provide NBMG with a much more robust understanding of multi-hazard risks in the city. It would also enable NBMG to accurately compare Ningbo with Bangkok and Manila, the other two pilot cities under this project. 74 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project The differences between the three cities is shown through summary values of metropolitan elements, including values for 2030 estimated for this project on a scenario basis79 (Table 5). Metro Manila Ningbo Bangkok total per capita total per capita total per capita Metropolitan Elements, 2010 People 11,062,311 2,715,072 5,710,883 Capital Stock (rv$) 86,684,173,003 7,836 64,427,707,764 23,730 188,180,343,245 32,951 Buildings (rv$) 60,215,780,832 5,443 40,708,608,048 14,994 170,706,006,933 29,891 Infrastructure (rv$) 23,082,482,196 2,087 23,719,099,716 8,736 17,474,336,312 3,060 Env-sensitive Areas (rv$) 3,385,909,975 306 48,268 0.02 - - Buildings (m2) 241,074,958 22 267,553,896 99 435,474,507 76 Metropolitan Elements, 2030 People 16,281,237 4,308,047 5,710,883 Capital Stock (rv$) 119,860,943,882 7,362 126,366,387,989 29,333 206,010,078,445 36,073 Buildings (rv$) 88,310,018,657 5,424 79,842,671,727 18,533 170,706,006,933 29,891 Infrastructure (rv$) 28,165,015,250 1,730 46,504,148,524 10,795 35,304,071,512 6,182 Env-sensitive Areas (rv$) 0 0 19,567,738 5 - - Buildings (m2) 346,008,704 21 420,242,706 98 435,474,507 76 - not calculable due to unavailability of data Table 5: Quantities of Metropolitan Elements what this table relates to in the multi-hazard city risk equation: MHCRI = ME x HI x VI There are significant differences between the cities in replacement value of capital stock. On a per capita basis, Metro Manila’s capital stock is 25% of Bangkok’s and Ningbo’s is 72%. Building stock in Metro Manila is 18% of Bangkok’s and 36% of Ningbo’s. Ningbo’s infrastructure stock is largest of all three cities: Manila has 25% of Ningbo’s and Bangkok has 35%. The reason for this large difference is that the replacement value of Ningbo’s port infrastructure is 8.2 times that of Bangkok’s and 9.5 times Manila’s. A concerted program of road construction in Ningbo over the last 15 years has resulted in stock that is 1.5 times larger than Bangkok’s and 3.5 larger than Manila’s80. There are also large differences in gross floor area of buildings. On a per capita basis, Ningbo is highest with 98 m2/person. Bangkok’s per capita area is 76 m2/person (78% of Ningbo). However, Manila, with 21 m2/person is far behind (21% of Ningbo). Ningbo’s higher stock is due to more than a decade of extensive residential real estate development and to the extensive industrial building stock that is typical of most cities in the Yangtze Delta region, China’s principal industrial powerhouse. 10.2 Natural Hazards Ningbo faces four natural hazards, three of which could be calculated for the first iteration in calculating MHCRI. Bangkok has five hazards as does Manila (Figs. 15-17). There are marked differences in Hazard Indices between the three cities with Manila generally having far larger values. Manila’s current index for a typhoon with a 30-year return period is twice Ningbo’s and 9.2 times Bangkok’s. Its 30-year earthquake index is 2.6 times that of Bangkok’s (Ningbo has none). Manila’s hazard index for 30-year monsoons is 1.2 times Ningbo’s and 1.4 times Bangkok’s. Its 30-year drought index is 1.4 times Ningbo’s and 1% lower than Bangkok’s. 79 Note that Bangkok’s population does not change in 2030; the national government’s regional development plan calls for no population growth in Bangkok to 2057. Development is to be channelled to suburban centres in adjoining provinces. 80 MHCRI project database. 75 Methodology Report: Calculating Multi-hazard City Risk For 2-year return periods, Manila’s typhoon index is twice that of Ningbo (typhoon events in Bangkok are rare, well beyond 2 years). However, Manila’s 2-year monsoon index is only slightly higher than Bangkok’s (1.06 times). Hazard indices will increase over time due to effects of climate change (see Appendix C). By 2030, typhoon indices in all three cities will grow by 2% for 30-year events. Monsoon indices for 30-year events are projected to grow by 12% in both Manila and Bangkok81. While sufficient data are not available to project drought indices, recent studies project more intense droughts in Bangkok and Ningbo associated with climate change. Manila Hazard Indices: 30- Hazard Indices: 2-yr, Hazard Indices: 30- Hazard Indices: 2-yr, Drought Storm Surge Monsoon 0.2968 2030 Typhoon 1.8003 Landslide Tsunami Earthquake Drought * Storm Surge 0.0425 yr, 2030 Monsoon Typhoon 0.3366 Landslide 0.0043 Tsunami Earthquake 0.0396 Drought Storm Surge Monsoon 0.2650 2010 Typhoon 1.7650 Landslide Tsunami Earthquake Drought 0.1304 Storm Surge 0.0380 yr, 2010 Monsoon Typhoon 0.3300 Landslide 0.0043 Tsunami Earthquake 0.0396 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 * insufficient data to estimate drought to 2030; 2010 index carried forward Hazard Indices what this graph relates to in the multi-hazard Figure 15: Manila’s Natural Hazard Indices city risk equation: MHCRI = ME x HI x VI 81 Data to calculate 2010 and 2030 monsoon indices for Ningbo are not accessible. 76 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Ningbo Hazard Indices: 30- Hazard Indices: 2-yr, Hazard Indices: 30- Hazard Indices: 2-yr, Drought * Storm Surge ** Monsoon ** 2030 Typhoon 0.0000 Landslide Tsunami Earthquake Drought * Storm Surge ** ** yr, 2030 Monsoon Typhoon 0.1427 Landslide Tsunami Earthquake Drought 0.0000 Storm Surge 0.0000 2010 Monsoon ** Typhoon 0.8450 Landslide Tsunami Earthquake Drought 0.0914 Storm Surge 0.0313 ** yr, 2010 Monsoon Typhoon 0.1399 Landslide Tsunami Earthquake 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 * insufficient data to estimate drought to 2030 ** data not available for monsoon calculation, future storm surge calculations, 2-yr typhoon calculation for 2030 Hazard Indices what this graph relates to in the multi-hazard Figure 16: Ningbo’s Natural Hazard Indices city risk equation: MHCRI = ME x HI x VI Bangkok Hazard Indices: 30- Hazard Indices: 2-yr, Hazard Indices: 30- Hazard Indices: 2-yr, Drought Storm Surge 0 Monsoon 0.2789 2030 Typhoon 0 Landslide Tsunami Earthquake Drought * Storm Surge 0.0055 0.0314 yr, 2030 Monsoon Typhoon 0.0365 Landslide Tsunami 0.0003 Earthquake 0.0200 Drought 0 Storm Surge 0 Monsoon 0.2490 2010 Typhoon 0 Landslide Tsunami Earthquake 0 Drought 0.1317 Storm Surge 0.0055 0.0281 yr, 2010 Monsoon Typhoon 0.0358 Landslide Tsunami Earthquake 0.0150 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 * insufficient data to estimate drought to 2030 Hazard Indices what this graph relates to in the multi-hazard Figure 17: Bangkok’s Natural Hazard Indices city risk equation: MHCRI = ME x HI x VI 77 Methodology Report: Calculating Multi-hazard City Risk 10.3 Vulnerability Indices Manila is considerably more vulnerable to flooding from typhoons and monsoons than the other cities (Fig. 18). For people, current flood vulnerability in Manila is 2.3 times greater than in Ningbo (Fig. 19) and 1.5 times greater than in Bangkok (Fig. 21). This is due to the large and much more densely inhabited area at flood risk in Manila, the city’s far larger number of poor residents, and the location of their residences in areas susceptible to flooding (Fig. 20). Bangkok’s buildings are also quite vulnerable to floods, and will become more vulnerable by 2030 due to an increase in the area susceptible to flood caused by subsidence and climate change effects on precipitation and SLR (Fig. 22). The vulnerability of Ningbo’s infrastructure and buildings to storm surge is high, reflecting the coastal location of high-value port infrastructure and large industrial estates. The MHCRI model calculations show the variations of vulnerability indices within each city for different natural hazards. This is important for risk management as it shows for which hazard an element is most vulnerable, enabling the design of measures to reduce vulnerabilities on a hazard- specific basis. Metro Manila Drought, 2010 Infrastructure Buildings People 30.3 Earthquake: Storm Surge, Storm Surge, Infrastructure 2030 Buildings People Infrastructure 2010 Buildings People Infrastructure 11.9 2010, 2030 Buildings 4.3 People 6.8 28.6 Flood, 2030 Infrastructure Buildings 24.6 People 24.7 31.3 Flood, 2010 Infrastructure Buildings 30.7 People 28.5 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 Vulnerability Indices what this graph relates to in the multi-hazard Figure 18: Manila’s Vulnerability Indices city risk equation: MHCRI = ME x HI x VI 78 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Ningbo Drought, 2010 Infrastructure Buildings People 3.2 Earthquake: Storm Surge, Storm Surge, Infrastructure 12.5 2030 Buildings 13.2 People 16.3 Infrastructure 1.2 2010 Buildings 11.1 People 9.2 Infrastructure 2010, 2030 Buildings People 12.3 Flood, 2030 Infrastructure Buildings 13.1 People 16.2 11.0 Flood, 2010 Infrastructure Buildings 12.4 People 9.2 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 Vulnerability Indices Figure 19: Ningbo’s Vulnerability Indices what this graph relates to in the multi-hazard Figure 20: Manila Households Susceptible to 30-year Flood, 2010 city risk equation: MHCRI = ME x HI x VI 79 Methodology Report: Calculating Multi-hazard City Risk Bangkok Drought, 2010 Infrastructure Buildings People 13.3 Earthquake: Storm Surge, Storm Surge, Infrastructure 2030 Buildings People Infrastructure 2010 Buildings People Infrastructure 1.1 2010, 2030 Buildings 3.7 People 2.1 4.7 Flood, 2030 Infrastructure Buildings 19.6 People 16.4 3.3 Flood, 2010 Infrastructure Buildings 18.3 People 15.1 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 Vulnerability Indices what this graph relates to in the multi-hazard Figure 21: Bangkok’s Vulnerability Indices city risk equation: MHCRI = ME x HI x VI what this map relates to in the multi-hazard Figure 22: Bangkok’s Additional Building Area in Extended Flood Inundation Zone by 2030 (blue city risk equation: MHCRI = ME x HI x VI boxes) 10.4 Multi-hazard City Risk Individual risks were calculated for each natural hazard applicable to each Indices city82 and then summed to create the MHCRI. Risks were calculated for four risk conditions: current risk from 30-year events; current risk from 2- year events; risk from 30-year events in 2030; and risk from 2-year events in 2030. Multi-hazard City Risk Indices were calculated for People (Fig. 82 Except for Ningbo’s monsoon risk for reasons already reviewed. 80 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 23), Capital Stock (based on replacement values)(Fig. 24), and its sub- elements of Buildings (Fig. 25) and Infrastructure (Fig. 26) based on the minimum infrastructure stock highlighted in yellow on Table 1. For all types of people, Manila has – by far – the highest levels of multi- hazard risk among the three pilot cities. The risk index for Manila residents is 55 times higher than in Ningbo83 and 9.4 times higher than in Bangkok (Fig. 60). 6.3 Bangkok 5.9 1.6 Ningbo* 1.0 70.3 Metro Manila 55.2 0 10 20 30 40 50 60 70 80 * partial risk analysis due to unavailability Multi-hazard City Risk Index (People) of meteorological data; see Ningbo City Profile Report 2030, 30 yr 2010, 30 yr what this graph relates to in the multi-hazard Figure 23: Multi-hazard City Risk Indices – People city risk equation: MHCRI = ME x HI x VI Note: * incomplete calculation due to restrictions in accessing meteorological and flood data in China MHCRI for capital stock in Manila is 7.7 times Bangkok’s and 12 times higher than Ningbo’s. On a per capita basis, Manila has 22 m2 of building area compared to 99 m2 in Ningbo and 76 m2 in Bangkok. Despite its much smaller population, Ningbo has a building stock that is sllightly larger than Manila’s, reflecting the extensive industrial development in its suburbs and booming residential construction. Due to price differences, the building replacement value in Ningbo is $15,000/m2 compared to $5,443/m2 in Manila and $30,000/m2 in Bangkok. Risk to Buildings is shown on Figure 62. Despite its low per capita stock of buildings, Manila currently has almost 12 times the 30-year risk in Ningbo and 5.3 times the risk in Bangkok. Ningbo has extensively developed its infrastructure over the last 15 years (including its deepwater port) such that per capita stocks are much higher than in Manila. For example, Ningbo’s replacement value of expressways is $724/person compared with $35/person in Manila and $158/person in Bangkok. Manila’s normalized capital stock is much lower than either of 83 All of Ningbo’s risk indices are undercounted as flood mapping was only made available for half the city area, and meteorological data to measure monsoon risk are considered State secrets in China. See Ningbo City Profile report for full discussion of data issues. 81 Methodology Report: Calculating Multi-hazard City Risk the other cities. In total, Manila’s capital stock is $7,733/person compared to $33,117 in Bangkok and $21,605 in Ningbo. Risk to infrastructure for the three cities is shown on Figure 63. Despite Manila’s low per capita replacement costs for capital stock, the city’s much higher Hazard and Vulnerability Indices push risk to capital stock to levels well ahead of the other two cities. In addition, a far larger quantity of infrastructure is exposed to multi-hazard risk than in Ningbo and Bangkok, where very little high-value infrastructure is in areas susceptible to flooding. Manila is clearly a city with very high risks from multiple hazards. 32.2 Bangkok 18.2 25.3 Ningbo* 14.1 246.9 Metro Manila 172.4 0 50 100 150 200 250 300 Multi-hazard City Risk Index: Capital Stock * partial risk analysis due to unavailability of meteorological data; see Ningbo City Profile Report 2030, 30 yr 2010, 30 yr what this graph relates to in the multi-hazard Figure 24: Multi-hazard City Risk Indices – Capital Stock city risk equation: MHCRI = ME x HI x VI Note: * incomplete calculation due to restrictions in accessing meteorological and flood data in China 82 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 30.4 Bangkok 17.7 12.8 Ningbo* 8.0 149.3 Metro Manila 93.2 0 50 100 150 200 250 300 Multi-hazard City Risk Index: Buildings * partial risk analysis due to unavailability of meteorological data; see Ningbo City Profile Report 2030, 30 yr 2010, 30 yr what this graph relates to in the multi-hazard Figure 25: Multi-hazard City Risk – Buildings city risk equation: MHCRI = ME x HI x VI Note: * incomplete calculation due to restrictions in accessing meteorological and flood data in China 1.7 Bangkok 0.5 12.5 Ningbo* 6.1 97.6 Metro Manila 79.2 0 50 100 150 200 250 300 Multi-hazard City Risk Index: Infrastructure * partial risk analysis due to unavailability of meteorological data; see Ningbo City Profile Report 2030, 30 yr 2010, 30 yr what this graph relates to in the multi-hazard Figure 26: Multi-hazard City Risk – Infrastructure (rv$) city risk equation: MHCRI = ME x HI x VI Note: * incomplete calculation due to restrictions in accessing meteorological and flood data in China In each pilot city, risks were calculated for each applicable hazard to calculate the aggregated MHCRI of each element. Manila results are shown in Figures 27-30. The shares of multi-hazard city risk to the constituent elements of capital stock were also assessed (Figs. 31, 32). In addition, sub-elements’ shares were identified, enabling local governments to prioritize more extensive analysis of sectors (Figs. 33, 34). 83 Methodology Report: Calculating Multi-hazard City Risk Detailed results from the MHCRI application in the three pilot cities are reported in Appendix E. MULTI-HAZARD 55 Wildfire Drought 26.23 Extreme Temperature Tornado Storm Surge Monsoon 2.85 Severe Thunderstorms Typhoon 25 Sudden Subsidence Landslide 0.02 Volcanic Event Tsunami Earthquake 1.26 0 10 20 30 40 50 60 Multi-hazard City Risk Index (30-year return period) Figure 27: Risks to Manila’s People, 2010 (30-year return periods) MULTI-HAZARD 172.4 Wildfire Drought Extreme Temperature Tornado Storm Surge Monsoon 17.0 Severe Thunderstorms Typhoon 148.2 Sudden Subsidence Landslide 0.3 Volcanic Event Tsunami Earthquake 6.9 0 20 40 60 80 100 120 140 160 180 200 Multi-hazard City Risk Index (30-year return period) Figure 28: Risks to Manila’s Capital Stock, 2010 (30-year return periods) 84 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project MULTI-HAZARD 93.2 Wildfire Drought Extreme Temperature Tornado Storm Surge Monsoon 9.5 Severe Thunderstorms Typhoon 82.6 Sudden Subsidence Landslide 0.3 Volcanic Event Tsunami Earthquake 0.9 0 20 40 60 80 100 120 140 160 180 200 Multi-hazard City Risk Index (30-year return period) Figure 29: Risks to Manila’s Buildings, 2010 (30-year return periods) MULTI-HAZARD 79.2 Wildfire Drought Extreme Temperature Tornado Storm Surge Monsoon 7.5 Severe Thunderstorms Typhoon 65.6 Sudden Subsidence Landslide 0.01 Volcanic Event Tsunami Earthquake 6.0 0 20 40 60 80 100 120 140 160 180 200 Multi-hazard City Risk Index (30-year return period) Figure 30: Risks to Manila’s Infrastructure, 2010 (30-year return periods) 85 Methodology Report: Calculating Multi-hazard City Risk Elements' share (%) of risk to all water transport infra. 1.3 Capital Stock all air transport 0.1 all inter-city rail transport 0.6 Buildings Infrastructure all public transport 18.6 all road transport 9.6 all solid waste infra. all wastewater infra. 2.8 45.9 all water supply infra. 11.0 54.1 all flood control infra. 1.6 all energy + power 0.3 all social service facilities 10.2 all employment buildings 20.0 all residential 23.9 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Metropolitan Sub-elements' share (%) of risk to Capital Stock Figure 31: Manila - Elements’ Share of Multi-hazard Risk to Capital Figure 32: Manila - Sub-elements’ Share of Multi-hazard Risk to Stock Capital Stock Building types' share Infra. types' share (%) of risk to (%) of risk to Buildings (RV) Infrastructure (RV) all residential 2.7 3.5 1.4 all energy + power all employment all flood control infra. 18.9 buildings all social service 24.0 all water supply infra. facilities all wastewater infra. 44.2 40.5 all solid waste infra. all road transport 6.1 all public transport 37.0 all inter-city rail 20.9 transport all air transport all water transport infra. Figure 33: Manila - Building Types’ Share of Multi-hazard Risk to Figure 34: Manila - Infrastructure Types’ Share of Multi-hazard Risk to Buildings Infrastructure 10.5 Sub-metropolitan Since city-level MHCRI is calculated through the aggregation of MHCRI MHCRI values for each 500 m cell in the pilot cities, more detailed analysis was able to be conducted to identify areas at highest risk from individual and multiple hazards. The MHCRI Model’s integration with GIS provides for the analysis of multi- and individual risks at the sub-metropolitan level. Risk values for each 500 m cell can be aggregated to the District scale to inform on their relative levels of risk. For example, in Manila local government units’ share of Manila’s MHCRI were calculated, enabling governments to better understand the extent of risk within their jurisdictions (Fig. 35). Barangays at highest risk were also identified (Fig. 36). 86 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Valenzuela Taguig San Juan Quezon City Pateros Pasig City Pasay City Paranaque Navotas Muntinlupa Marikina Manila City Mandaluyong Malabon Makati City Las Pinas Kalookan City 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 % of Metro Manila's MHCRI Values Poor Households People (reg. residents) Figure 35: Distribution of Multi-Hazard City Risk for People among Cities and Municipalities in Metro Manila, 2010 Figure 36: Barangays in the top decile of MHCRI to People, 2010 Given the granularity of analysis possible with the 500 m cell, detailed risk mapping was conducted that allows for the identification of specific areas at highest level of risk, and for the identification of areas in which risk will increase the most between 2010 and 2030 (Figs. 37 to 40). The resulting maps are a guide for local governments to identify areas requiring more detailed risk analysis than is possible with the MHCRI; they are not meant to be used as an investment planning tool on their own. 87 Methodology Report: Calculating Multi-hazard City Risk Figure 37: Manila - Change in MHCRI: People, 2010-2030 Figure 38: Ningbo - Change in MHCRI: Capital Stock, 2010 – 2030 88 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Figure 39: Bangkok - Change in 30-yr MHCRI for Capital Stock, 2010 – 2030 Figure 40: Six Areas in Bangkok Requiring Detailed Assessment of Multi-hazard Risks 10.6 Exploring Risk The MHCRI model provides for the simulation of risks both in the future Management Options and in the past. It allows broad conclusions to be reached on possible risk management measures to influence vulnerability, either by reducing physical susceptibility (e.g. through new flood control works), reducing fragility, or by improving resilience. 89 Methodology Report: Calculating Multi-hazard City Risk This was illustrated in the Manila City Profile Report where a scenario was run in which governments invest in flood control infrastructure so that physical susceptibility to 30-year floods is reduced by 30% by 2030. Running the MHCRI model for this scenario shows that these investments would reduce the total Physical Susceptibility Index for People by 22% and the overall Vulnerability Index for People by 28% (Fig. 41). However, overall MHCRI for People would only be 95% of the MHCRI in 2010 (Fig. 42). For Capital Stock, the MHCRI would be 90% (Fig. 43). This illustrates the complexities inherent in DRR in large and rapidly-urbanizing cities. Hazard Indices in Metro Manila will not change dramatically from 2010 to 2030. A 30% reduction in physical susceptibility to floods would therefore be expected to have a greater affect on reducing MHCRI than 5% (for People). However, the driver of overall risk is clearly growth in population: in a do-nothing scenario, 48% more residents would be living in areas susceptible to 30-year floods by 2030. VULNERABILITY INDEX 18.98 Resilience Index (RI) 9.00 financing for CCA 0.00 local CCA policies 0.00 national CCA policies 0.00 continuity of operations 0.00 technological capital 0.30 human capital 0.30 procedures + processes 0.60 institutions 0.60 early warning systems 3.00 floodplain management standards 0.00 land use + intensity controls 0.60 environmental standards 0.30 infrastructure standards 0.60 building standards 0.60 environmental planning 0.60 infrastructure sector planning 0.60 urban + district planning 0.60 information base 0.30 Fragility Index (FI) 7.77 risk transfer mechanisms (% homes or infra insured) 2.55 mun. infra. investment as % total expenditures for last 3 yrs 0.00 territorial function (national; provincial for large countries) 2.50 % of territorial GDP (national; provincial for large countries)/100 0.91 per capita GDP 0.20 <4 and >65 yrs as % of susceptible population 0.85 females as % of susceptible population (/100) 2.26 poor as % of susceptible population (/100) 1.00 Physical Susceptibility Index (PSI) 20.21 extent of safe open space (% cell area/100) 1.45 susceptibility to flooding (index) 13.35 % of infrastructure in poor condition (/100) 3.20 % of bldg area in poor condition (/100) 2.21 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 Vulnerability Values: People (Flood) Figure 41: Vulnerability Index in 2030 with 30% Reduction in Depth of 30-year Floods (Metro Manila) 90 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project MULTI-HAZARD 57.99 55.16 Wildfire Drought 26* 26.23 Extreme Temperature Tornado Storm Surge Monsoon 3.42 2.85 Severe Thunderstorms Typhoon 27.06 24.79 Sudden Subsidence Landslide 0.02 0.02 Volcanic Event Tsunami Earthquake 1.26 1.26 0 10 20 30 40 50 60 70 * insufficient data to model to 2030; Multi-hazard City Risk Index (30-year return period) at least 2010 value in 2030 2030 with 30% reduction in flood-prone area 2010 Figure 42: MHCRI for People in 2030 with 30% Reduction in Depth of 30-year Floods MULTI-HAZARD 191.1 172.4 Wildfire Drought Extreme Temperature Tornado Storm Surge Monsoon 20.6 17.0 Severe Thunderstorms Typhoon 163.3 148.2 Sudden Subsidence Landslide 0.3 0.3 Volcanic Event Tsunami Earthquake 6.9 6.9 0 50 100 150 200 250 Multi-hazard City Risk Index (30-year return period) 2030 with 30% reduction in flood-prone area 2010 Figure 43: MHCRI for Capital Stock in 2030 with 30% Reduction in Depth of 30-year Floods 91 Methodology Report: Calculating Multi-hazard City Risk 92 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 11 Further Testing and Refinement 11.1 Lessons Learned from By applying the MHCRI model described in this report to the three pilot the Pilot Testing cities, several lessons have been learned that should inform future iterations in the three cities, and the application of the model in other cities. Module 1: Metropolitan Elements The ‘desired’ data for constructing a comprehensive inventory of metropolitan elements was deliberately set to a high level. In Manila and Ningbo, information was available that provided for a much more comrehensive assessment of sub-elements at potential risk than in Bangkok. However, removing data beyond the minimum required to calculate the MHCRI only resulted in a 1.8% reduction in replacement values to capital stock in Manila. The reduction in Ningbo was more significant (9%). Despite having a relatively small effect on the overall MHCRI for a city, the additional data needed to populate the ‘desired’ inventory provides local governments with the means to dig much deeper into sectoral and sub-sectoral risk. Local governments will need to decide how much data, beyond the minimum required to calculate MHCRI, yields information that is useful to their management of urban risk. Obtaining information on the structural conditions of buildings and infrastructure was difficult in all three cities. For buildings, informal settlements were used as a proxy for ‘worst’ building condition (a parameter in the Vulnerabiilty Module), but no further differentiation could be made of ‘average’ and ‘good’ conditions in the rest of a city’s building stock. For infrastructure, proxies again were used. In Manila, the national government had just published a detailed condition assessment of the national highway system, including highways in Metro Manila. In Ningbo, government staff were able to provide percentages in ‘poor’ condition for specific types of infrastructure, but without locational references. These percentages were therefore applied equally to all 500 m cells holding the particular type of infrastructure. In Bangkok, no information could be obtained on infrastructure conditions. In future iterations and for additional cities, the Bank should make it clear to local governments that the assessments of buildling and infrastructure conditions are a prerequisite for calculating the MHCRI. These need not be detailed inventories; they could simply be expert judgements of staff in various operating departments working with sketch maps84. Data were not available on all metropolitan elements for the same year. While a maximum period of two past years was set initially, data on some parameters were only available for earlier years, going back to 2004 in some cases. This temporal variability in data probably cannot be avoided: few cities in the world regularly collect information on the range of 84 Evaluation of building and infrastructure conditions in the MHCRI is, by necessity, a social construct. There is no way to establish comparable baselines applicable to all cities without detailed engineering assessments of all capital stock using common measures; this is far beyond the scope of the MHCRI. Ultimately, what matters to residents and local governments is the relative quality of capital stock within their city according to local socio-cultural standards. A building in ‘poor’ condition in New York City might well be a ‘good’ building in many developing countries. This is an example of the methodological trade-offs required in the construction of an index. 93 Methodology Report: Calculating Multi-hazard City Risk parameters in the MHCRI model. An outside period of eight years for data is more realistic. However, much can happen in a rapidly-developing city in eight years that will be missed. In terms of inter-city comparability, since all three cities had some data that was more than two years old, there is likely not much of an impact on the integrity of comparisons. The lesson learned, however, is that the degree of precision in the MHCRI is variable. Although appropriate for strategic assessment, the MHCRI should not be construed or used as a planning tool. Module 2: Hazard Indices Inaccessibility of weather-related data in China due to its ‘State secret’ status means that the only way the MHCRI can be applied in any of the country’s 654 cities is for local governments to themselve calculate Hazard Indices following the approach described in this report. They need not divulge the data; rather, they should use the data to calculate the indices. If the MHCRI is to be applied further in Chinese cities, the Bank should first obtain clear agreement from local governments that they will retain local experts to calculate the Hazard Index values. Calculation of the Hazard Indices requires technical expertise in meteorology and statistics that is not likely to exist within local governments85. Local technical expertise required to calculate these indices is available in institutes and universities in Thailand, China and the Philippines. It is likely available for most of East Asia. However, there may be countries where there is insufficient capacity and data which would mean that MHCRI could not be calculated using the current approach. This would require that the current approach be replaced with one that provides far less rigorous results. A trade-off would need to be made between including cities by following a simplistic approach to hazard estimation, and losing the rigor in the majority of cities where technical capacities and data exist to calculate more robust Hazard Indices. Module 4: Vulnerability A critical and highly-weighted parameter of vulnerability is physical susceptibility to individual hazards. Spatial precision in defining susceptibility is needed to identify people and capital stock at actual risk from those hazards that do not manifest at the city scale. This requires current flood mapping in cities facing typhoon, storm surge, monsoon and severe thunderstorm hazards. It also requires seismic risk mapping for cities facing earthquake and landslide risk. Such mapping was available for Manila and Bangkok, but not for all of Ningbo. A pre-condition for applications of the MHCRI in other cities should be the availability of susceptibility maps, or that governments have and are willing to mobilize domestic capacities to prepare them. 11.2 Improving the MHCRI At this initial stage in development of the MHCRI, the model does not Model incorporate loss or damage functions since the costs of doing so in multiple cities with varying types of buildings and infrastructure and 85 Appendix C describes how Hazard Indices are calculated. 94 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project hazard profiles are far beyond what is envisioned for the MHCRI. In the next phase of MHCRI development, identifying and incorporating appropriate proxies for affordable loss estimation86 for each natural hazard needs to be explored. Alternately or in tandem, a second look needs to be taken at critical thresholds beyond which a hazard event causes injuries or losses of life and to property that are considered significant to local communities. The methodological challenges to this approach are: achieving a consensus among local stakeholders on what constitutes ‘significant’ damage from each type of hazard; and standardizing critical thresholds such that they are comparable between cities87. So far, the model only assesses hazard events with 30-year and 2-year return periods. Longer periods need to be modeled, especially earthquakes and 100-year floods. The model only assesses single events, not coupled events such as earthquakes and resulting fires. Only natural hazards are modeled; anthropogenic hazards that accelerate or amplify the effects of natural hazards are not assessed (e.g. poor maintenance of upstream water diversion infrastructure). The possibility of incorporating coupled hazard events, including anthropogenic events, needs to be explored in the next phase. Finally, the current spreadsheet structure of the model (Appendix E) is not conducive to running rapid sensitivity analysis to explore implications of changes to vulnerability indicators (including their weights), and changes to quantities of metropolitan elements. This will be greatly facilitated once the Bank completes the migration of the model into a relational database that will drive the web portal that is under development. 11.3 Criteria for Further Further testing of the MHCRI model should be conducted to explore its Testing limits and utility, and to refine the model based on lessons learned in the initial three pilot cities. Selection of additional pilot cities can be guided by three decision criteria:  expanding the representativeness of cities in terms of the size and characteristics of metropolitan elements (e.g. population size, scope and quality of infrastructure);  expanding the range of natural hazards that cities face;  expanding the range of vulnerability conditions in cities. The first criterion would suggest focusing on smaller or larger cities. We do not think that much would be gained from focusing solely on city size at this point. As the model is refined, its utility should be assessed for very large metropolitan regions such as Shanghai, Tokyo, and Mumbai. However, further testing is needed before investing the resources that 86 Resources will not be available to most local governments to follow an actuarial approach in calculating the MHCRI. Loss curves for different types of buildings and infrastructure for individual hazards are inherently very costly to prepare and maintain. 87 Critical thresholds were included in the MHCRI at the start of the project, but were removed after these concerns were raised by the Bangkok Metropolitan Administration and Ningbo Municipal Government. 95 Methodology Report: Calculating Multi-hazard City Risk analysis of these cities would require. The second criterion would suggest exploring cities facing higher levels of risk from a broader range of hazards. There is significant merit in doing this to test the Hazard Index model in Module # 2. Cities in Indonesia that face earthquake, tsunami, landslide and volcanic hazards could be candidates. The third criterion – vulnerability differences – would suggest exploring cities likely to have very high vulnerabilities or low vulnerabilities. This selection filter would, in our view, yield the greatest benefits for model development as it would serve to further test the vulnerability parameters and measurement approaches in Module # 4. Cities with low vulnerabilities in EAP could be Singapore, Hongkong and Seoul. Beyond EAP, there could be merit in selecting more advanced cities in the US (e.g. New York, Los Angeles) and Europe (e.g. London, Amsterdam). Cities with higher vulnerabilities in EAP could be selected cities in Indonesia (that also face a broader range of hazards), Philippines (e.g. Davao), Cambodia, and Vietnam. Beyond EAP, cities in MENA (e.g. Cairo, Alexandria, Sana’a) could be considered as well as in South Asia (e.g. Chittagong, Dhaka, and a range of cities in India). Circumstances at the Bank will of course determine the selection of additional pilot cities. However, in terms of model development, we suggest that expanded ranges of vulnerability conditions be the key criterion for city selection. 96 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Appendix A: Terms of Reference RESILIENT CITIES: MULTI-HAZARD CITY RISK INDEX Terms of Reference A. Background and Context The countries of East Asia and the Pacific (EAP) are among the most vulnerable in the world to the physical, social, and economic impacts of disasters. The frequency of natural disasters has quadrupled in the past two decades and climate change is expected to lead to further increases in the future. Urban centers in the region stand to be hit hard by these disasters because of their location (many are located in floodplains or coastal zones) and concentration of economic activity and people. With urbanization continuing at unprecedented rates – cities in East Asia absorb 2 million new urban residents every month and are projected to triple their built-up areas in the coming two decades – exposure is increasing and will translate into heavy loss of life and property unless proactive measures are mainstreamed into urban planning processes. These losses are particularly high in densely populated peri-urban and informal settlements, whose residents live on marginal lands in poorly constructed shelters and lack the financial resources to cope with the loss of property in the absence of adequate social safety nets. To help policymakers with these challenges, the World Bank has produced several tools at the intersection of disaster risk management, climate change, and planning in urban areas. One such tool is the Climate Resilient Cities: A Primer on Reducing Vulnerabilities to Disasters88 (February 2009), which provides guidance to local governments in East Asia on the concepts of climate change and disaster risk reduction; how climate change consequences contribute to urban vulnerabilities; and what is being done by city governments around the world to actively engage in capacity building and capital investment programs for building resilience. Following the launch of the Primer, six city governments – in Vietnam (Hanoi, Can Tho, and Dong Hoi), Indonesia (Yogyakarta), Philippines (Iloilo), and China (Ningbo) – have received technical assistance from the World Bank to apply the broad concepts from the Primer to their local contexts and produce Local Resilience Action Plans (LRAPs).89 Experience with the pilot program highlights the need for such localized tools and strategies for managing climate and disaster risks. At present, much of the disaster mitigation and climate adaptation planning and allocations take place at the national level (e.g. based on SNAPs, NAPAs, and national policies), while much of the action is inherently local. To help fill this gap, the ongoing technical assistance and city partnerships that result in LRAPs is intended to provide a tool for investment planning at the urban level. The LRAP reflects a risk assessment based on current hazards the city faces and potential impacts of future climate changes in the context of urban expansion; an institutional and policy analysis to determine whether plans are in place to mitigate some of these risks; various option to mitigate risks that remain unaddressed; and results in a set of specific prioritized structural investments and nonstructural measures, with rough cost approximations and timelines, that the city would like to undertake to increase its resilience. The World Bank is now preparing a Workbook on Developing LRAPs that captures the methodology and experiences from the initial pilot cities so that other cities can undertake the process on their own. There is also a need for the quantitative tools that enable local and national officials to assess a city’s aggregate vulnerability to potential climate and disaster impacts and benchmark progress toward enhanced resilience over time. While each city will need to think of a more detailed and locally customized monitoring and evaluation program to track the implementation status of individual action items from the LRAP, and evaluate their impacts, there is also the need for a standardized Multi- Hazard City Risk Index (MHCRI). This Index could be one component of a city’s detailed monitoring system but would not 88 The publication can be downloaded at www.worldbank.org/eap/climatecities. The hotspots exercise (chapter 3) provides some suggested inputs into the MHCRI. 89 These programs are being funded through a variety of sources, including GFDRR, TFESSD, AusAID, BB, and the cities’ own resources. The Vietnam program has been completed, and the remaining programs will be completed by April 2011. Two additional cities in Indonesia will also undertake the program. Cities in MENA and SSA also applied the Primer’s “hotspot� methodology in 2009. 97 Methodology Report: Calculating Multi-hazard City Risk substitute for it. However, there should be a link between the city’s LRAP, or other discrete actions to reduce disaster and climate risks, and its MHCRI score. The MHCRI would provide a standardized metric to capture aggregate risk at the city level, and help establish a baseline and measure performance over time and relative to peers. The assignment described in these TOR is for the methodology development of the MHCRI and its application (as part of methodology refinement) to selected pilot cities in East Asia. This project is managed by the World Bank’s East Asia and Pacific Region, Sustainable Development Department – Infrastructure Unit (EASIN). Among other sectors, EASIN manages the World Bank’s regional portfolio on the Urban Sector as well as Disaster Risk Management, including climate adaptation. The following donors/ trust funds are acknowledged for their support to this assignment:  Australian Agency for International Development (AusAID) Infrastructure for Growth Trust Fund  Global Facility for Disaster Reduction and Recovery (GFDRR)  Korean Trust Fund (KTF)  UK’s Department for International Development (DFID) Trust Fund supporting pillar 3 (adaptation to climate change) of the Clean Energy Investment Framework (CEIF) B. Objectives The objectives are to provide (i) policymakers and planners in East Asian cities with decision support tools on assessing, mitigating, and benchmarking urban risk arising from natural hazards and climate change; (ii) national governments with tools to guide inter-governmental fiscal transfers geared toward translating national policy on disaster risk reduction and climate adaptation into local action; and (iii) international development partners with a standardized and rapid risk assessment at the city level that can help identify areas of greatest need as well as capacity to effectively utilize funding support. The MHCRI should reflect a robust methodology that adequately captures and aggregates risk at the city level; is comparable across cities and produces logical scores/ ranks; is credible, objective and standardized; is cost-effective and easily replicable using free data and standard software; is simple and clearly understandable. The objective is to provide an evidence-based, verifiable metric to support policymaking that could be updated at regular intervals. C. Scope of Work 1. A conceptual framework and model: A methodology will be developed for the MHCRI that will capture risk at the urban level in terms of:  Hazards: both slow and rapid onset events, seismic as well as hydromet – e.g. earthquake, volcanic eruption, tsunami, typhoon/ windstorm, snow storm, precipitation-based flooding, drought, landslide, sea level rise, temperature increases, changing precipitation patterns – and their associated intensive and extensive risks looked at separately;  Vulnerability: qualitative (proxy) and quantitative measures for the degree of susceptibility of the urban system – including vulnerability of built structures, environment, economy, people, and institutions – to damaging effects of natural hazards;  Exposure: value of life, infrastructure, and economic assets at-risk; with options for disaggregation of incidence by income level (i.e. impact on the poor), geographic area (i.e. to identify areas for urgent intervention), and sector (e.g. government/ public, commercial/ industrial, residential); and  Adaptive capacity: national and local policies and institutions for overall metropolitan management and specifically for adequate hazard risk reduction, preparedness, and response; ability to raise finances from local sources, contingency planning, transparency in resource use – at the city level and also at the level of the disaster management department; technical capacity/ links to technical institutes. A distinction should be made between 98 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project current adaptive capacity and potential adaptive capacity (given additional financial resources). The focus here is on government institutions, rather than on community-based or private sector initiatives. Leverage existing methodologies. There are several frameworks for urban risk that already exist (see Annex 1). While the MHCRI would establish a new metric for urban risk arising from hazards, a review should be undertaken to determine whether aspects of existing frameworks can be adapted/ leveraged. The focus is on value added and not on reinventing the wheel. Some of the indices provide insights into aggregation of risk, others on dealing with uncertainty, and still others on interaction with urban dynamics (e.g. urban expansion, land subsidence); some focus on one type of hazard, others adopt a multi-hazard approach, and still others focus on the set of climate impacts only; some look at the national level, and a few at the urban level. Indentify indicators and proxies for variables. The input and output variables for the MHCRI model need to be determined and specific indicators need to be identified. Part of the task is to determine the appropriate variables on a conceptual level, but the challenge also lies in being able to identify indicators (or their proxies) that are objective and standardized, and for which data that would be widely available for a large number of cities. There may be some indicators that would require in- country engagement, field work, and validation; however, the bulk of the assessment should be designed in a way that it could be undertaken remotely using open source data, software, and models. Level of analysis and approximation. To work toward a cost-effective approach to replicating the index, an exhaustive inventory of all specific assets within the city is not expected. Approaches have been developed to derive indications of population/ occupancy, infrastructure/ buildings, and economic activity from Earth Observation data on structural characteristics of the built-up area, demographic databases, land use plans/ location of industrial zones. This information can be used to also disaggregate economic activity as well as population by income level. Establishing unit costs and layering various sources of information can produce an exposure database. It is important to factor in sources of future exposure, such as projected expansions of the urban built-up area and population, as well as trends in subsidence due to excessive groundwater extraction or other factors related to urban management more broadly. A note on resolution: it is important to note that the MHCRI is intended for use at city-scale; while it could provide general guidance on areas, populations, or sectors within the city that are particularly at-risk, the Index is not meant to substitute for detailed local planning.90 Assigning appropriate weights. Once the variables have been identified, appropriate weights will need to be assigned to each based on conceptual frameworks and local context – for instance, different weights may be assigned to the relative importance of variables according to a classification of city attributes. Determining probabilities. There are several uncertainties that need to be accounted for, including the likelihood that the event will occur within a given return period; its magnitude/ intensity, frequency, and duration; interaction effects, the probability that the primary event will trigger a secondary event, or that multiple events will occur simultaneously; and the types of impacts (losses and damages, and their distribution) that are likely to occur. While historical data and trends are necessary, they are no longer sufficient given the changes in urban land cover and population, as well as the changes to the climate that are expected – these factors will need to be accounted for in the model. Accounting for uncertainty will require both scenario-based analysis as well as Monte Carlo simulations. With regard to climate change, appropriate IPCC climate change scenarios will need to be investigated in two different time periods in the future – and the choice of these points in time will need to be justified. An assessment of city risk today should reflect future risks as well; actions taken today will reduce these future risks and improve Index scores. Constructing sub-indices. In addition to presenting the risk assessment in the aggregated form of the MHCRI, sub-indices would also be important for presenting different filters/ cuts of the data. At least three appropriate filters would need to be determined in light of end-user demand/ policy needs such as: 90 The objective is not to identify the particular bridge that needs reinforcement or the levee that needs to be raised (these types of interventions could be identified through the LRAP – see background section). Yet, the resolution must be sufficient for analyzing city-wide risk, rather than a broader regional assessment. 99 Methodology Report: Calculating Multi-hazard City Risk  Filter by hazard type or a set of hazards: What are the main hazards that my city is prone to? What is my city’s risk in terms of sea-level rise? What about earthquakes? What about overall climate impacts?  Filter by population exposure: How many lives are at-risk? What is the exposure of the urban poor?  Filter by asset exposure: What is the value of overall infrastructure at-risk? Is there sufficient economic diversification or is my city’s relying on a single sector that is exceptionally exposed, and how much of this exposure is inherent to the sector and how much can be affected by public policy and regulation?  Filter by region: Should my city focus interventions on a specific geographic area that is particularly at-risk? Is this area where most of my city’s GDP is generated or is it abandoned land?  Filter by capacity: Does a particular city have adequate technical capacity to reduce risk but is lacking financial resources? Is the city set up to make effective use of additional resources if they were to be provided?  Filter by overall factors: Where are my city’s main risks stemming from? How much of this can be influenced by policy decisions and investments? Should the focus be on institutional capacity, focus on preparedness for a specific hazard, or measures targeting a specific area of the city? 2. Application to the pilot cities: The conceptual framework for the MHCRI will only be as good as its application. Part of the exercise in refining the conceptual model will be to apply it to at least three pilot cities in East Asia. The cities that have been identified are as follows:  Bangkok, Thailand  Ningbo, China  Metro Manila, Philippines These pilots have been selected because of regional/ national importance, availability of data/ existing studies, susceptibility to a variety of hazards, challenges in terms of urban poverty/ informal settlements, size, and donor funding for this exercise. The objective is to be able to use the pilots to compare cities to assess the robustness of the methodology before possible scale-up. The MHCRI methodology will first be applied to these pilots based on desk research, to the extent possible. Consistent with the conceptual framework developed, the city-level analysis will use historical data based on past disasters (location, intensity, structural damage, and loss of life), time series on urban expansion (population densities, changes in land cover, and available poverty maps), and probabilistic risk modeling for future impacts of potential disasters and climate change in light of urban dynamics. Field visits may then be undertaken for verification and to refine the data collection and assessment in collaboration with government partners and stakeholder inputs. Based on the results, the methodology may need to be fine- tuned. Some of the previous/ ongoing work in these cities on more intensive within-city planning on climate adaptation and disaster mitigation measures91 has produced a series of recommendations/ priorities in terms of structural and nonstructural measures to increase the city’s resilience. These measures can also provide inputs for testing and calibrating the index methodology – e.g. if the city implemented some of the most important measures in these plans, would it would affect the MHCRI. In addition, the MHCRI will provide insights as to what aspects of risk city managers should be focusing on, based on sensitivity analysis identifying the parameters that give the most “bang-for-the-buck� for risk reduction. D. Deliverables The following deliverables are expected as part of the assignment: 1. Work plan and Mobilization of Team 91 For instance: ongoing work on LRAPs in the Indonesian cities, Iloilo, and Ningbo; the Climate Change Impact and Climate Adaptation Study for the Bangkok Metropolitan Region, as well as some work on poverty mapping; an ongoing “climate change and the urban poor� case study for Jakarta, as well as background studies for the JEDI Project (Jakarta Emergency Dredging Initiative on flood potential, drainage, subsidence, and sea level rise; the OECD (2007) on port cities, etc. 100 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project A detailed consultant team work plan to execute the assignment will need to be formulated, with specific timelines, outputs, and involvement from core consultant team members. The workplan proposed in the Consultant proposal may need to be modified based on discussions prior to contract signing. 2. Methodology Report A report detailing the proposed conceptual methodology and recommended construction of the MHCRI should be presented, as well as the rationale for each aspect of the model as described above. It should be well-written and comprehensive, with sufficient detail to be understood by non-technical practitioners. Annexes should include, among others: a review of the literature/ existing methodologies and explain how and why the MHCRI model departs from these; a discussion of the requirements for adapting the methodology for application at the intra-city level (e.g. neighborhood/ ward); a discussion on how, if at all, application to the pilot cities (and workshop application – see below) led to changes in the formulation of the index; and a guide on a low-cost method for application of the MHCRI92 to all cities in the region with populations exceeding one million, with a section describing special considerations for replication in other regions. 3. Model and Code The model should run on open-source software; if this is not possible, the Consultant shall provide the customized software for a cross platform environment, including for MS-Windows XP environment. Both the source code and the complied binaries shall be delivered, as well as a manual for the model to help with replication/ update. The Consultant shall submit all inputs developed under this project that would be needed for possible replication of the MHCRI to additional cities at some point in the future, if the World Bank chooses to scale up the pilot exercise. 4. City Profiles For each of the three pilot cities, a brief city profile should be submitted which provides each city’s current MHCRI and, based on the Index, the measures that would improve its score. The analysis should also detail which of these measures the city is already planning to undertake (the MHCRI score should reflect ongoing measures as well as those that are budgeted for, but not those that are planned but largely unfunded mandates) and where the gaps exist – i.e. where the city should be focusing. The MHCRI cannot be the sole basis for making investment decisions at the city level; however, it can provide one input and the recommendations in terms of improving MHCRI scores will be important in this regard. Each profile should disaggregate the index and provide explanation to the input variables and weights, as well as a rationale for the scores. Annexes should include: data sources and indicators used; and an inventory of relevant documents/ analysis for the city and how these have been used in the construction of the index. 92 See www.doingbusiness.org for an example of a very low cost index that is updated annually based on a virtual network of professionals around the world based on an objective methodology. 101 Methodology Report: Calculating Multi-hazard City Risk Annex 1: Examples of Existing Frameworks on Climate and Disaster Risk Indices Index/ Framework Year Level Application Organization Link Climate Change Index 2007 National Global Michele B. http://www.up.ethz.ch/publicati (CCI) Baettig, Martin ons/documents/Baettig_2006G Wild, Dieter M. L028159.pdf Imboden Climate Risk Index (CRI) 2009 National Global Germanwatch http://www.germanwatch.org/kl ima/cri.htm Climate Vulnerability 2009 National ECA World Bank http://www.worldbank.org/eca/ Index climate/ECA_CCA_Full_Repor t.pdf Climate Vulnerability 2009 City EAP WWF http://assets.panda.org/downloa Scorecard ds/mega_cities_report.pdf Disaster Deficit Index 2009 National , LAC IDB http://www.iadb.org/exr/disaste (DDI), Local Disaster update possible at r/ddi.cfm?language=EN&parid Index (LDI), Risk subnational =2 Management Index (RMI), Prevalent Vulnerability Index (PVI) Disaster Preparedness 2006 City, USA Fritz Institute/ http://www.fritzinstitute.org/PD Index (Dpi) & Resiliency community University of Fs/WhitePaper/DaveSimpson% Index (DRi) Louisville 20IndicatorsRepor.pdf Disaster Risk Index 2009 National Global UNDP http://www.undp.org/cpr/disred (DRI) revision /english/wedo/rrt/dri.htm Earthquake disaster risk 1997 City Global Stanford Blume http://blume.stanford.edu/pdffil index (EDRI) Earthquake es/Tech%20Reports/TR121_Da Engineering vidson.pdf Center Hotspots indexing project 2005 Sub- Global Columbia http://www.ldeo.columbia.edu/ national University and chrr/research/hotspots/ the World Bank Hurricane Disaster Risk 2001 County USA Davidson & http://scitation.aip.org/getabs/se Index (HDRI) Lambert rvlet/GetabsServlet?prog=norm al&id=NHREFO00000200000 3000132000001&idtype=cvips &gifs=yes&ref=no Livelihood Vulnerability 2008 District AFR - CARE- http://pine.sage.wisc.edu/pubs/a Index (LVI) Mozambique Mozambique, rticles/F- Emory L/Hahn/hahn2009GEC.pdf University Risk Index for Megacities 2006; City global Munich Re www.munichre.com/publicatio regularly ns/302-04271_en.pdf Social Vulnerability 2005, Metro USA Lisa Rygel, http://www.springerlink.com/co Index repeat Region David O’sullivan ntent/h011437x16330471/ and Brent Yarnal 102 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Appendix B: Comparison of MHCRI with Other Risk Indices There are numerous studies of disaster risk, hazard and vulnerability, conducted through a variety of approaches. This Appendix reviews some of that are based on an index approach. Cutter et al (2003) developed a social vulnerability index at the county level for the United States, based upon 11 factors that accounted for 76% of the variance. The factors included various infrastructure and social factors, including race. An additive model was used to accumulate vulnerability, with equal weights being given to the various factors (due to the difficulty and subjectivity of assigning weights for many counties at a large scale). We have adopted the additive vulnerability approach used by Cutter, but provide the option of assigning weights. Some factors of relevance to the Cutter study, such as racial background, are not used because they are not applicable beyond Cutter’s area of study. The UNDP has developed a Disaster Risk Index (Peduzzi et al., 2009) that, by setting risk values for the period 1980-2000, can be used as a reference for estimating future trends in risk. It is based upon EM-DAT fatality data from droughts, earthquakes, tropical cyclones and floods, as a proxy for risk and operates at a country scale. Risk is defined as a multiplicative function: Risk = (hazard frequency) x (population) x (vulnerability). Vulnerability is quantified using an indicators approach involving 32 socio-economic and environmental variables, with risk being calculated via a correlation analysis. This is a very restrictive proxy for risk that excludes economic and most social impacts, and differs from MHCRI which does not assume a priori estimates of risk that can be used to generate risk equations. MHCRI uses a similar definition of risk, except for creating a more sophisticated hazard index. The major difference is that we define risk as a dependent variable instead of using EM-DAT data as a proxy, since the EMDAT data is subject to a large number of significant biases and trends, and excludes potentially serious disasters that have not yet occurred. As well, using only fatality data as a proxy for risk is far too restrictive for the purposes of MHCRI, which considers risk to include broad-based social and economic aspects. A Global Urban Risk Index (Brecht, 2007) for cities with populations over 100,000 used a multiplicative definition of risk based upon hazards (earthquakes, volcanoes, landslides, floods and cyclones), exposed elements (city population and GDP) and vulnerability (EM-DAT death and economic loss data). Empirical data is used as a proxy to estimate vulnerability, similar to the UNDP approach for risk. Note that EM-DAT data for economic loss are particularly suspect, and should be used with caution (Guha-Sapir and Below, 2005). Again, we have used a similar definition of risk, but differ in how vulnerability is estimated. . An urban earthquake disaster risk index was derived by Davidson (1997) and operates at a city scale. It uses a weighted additive approach where Risk is a function of dimensionless indicators of hazard, exposure, vulnerability, external context and emergency 103 Methodology Report: Calculating Multi-hazard City Risk response and recovery. The significant difference in methodology with this approach, compared to MHCRI, is the additive as opposed to multiplicative calculation of risk. This study is unique in that regard, and appears to run counter to most disaster theory. The Environmental Vulnerability Index (EVI) works at a country scale and uses an additive approach based on 32 indicators of hazards, 8 of resistance and 10 of damage (Kaly et al., 2005). It is designed to be used with economic and social vulnerability indices to provide insights into the processes that can negatively influence the sustainable development of countries. The hazard indicators relate to the frequency and intensity of hazardous events. The resistance indicators refer to the inherent characteristics of a country that would tend to make it more or less able to cope with natural and anthropogenic hazards. Damage indicators relate to the vulnerability that has been acquired through loss of ecological integrity or increasing levels of degradation of ecosystems. All of the EVI’s indicators are transformed to a common scale so that they can be combined by averaging, and to facilitate the setting of thresholds of vulnerability. This approach is conceptually very similar to the way vulnerability is calculated in the MHCRI. The main difference is the way that hazard is included, which here is considered to be a separate vulnerability factor that can be scaled and added to other factors. We prefer to use the more standard definition of risk that incorporates hazard in a multiplicative way. The Central American Probabilistic Risk Assessment (CAPRA) is based upon four indices (Cardona, 2005). Of relevance to this project for comparisons are the Local Disaster Index and the Prevalent Vulnerability Index:  “The Disaster Deficit Index measures country risk from a macroeconomic and financial perspective according to possible catastrophic events. It requires the estimation of critical impacts during a given period of exposure, as well as the country’s financial ability to cope with the situation.�  “The Local Disaster Index identifies the social and environmental risks resulting from more recurrent lower level events (which are often chronic at the local and subnational levels). These events have a disproportionate impact on more socially and economically vulnerable populations, and have highly damaging impacts on national development.� This index is additive in nature, and is based upon empirical records of deaths, people affected and losses at a municipal level. Because it uses historical data to estimate the disaster index, this methodology differs from MHCRI.  “The Prevalent Vulnerability Index is made up of a series of indicators that characterize prevalent vulnerability conditions reflected in exposure in prone areas, socioeconomic weaknesses and lack of social resilience in general.� This index is additive, and is based upon a set of weighted environmental, social and resilience indicators. This index is very similar to how MHCRI estimates vulnerability, though without the emphasis on 104 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project infrastructure.  “The Risk Management Index brings together a group of indicators that measure a country’s risk management performance. These indicators reflect the organizational, development, capacity and institutional actions taken to reduce vulnerability and losses, to prepare for crisis and to recover efficiently from disasters.� Rygel et al (2006) constructed a social vulnerability index to storm surge using a composite index of poverty, gender, race, ethnicity, age and disabilities. They attempted to avoid issues related to assigning weights through the use of principal component analysis of census data and Pareto ranking. They ended up with 3 components accounting for 51% of the variance in the data in their case study. Note that this method is case specific, in that different geographical regions would have different vulnerability components. This method is not appropriate for MHCRI because it would need to create a different set of indices for every location, using a principal component analysis. Comparability would therefore be severely compromised. Hahn et al (2009) discuss the CARE Livelihood Vulnerability Index (LFI) as applied to climate variability and change in Mozambique, using socio-demographics of livelihoods, social networks, health, food and water security, natural disasters and climate variability. From this they develop a composite vulnerability index through an additive approach where all factors are equally weighted. The unit of analysis is household data from local surveys. The methodology used in this study is not appropriate to MHCRI, given its requirement for large-scale household level surveys. The additive approach is similar to the vulnerability calculations in MHCRI, though we include weightings. Simpson (2006), in a thorough review of issues related to the creation of indices, proposes a Disaster Resiliency/Preparedness Index (Dpi) that combines hazard probability and frequency with various measures of vulnerability in a multiplicative formula. The proposed methodology for constructing vulnerability is additive and uses weights. A useful list of possible indicators is listed in an appendix. The methodology is conceptually very similar to MHCRI, except that Simpson uses the term ‘vulnerability’ where MHCRI uses the term ‘risk’. The World Wildlife Federation (WWF, 2009) ranked climate vulnerability for a set of coastal cities in Asia, using exposure, sensitivity, and adaptive capacity. These three categories were then averaged to calculate an overall score. “Exposure is the average of the three highlighted environmental categories including the susceptibility of the city to be impacted by 1 m sea-level rise, historical frequency of extreme weather events, including flooding and drought, and frequency of tropical storms and surges. Sensitivity is based on population, GDP, and the relative importance of that city to the national economy. Adaptive capacity is calculated by examining the overall willingness of the city to implement adaptation strategies (calculated by the available adaptation examples and/or responses to previous impacts) and the 105 Methodology Report: Calculating Multi-hazard City Risk per capita GDP.� This methodology is different than MHCRI in that hazard data is added into the vulnerability calculation and uses equal weights for the three categories of exposure, sensitivity and adaptive capacity. In that sense it contradicts the basic risk theory on which MHCRI is based, where risk is multiplicative of hazard, exposure and vulnerability. German Watch93 publishes an annual climate risk index based upon historical socio-economic impacts of extreme weather events, not including fatalities and number affected. It only addresses direct impacts and operates at a country scale. This is quite unlike MHCRI, as it is only empirically based using economic impact data, and is not an appropriate approach to calculating a broad based disaster risk index. Baettig et al. (2007) developed a climate change index, based upon IPCC scenarios of annual temperature and precipitation changes and changes in 20 year return period extreme events, including droughts. These are hazard indices only and do not consider vulnerability or impacts and are calculated using a weighted mean of the indictors. This study could potentially provide some guidance in terms of a 2030 scenario, but is not relevant to our methodology. The studies described in this section illustrate the variety of approaches that can be taken in calculating a risk or vulnerability index. Based on several of the approaches described above that are best suited for a multi-hazard city risk index, MHCRI applies the following approach:  All variables are converted to dimensionless numbers  Risk is defined as a multiplicative function of hazard, exposure and vulnerability (RI=HIxEIxVI)  Hazard is defined as a multiplicative function using physical parameters.  Vulnerability is defined as an additive function using a sectoral approach.  Vulnerability calculations are based upon social, economic, environmental and infrastructure variables that determine potential damage or harm, as opposed to using historical disaster impact data as a proxy.  Vulnerability variables are weighted. 93 http://www.germanwatch.org/klima/cri.htm 106 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Appendix C: Hazard Indices and Climate Change94 94 This Appendix was prepared by David Etkin, Coordinator of the Program in Emergency Management at York University in Toronto, and Kaz Higuchi, Research Scientist at Environment Canada and Adjunct Professor in the Faculty of Environmental Studies, York University. 107 Methodology Report: Calculating Multi-hazard City Risk This Appendix illustrates how Hazard Indices were calculated for typhoons, monsoons, drought, and storm surge in the three pilot cities, and explains issues related to flood, sea level rise, heat waves, and the Gumbel distribution. Typhoon Hazard The parameter H1 =1, since the spatial scale of typhoons greatly exceeds city size. Estimates of the probability of typhoon occurrence (H3, H5) of different magnitudes (H2, H4) can be made through the use of historical data, such as plotted on Figure 44. With respect to the cities currently being evaluated, this figure shows that typhoons are rare in Thailand, more common on the east China coast, and frequent near the Philippines. Figure 44: Tropical Cyclone Tracks from the beginning of records to September 2006 Source: http://www.physicalgeography.net/fundamentals/7u.html) The three cities of Bangkok, Manila and Ningo were assessed for typhoon risk. Specific typhoon track information is available from NOAA, and lists of typhoons and their impacts are available from EMDAT and various papers that have been published in academic and institutional literature. A gridded Western North Pacific typhoon data set, which includes wind speeds, was used to generate the statistics in Table 6 and is available from the US Naval Oceanography Portal (JTWC Dataset) at http://www.usno.navy.mil/NOOC/nmfc- ph/RSS/jtwc/best_tracks/wpindex.html. Of particular relevance is the number of typhoons passing within 200- 250 km of the city (Chavas and Emanuel, 2010). The JTWC dataset was sorted for typhoon observations within 2 degrees latitude/longitude for each city, which corresponds to the range identified above. The maximum typhoon wind speed95 for each year was selected and used as an analysis variable (precipitation data are not available). Over the 41 years analyzed, Bangkok had 6 years with typhoon observations within the 2 degree box, Ningbo had 17 and Manila had 41. The most 95 Wind speeds in the database are estimated to the nearest 5 knots, but converted to km/h for this analysis 108 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project frequent and strongest typhoons occurred near Manila, followed by Ningbo and then by Bangkok (Table 6 and Figure 45). MANILA NINGBO BANGKOK N of Cases 41 17 6 Minimum (km/h) 28 74 28 Maximum (km/h) 259 176 46 Mean (km/h) 154 117 37 Standard Deviation 63 32 8 (km/h) Frequency of years 100% 41% 15% with typhoons Table 6: Statistics of Typhoon Annual Maximum Wind Speed (km/h), within Two Degrees of City (1970-2010) using JTWC Dataset Note: statistics include only years with typhoon occurrence 300 250 200 Wind Speet (km/h)  150 MANILA BANGKOK NINGBO 100 50 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentile Figure 45: Percentiles of Typhoon Maximum Wind Speeds for Manila, Bangkok and Ningbo, for years when Typhoons Occur 109 Methodology Report: Calculating Multi-hazard City Risk Figure 45 indicates that, for years when typhoons do occur, equivalent statistical wind speeds at the 30th percentile or higher at Bangkok are about 20% of those at Manila, while Ningbo speeds are about 70% of the Manila ones. Only years with typhoons were included in the calculation. The methodology to calculate design typhoon magnitudes is based upon an extreme value analysis using (1) a 30-year return period event (frequency = 0.033) or the storm of record, and (2) a 2-year return period event (frequency = 0.5). It was not possible to do an extreme value analysis for Bangkok or Ningbo because of the large number of non-typhoon years, and therefore the maximum estimated typhoon wind speed (i.e. storm of record) from the JTWC dataset was used to estimate the PME for (1), while the 50th percentile wind speed was used for (2). Wind frequencies for the storm of record for Ningbo and Bangkok were estimated by dividing the number of years with occurrences of that wind maximum by the number of years of data (Bangkok = 2/41, and Ningbo = 1/41). Table 7 shows Manila return period wind speeds calculated using a Gumbel distribution. Manila Maximum Typhoon Wind Return Period Speed (km/h) (years) 2 145 5 164 10 238 20 273 30 295 Table 7: Manila Typhoon Return Period of Annual Maximum Wind Speeds The wind speeds were then used to estimate H2 and H4 using the Saffir-Simpson Hurricane Scale (Table 8). Note that H2/H4 remains at 10 for all speeds above 249 km/h. The resulting Hazard Indices for typhoons for the three cities are shown on Table 9. Minimum Hurricane wind speed H2 Scale Scale (km/h) 5 249 10 4 210 8 3 178 6 2 154 4 1 119 2 Tropical 63 1 110 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Storm Tropical 0 0 Depression Table 8: Saffir-Simpson Hurricane Scale as compared to H Scale Manila  Ningbo  Bangkok              H1  H2/H4  H3/H5  HI  H1  H2/H4  H3/H5  HI  H1  H2/H4  H3/H5  HI  Typhoon  PME (30 yr  1.0  0.7  0.05  0.04  1.0  10.0  0.033  0.33  1.0  5.8  0.024  0.14  or storm of  record)  Typhoon  1.0  0.0  0.5  0.00  1.0  3.5  0.5  1.8  1.0  1.7  0.5  0.85  (2 yr)  Table 9: Summary of HI Typhoon Calculations for Bangkok, Manila and Ningbo (based upon annual maximum wind speeds) Note that the hazard indices for Manila and Ningbo are larger for the more frequent 2-year event than for the 30 year event. This is because, although the 2-year events are weaker they are multiplied by their frequency, which is much larger. This does not necessarily mean that they create more risk to the city however, since that can only be estimated after inserting vulnerability into the calculation. Research on how climate change will affect tropical cyclones is unclear, though there is some evidence that they may become more frequent or severe (Knutson et al., 2008, Knutson et al., 2010). Knutson et al. (2010), in a review of model outputs, note a global decrease in frequency by 6-34%, but with a corresponding increase towards stronger storms of 2-11% by 2100. Füssel (2009) says that “In summary, substantial uncertainties about past and future changes in cyclone activity remain, and the scientific debate on this subject is expected to remain very active.� Murakami et al (2011) found the following changes in North Pacific Typhoons using a high resolution model: “The future (2075-99) projection indicates (i) a significant reduction (by about 23%) in both TC genesis number and frequency of occurrence primarily during the late part of the year (September-December), (ii) an eastward shift in the positions of the two prevailing northward-recurving TC tracks during the peak TC season (July-October), and (iii) a significant reduction (by 44%) in TC frequency approaching coastal regions of Southeast Asia.�, and “fewer TCs will form in the western portion of the WNP (west of 145°E), whereas more storms will form in the southeastern quadrant of the WNP (10°-20°N, 145°-160°E).� In the cities under consideration, typhoon occurrence will depend upon how climate change affects sea surface temperature and ENSO events, which is not well understood. Given the findings of the literature review, it is reasonable to assume a 2030-low of 111 Methodology Report: Calculating Multi-hazard City Risk fewer typhoons (about 10%) with little change in intensity, and a 2030-high of no change frequency (0%), but a slight increase in intensity (about 2%). Monsoon Hazard Each of the three cities experiences a summer monsoon season (Figure 46) with increased rainfall, though the period of maximum rainfall varies between the cities. Because the scale of monsoon is much larger than city scale H1=1. To evaluate the monsoons, 3- month total rainfall during the peak of monsoon season was calculated from weather stations located within the city, with the exception of Ningbo, which had insufficient data. This statistic was chosen instead of another circulation-type monsoon index because it is physically more relevant to flooding. Figure 46: Asian Monsoon Wind Patterns There is no standard scale for comparing monsoon precipitation (as there is for typhoons), so a 3-month maximum of 15,000 mm was selected based upon data96 from an area in northern India that has the world’s most extreme recorded monsoon precipitation. Tables 10-12 and Figure 47 below summarize the data and HI calculations. 96 Source: Indian Institute of Tropical Meteorology ftp://www.tropmet.res.in/pub/data/rain/iitm-regionrf.txt. Data from NEIND1871R, 1871-2006. 112 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project MANILA BANGKOK Northern (July, (August, INDIA August, September, (June, July, September) October) August) N of Years 25 46 138 Minimum (mm) 888 355 8,911 Maximum (mm) 1,736 1,503 15,218 Arithmetic Mean 857 758 11,324 (mm) Standard Deviation 396 236 1,155 (mm) Table 10: Summary statistics of maximum 3-month precipitation (mm) during monsoon 2000 1800 1600 Monsoon Precipitation (mm) 1400 1200 1000 MANILA 800 BANGKOK 600 400 200 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentiles Figure 47: Percentiles of 3-month Monsoon Precipitation at Manila and Bangkok 113 Methodology Report: Calculating Multi-hazard City Risk Bangkok:Manila    Bangkok  Manila  Ratio  Amount  Amount  Return Period  (mm)  (mm)  %  (yrs)  2 747 794  94% 5 816 914  89% 10 1081 1376  79% 20 1207 1594  76% 30 1281 1728  74% Table 11: Return Periods of 3-month Monsoon Precipitation Totals for Manila and Bangkok, calculated using a Gumbel distribution Manila  Bangkok          H1  H2/H4  H3/H5  HI  H1  H2/H4  H3/H5  HI  Monsoon  (30 year  1.0  0.85  0.033  0.03  1.0  1.15  0.033  0.04  PME)  Monsoon  1.0  0.48  0.5  0.24  1.0  0.53  0.5  0.27  (2 yr)   Table 12: Summary of HI Monsoon Calculations for Bangkok and Manila based upon 3- month precipitation totals The effect of climate warming on monsoons appears to be a likely intensification, though with a high degree of uncertainty. Annamalai et al. (2007) note, on an analysis of monsoons in climate models, “From these “best� models in the double CO2 simulations there are increases in both the mean monsoon rainfall over the Indian subcontinent (by 5%–25%) and in its interannual variability (5%–10%). … This result, though plausible, needs to be taken with some caution because of the diversity in the simulation of ENSO variability in the coupled models that have been analyzed.� Although Kripalani et al. (2007) have found that “Extreme excess and deficient monsoons are projected to intensify“ these results should be taken with caution as noted by Bollasina and Nigam (2009); “In our opinion, current models cannot provide durable insights on regional climate feedbacks nor credible projections of regional hydroclimate variability and change, should these involve ocean–atmosphere interactions in the Indian basin�. Given the strong dynamical connection between monsoons and ENSO events, this cautionary note has been reinforced recently by a review article by Collins et al. (2010) who stated that “despite considerable progress in our understanding of the impact of climate change on many of the processes that contribute to El Niño variability, it is not yet possible to say whether ENSO activity will be enhanced or damped, or if the frequency of events will change.� The above suggests the following changes in monsoon precipitation: a 114 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 2030-low scenario of +2% and a 2030-high scenario of +12%, for given return periods. A summary of the effects of climate change on monsoon are given in Table 13. Effect of  Bangkok  Manila  Climate  Ningbo  Change (low)              H1:  H2:   H4:  HI  H1:  H2:   H4:  HI  H1:  H2:   H4  HI    Monsoon   0.254  0.270  Insufficient  0%  Insufficient  1.000  2% 0%  1.000  2% 0%  1.000  Data  Data  (2 yr)  Monsoon   0.029  0.039  Insufficient  0%  Insufficient  1.000  2% 0%  1.000  2% 0%  1.000  Data  Data  (30 yr)  Typhoon   0.000  1.588  �10%  0.763  1.000  0%  10%  1.000  0%  �10%  1.000  0%  (2 yr)  Typhoon   0.032  0.297  �10%  0.128  1.000  0%  10%  1.000  0%  �10%  1.000  0%  (30 yr)    Bangkok  Manila  Effect of  Ningbo  Climate      Change (high)          H1:  H2:   H4:  HI  H1:  H2:  H4:  HI  H1:  H2:   H4:  HI    Monsoon   Insufficient  Insufficient  1.000  12%  0%  0.279  1.000  12%  0%  0.296  1.000  Data  0%  Data  (2 yr)  Monsoon   Insufficient  Insufficient  1.000  12%  0%  0.032  1.000  12%  0%  0.043  1.000  Data  0%  Data  (30 yr)  Typhoon   1.000  2%  0%  0.000  1.000  2%  0%  1.800  1.000  2%  0%  (2 yr)  0.864  Typhoon   1.000  2%  0%  0.036  1.000  2%  0%  0.337  1.000  2%  0%  0.145  (30 yr)  Table 13: Summary of the Climate Change Effect on HI Calculations Drought Drought is a slow onset, creeping hazard that has no universal definition. It can be thought of as a prolonged period of dry weather that depletes water resources needed for human and environmental needs. There are three kinds of drought, meteorological (related to rainfall amounts), hydrological (related to water levels) and agricultural (related to water availability in the soil for crops). There are various indices devised to measure drought, based upon rainfall anomalies, hydrological measurements or soil moisture. Examples are the Standardized Precipitation Index, the Palmer Drought Severity Index (PDSI), the Crop Moisture Index and Deciles97. Some of these methods define drought relative to local climatic conditions and 97 see http://www.drought.gov/portal/server.pt/community/drought_indicators/223/crop_moisture_index/277 115 Methodology Report: Calculating Multi-hazard City Risk therefore cannot be used for comparative purposes on a global scale. Others require long time series of data and running complex algorithms or models. The National Drought Mitigation Center describes PDSI as “a meteorological drought index, and it responds to weather conditions that have been abnormally dry or abnormally wet. When conditions change from dry to normal or wet, for example, the drought measured by the PDSI ends without taking into account streamflow, lake and reservoir levels, and other longer-term hydrologic impacts. The PDSI is calculated from precipitation and temperature data, as well as the local available water content of the soil. From these inputs, all the basic terms of the water balance equation can be determined, including evapotranspiration, soil recharge, runoff, and moisture loss from the surface layer. Human impacts on the water balance, such as irrigation, “are not considered� (Hayes, n.d.). The National Centre for Atmospheric Research (NCAR), Climate Analysis Section provides a global drought dataset98 (Dai et al., 2004) based upon the Palmer Drought Severity Index99; we have used this dataset for comparative purposes. However it must be noted that this index, which uses temperature and rainfall to determine soil moisture (developed by Palmer, 1965) works best for homogeneous non- mountainous regions experiencing long-term drought (Alley, 1984). The number 0.0 represents normal conditions, with drought being represented by negative numbers: -2.0 being moderate drought, -3.0 severe drought, and -4.0 extreme drought. It must be noted that these values were assigned within the context of U.S. climate, and may not be as representative of tropical climates. Regional drought frequencies are highly variable across the region (Figure 48; Yusuf and Francisco, 2009), as a result of variations in topography, geography relative to storm tracks, etc. 98 This dataset is available for download at http://www.cgd.ucar.edu/cas/catalog/climind/pdsi.html 99 See http://www.drought.noaa.gov/palmer.html for the NOAA description of its use. 116 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Figure 48: Drought frequency (events/yr 1980-2000) The NCAR Global Palmer Drought Data Severity Index was used to estimate drought hazard for the three cities. A grid point over land closest to the city location was chosen as a representative value. A time series of the monthly drought indices from January 1970 – December 2005 is shown in Figure 49, and the associated histograms in Figure 50. Indices near Ningbo should be interpreted with caution, due to the sparseness of the meteorological data from the weather station used to represent the meteorological condition of the city. The data suggests that all three locations have experienced severe drought events with PDSI values lower than -4.0 (Bangkok about 18% of the time and Manila about 9% of the time). 117 Methodology Report: Calculating Multi-hazard City Risk Figure 49: Time series of monthly PDSI values (1970-2005) Source: Data from NCAR Figure 50: Histograms of monthly PDSI (1970-2005) Source: Data from NCAR. 118 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Percentile values of the PDSI are shown in Figure51. 50th percentile values indicate that drought does not occur often enough at any of the cities to be of concern at a 2-year return period. Rare events are of potential significance however. Percentiles of PDSI 1% 5% 10% 20% 25% 30% 40% 50% 0 �1 �2 �3 BANGKOK PDSI MANILA �4 NINGBO �5 �6 �7 �8 Figure 51: Figure 8: Percentiles of the PDSI for Bangkok, Manila and Ningbo Note: Results for Ningbo should be taken with caution due to incompleteness of the meteorological data – the 1% value is not included for this reason. Data from NCAR. As well as having impacts on agriculture in the Philippines, Manila has experienced blackouts due to drought, as a result of hydroelectric power shortages (Conde, 2010). Bagalihog and Siapno (n.d.) indicate that as well as impacting crops and fisheries, droughts that normally occur in El Niño years have contributed towards forest fires and water shortages in Metro Manila. Severe droughts during the 1997-98 event resulted in 24 consecutive months with PDSI indices below -2.0 at Manila, going as low as -7.9 (with the exception of November 1996, which had a PDSI of -1.5). Drought associated with the 1982 El Niño event had a run of PDSI values below -2.0 from August 1981 to May 1984, with values dropping to -7.23 in the summer of 1983. 30-year return period PDSI values are: Bangkok -7.98, Manila -7.90 and Ningbo -5.54. Various papers such as Dai (2010) use extreme PDSI values of -20, so we have chosen that as the drought reference value in our calculation. Table 14 shows the HI calculations for 30-year droughts based upon the PDSI. Bangkok Manila Ningbo H1 H3 H5 HI H1 H3 H5 HI H1 H3 H5 HI 30 year 3.9 3.9 PDSI 1.0 9 0.033 0.13 1.0 5 0.033 0.13 1.0 2.77 0.033 0.09 Table 14: HI calculations for drought based upon PDSI 119 Methodology Report: Calculating Multi-hazard City Risk The frequency and/or intensities of droughts could shift due to changes in precipitation patterns under climate change. Dai (2010) in a review of drought under global warming notes that in the past 500 years there have been dry periods that lasted for decades over East Asia, probably triggered by anomalous sea surface temperatures, and that global PDSI trends over the past 30 years have been downwards by about - 0.07 . He also notes that “Climate models project increased aridity in the 21st century over … Southeast Asia.� Burke et al (2006) and Burke and Brown (2008) found, using the Hadley GCM, that drought in southeast Asia increased from 10-18%, though south and central Asia had a decrease in drought. Dai (2010) also found a general increase in aridity (Figures 52, 53). At a local level uncertainties are high due to the importance of local feedbacks and the difficulty GCM models have in resolving precipitation, monsoons, ENSO events and typhoons in the western Pacific. In this regard, Arnel (n.d.) notes the large difficulty in providing quantitative estimates; thus the PDSI estimates in this report must be taken with caution. Figure 52: East Asia PDSI (2030-2039) (adapted from Dai, 2010) Note: Note the drying in east China and Thailand, and wetter conditions in the Philippines. Mean annual PDSI for years 2030–2039 calculated using the 22-model ensemble-mean surface air temperature, precipitation, humidity, net radiation, and wind speed used in the IPCC AR4 from the 20th century and SRES A1B 21st century simulations. Red to pink areas are extremely dry (severe drought) conditions while blue colors indicate wet areas relative to the 1950–1979 mean. Figure 53: Drought in East Asia for an ensemble of models for doubled carbon dioxide Source: adapted from Burke and Brown, 2008 Note: Spatial distribution of the likelihood of increase or decrease of moderate drought for a multi-model experiment. Locations where more than 70% of the ensemble members show a decrease (increase) in moderate drought are in blue (red). Places where less than 70% of the ensemble members agree on either an increase or a decrease are in gray. 120 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Model outputs do not provide information regarding changes in the 30- year return period droughts for 2030, and due to the complexity of the problem and strong nonlinearity in the climate response, quantitative estimations of that statistic are beyond the scope of this project. Current literature does, however, suggest that drought could become more severe in Bangkok and Ningbo in the future, but not in Manila. Storm Surge Winds around a typhoon act on the water surface, through friction, to induce a rise in sea level. Stronger winds and lower pressures result in greater surges, as shown in Figure 54. These surges occur to the right of the storm track direction where storm-relative wind speed and storm- motion speed combine to create maximum surface winds (Figure 55), and therefore the impact of storm surge is very dependent upon the path of a storm. In the Philippines typhoon paths are mainly from east to west and therefore the eastern coast is exposed to storm surge hazard, whereas the western coast where Manila is located, is not. Bangkok, at the north end of the Gulf of Thailand, is also not very vulnerable to storm surge because it is exposed to only very few typhoons. By contrast, Ningbo, on the east coast of China, is exposed to storm surges. Climate change will alter storm surge risk in the same way that typhoon risk is altered. Figure 54: Storm Surge Height as a Function of Hurricane Wind Speed (Franck, 2009) Figure 55: Effect of Storm Motion on Surface Wind Speeds Note: In the northern hemisphere winds are stronger to the right of storm track, since the storm motion and storm-relative wind speeds combine. This produces an area with the largest storm surge. Winds are weakest to the left of storm track, since the storm motion diminishes the storm-relative wind speed. 121 Methodology Report: Calculating Multi-hazard City Risk Historical storm surge information is available for the cities of Bangkok (Songklanakarin, 2009) and Manila (Muto et al, 2010), though they are not at high risk. A historical value for Ningbo was estimated from Figure X1 above (Franck, 2009), using a wind speed of 176 km/hr, which is the maximum wind speed listed in the typhoon data set (available from the US Naval Oceanography Portal (JTWC Dataset) at http://www.usno.navy.mil/NOOC/nmfc- ph/RSS/jtwc/best_tracks/wpindex.html) within 2 degrees latitude and longitude100; Table 15 summarizes the HI calculations. Bangkok Manila Ningbo H1: Fraction 0.4 0.4 0.4 H3: Storm Surge 0.76 1.14 3.13 H5: Probability 0.018 0.017 0.025 HI: 0.0055 0.0076 0.0313 Reference height (m) 8 8 8 Historical Value (m) 0.61 0.91 2.5 N 55 60 40 Table 15: Storm Surge HI Calculations The Bangkok historical event resulted from Typhoon Linda in 1997 and in Manila from a typhoon in 1957. It was unclear in Muto et al. (2010) how many years of data were assessed, and therefore H5 of 0.017 must be considered an estimate. Sea Level Rise Sea levels have changed dramatically in the past (Figure 56) as a result of changes in climate, and are expected to rise over the coming century due to melting of land ice and thermal expansion of water, as noted in the latest IPCC assessment (Solomon et al., 2007). This, combined with subsidence, has the potential to significantly affect flood risk in many major coastal cities. Figure 56: Sea Level Changes Since the Last Glacial Maximum Source: Image created by Robert A. Rohde / Global Warming Art" http://www.globalwarmingart.com/wiki/File:Post- Glacial_Sea_Level_png, Data from Fleming et al., 1998 100 Note that the Ningbo storm surge height is not directly observed at the city location, as are the observations for Bangkok and Manila, but is inferred from modeled wind speeds within a square, 2 degrees latitude by 2 degrees longitude, associated with the JTWC typhoon dataset. 122 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Low lying cities near the coast are particularly at risk, which includes parts of Bangkok, Ningbo and Manila. The 2007 IPCC assessment produced a probabilistic estimate of global average sea level rise (Figure 57). Though it can vary regionally (Figure 14) changes are expected to be minimal near the cities of Bangkok, Manila and Ningbo. By 2030, predictions of the 90% confidence limits of the sea level rise (SLR) will likely range from about 0.05 m to 0.15 m. Some recent publications suggest that the IPCC estimate is too low (Rahmstorf, 2007; Jevrejeva, 2010), perhaps by a factor of 3 or more. These estimates do not include possible catastrophic scenarios (which are considered unlikely within the next century), such as the slippage of part of the Antarctic ice sheet into the ocean. Figure 57: Projected sea-level rise for the 21st century Source: http://maps.grida.no/go/graphic/projected-sea-level-rise-for-the-21st-century The projected range of global averaged sea-level rise from the IPCC 2001 Assessment Report for the period 1990 to 2100 is shown by the lines and shading. The updated AR4 IPCC projections made are shown by the bars plotted at 2095, the dark blue bar is the range of model projections (90% confidence limits) and the light blue bar has the upper range extended to allow for the potential but poorly quantified additional contribution from a dynamic response of the Greenland and Antarctic ice sheets to global warming. Note that the IPCC AR4 states that “larger values cannot be excluded, but understanding of these effects is too limited to assess their likelihood or provide a best estimate or an upper bound for sea-level rise.� The inset shows the observed sea levels from tide gauges (orange) and satellites (red) are tracking along the upper bound of the IPCC 2001 projections since the start of the 123 Methodology Report: Calculating Multi-hazard City Risk projections in 1990. Changes in sea level are not incorporated into the MHCRI model at the hazard input stage, but can impact vulnerability in the 2030 scenarios in terms of physical susceptibility to flooding. Heat Waves (Extreme Heat as a natural hazard has not been recognized to the same level as Temperature other severe meteorological events, such as hurricanes (typhoons), thunderstorms and tornadoes (Sheridan, 2004). One of the reasons for this is that there is no universally accepted definition of a heat wave. Definitions vary from study to study and from region to region; this is consistent with the fact that human thresholds to heat do in fact show spatio-temporal variations on a global basis. Assessment of human vulnerability to heat is much more than just a function of maximum temperature and humidity. It needs references to human impacts (Souch and Grimmond, 2004). The following is a meteorologically-based and “objective� definition of the occurrence of a heat wave recommended by the World Meteorological Organization: “the daily maximum temperature of more than five consecutive days exceeds the average maximum temperature by 5 Celsius, the normal period being 1961–1990�. Frich et al. (2002) used this definition to derive the Heat Wave Duration Index (HWDI) to monitor changes in extreme temperature events worldwide. The authors of the study, however, do acknowledge that “such an index has little value in warm climates with low day-to-day variability� (Souch and Grimmond, 2004). Maritime tropical climates that have very low temperature variability, such as Bangkok and Manila, are not susceptible to heat waves. Ningbo does apparently experience this hazard, but due to a lack of meteorological data, its frequency could not be properly assessed. Sheridan (2004) developed the Spatial Synoptic Classification (SSC) method that assigns, at a given location on a certain day, one of eight different weather types. Each weather type is associated with a certain level of human health. Climate change is highly likely to increase the frequency and intensity of heat waves, and their occurrence responds in a very non-linear way to changes in mean temperature (Columbo et al, 1999). However, a rise in the mean temperature does not necessarily result in a rise of heat wave events (Frich et al., 2002). For China, instrumental records over the past half century have shown an increase in mean summertime temperatures, with nighttime temperatures increasing more than daytime temperatures (Wang and Gaffen, 2001). They also show an increase in the number of extremely hot and humid days, with heat waves lasting several days. In contrast, it is interesting to note that southern China has been experiencing shorter heat waves (Frich et al., 2002). On a global basis, however, a warmer world would likely lead to an increase in heat wave duration, resulting in significantly increased heat vulnerability. Tol (2002), for example, suggested that for very 1oC increase in the global mean temperature, about 350,000 people could die from heat-related cardio-vascular and respiratory problems. Heat vulnerability could become a significant issue for Ningbo as climate warming is compounded by the urban heat island effect as the city’s 124 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project population continues to grow. Annex 1: Extreme value The Gumbel distribution is a special case of the Generalized Extreme analysis (Gumbel Value distribution (Gumbel, 1958). It is a relatively common probabilistic Distribution) model used in the environmental sciences to model extreme values associated with such environmental extreme events flooding and wind gusts and is recommended by the National Building Code of Canada. Extreme values are selected as maximum or minimum values from sets of data. For example, the annual maximum rainfall at a location constitutes the largest recorded value during a year. A series of these annual maximum rainfall values will then make up a time series of extreme values that can be analyzed statistically. Probability distributions of these extreme values will converge to one of three forms of extreme value distributions, called Types I, II and III. The Gumbel distribution is a Type I extreme value distribution (Figure 58). One of the attractive features of the Gumbel distribution is that its parameter equations provide an estimate of the mode. From the Gumbel probability distribution equation, one can also derive a simple equation, with certain assumptions, to estimate return periods of any extreme natural events. All calculations of return periods were done in MATLAB using this distribution. Summaries of the Gumbel distribution can be found on a number of web sites, including the Engineering Statistics Handbook101. Figure 58: Type I Gumbel distribution for the Maximum Case Source: Engineering Statistics Handbook 101 http://www.itl.nist.gov/div898/handbook/eda/section3/eda366g.htm 125 Methodology Report: Calculating Multi-hazard City Risk 126 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Appendix D: Spreadsheet Structure of the MHCRI Model 127 Methodology Report: Calculating Multi-hazard City Risk 128 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Appendix E: Indices for Bangkok, Manila and Ningbo 129 Methodology Report: Calculating Multi-hazard City Risk 130 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 131 Methodology Report: Calculating Multi-hazard City Risk 132 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project 133 Methodology Report: Calculating Multi-hazard City Risk 134 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Appendix G: Project This project was prepared through the efforts of a team comprised of Bank staff and Consultants with valuable inputs from an International Team Advisory Group of specialists from around the world. World Bank World Bank Project Team: Fatima Shah, Team Leader, East Asia Infrastructure Sector Unit Henrike Brecht, East Asia Infrastructure Sector Unit Yuri Dikhanov, Development Economics Data Group Federica Ranghieri, World Bank Institute Urban Unit Peter Jipp, Southeast Asia Sustainable Development Unit Chris Pablo, Philippines Sustainable Development Unit Cathy Vidar, Philippines Sustainable Development Unit World Bank Management: John Roome, Sector Director, East Asia Sustainable Development Department Vijay Jagannathan, Sector Manager, East Asia Infrastructure Sector Unit World Bank Final Review Committee: Uwe Deichmann Dan Hoornweg Stuart Gill Abedalrazq Khalil Abhas Jha Additional World Bank Peer Reviewers: Sudipto Sarkar Edward Anderson Tony Bigio Niels B. Holm-Nielsen George Soraya Veasna Bun Additional Feedback and Support in the World Bank From: Victor Vergara Stephane Hallegatte Alexander Lotsch Judy Baker Zoe Trohanis Yoshiyuki Imamura Oscar Ishizawa Escudero Iwan Gunawan Markus Kostner Paul Kriss Rey Ancheta Joop Stoutjesdijk Zuzana Svetlosakova Sandra Walston Dainty Ignacio 135 Methodology Report: Calculating Multi-hazard City Risk International Advisory Group Experts Roundtable in February 2011 Included Participants: Adam Freed, City of New York Amanda Eichel, City of New York Anna Burzykowska, European Space Agency Robert Kehew, UN Habitat Cristina Rumbaitis del Rio, Rockefeller Foundation David Simpson, University of Louisville Arturo Samper, Global Adaptation Institute Robert Belk, Deltares /Delft Hydraulics Kees Bons, Deltares/Delft Hydraulics Nikhil Krishnan, McKinsey & Co Greg Hintz, McKinsey & Co Alexander Kaplan, Swiss Re Nikhil Da Victoria Lobo, Swiss Re Rob Lempert, RAND Shagun Mehrotra, Columbia University Marc Levy, Columbia University Shane Mitchell, Cisco Sven Harmeling, GermanWatch Helen Hodge, Maplecroft Additional Written Inputs Also Received From: Stephen Coulson, European Space Agency Yuki Matsuoka, UN-ISDR Richard Sanders, Willis Re Patrick McSharry, University of Oxford/ Willis Research Network Consultants Chreod Ltd. Edward Leman (Team Leader), Toronto Zhang Rufei (Deputy Team Leader), Shanghai Cameron Proctor (GIS Specialist), Toronto Shen Weinan (Database Specialist), Shanghai Imran Hasan (Database Specialist), Toronto David Etkin (DRM Specialist), York University, Toronto Kaz Higuchi (Climate Modeling Specialist), York University, Toronto Frank Zhang (Climate researcher), York University Annemarie Schneider (Geospatial Specialist), University of Wisconsin, Madison Saengabha Srisopaporn (Social and Institutional Specialist), Bangkok Jakkaphun Nanuam (Environment and Infrastructure Specialist), Burapha University Leo Barua (GIS Specialist), Manila University of the Philippines Team: Mahar Lagmay (Geophysical Specialist) Candido A. Cabrido Jr. (DRM Specialist) Jun Castro (Infrastructure and GIS Specialist) Mark Morales (Land Use and Building Specialist) Evelyn Lorenzo (Population and Local Government Specialist) Jennifer Barreto (Hydro-geologist) Sogreah Ltd. (Artelia) Gary Moys (Hydrologist), Shanghai 136 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project REFERENCES Adams, J. (1995). Risk. London: University College Press. Alley, W.M. (1984). The Palmer Drought Severity Index: Limitations and Assumptions. Journal of Climate and Applied Meteorology. Vol. 23, 1100-1109. Air Worldwide (2010). Anatomy of a Damage Function: Dispelling the Myths, March 16, 2010. http://www.air- worldwide.com/PublicationsItem.aspx?id=19052 Annamalai, H., K. Hamilton, K. R. Sperber, (2007). The South Asian Summer Monsoon and Its Relationship with ENSO in the IPCC AR4 Simulations. J. Climate, 20, 1071–1092. Arnell, N.W. (n.d.). Climate Change and Drought. Options Mediterraneennes, Series A, No. 80, 13-21. Baettig, M. B., Wild, M. & Imboden, D. M. (2007). A climate change index: Where climate change may be most prominent in the 21st century. Geophysical Research Letters, 34, L01705, doi:10.1029/2006GL028159. Bagalihog, E. and Siapno, J. (n.d.). Drought Management in the Philippines. Downloaded from www.adb.org/water/NARBO/2005/.../presentation-session3-PHI.pdf June 1, 2011. Birkman, J. & Wisner, B. (2006). Measuring the un-measurable: The challenge of vulnerability. Bonn: United Nations University, Institute for Environment and Human Security. (SOURCE Publication Series of UNH-EHS No. 5/2006) Birkman, J. (2006). Indicators and criteria for measuring vulnerability: Theoretical bases and requirements. In J. Birkman (Ed.), Measuring vulnerability to natural hazards: Towards disaster resilient societies (pp. 55–78). New York: United Nations University Press. Blaikie, P., Cannon, T., Davis, I., & Wisner, B. (1994). At risk: Natural hazards, peoples’ vulnerability, and disasters. London: Routledge. Bollasina, M. and Nigam, S. (2009). Indian Ocean SST, evaporation, and precipitation during the South Asian summer monsoon in IPCC-AR4 coupled simulations. Climate Dynamics Vol. 33 (7-8), 1017-1032. Brecht, H., & World Bank Development Research Group (2007). A global urban risk index: Analyzing mortality and economic threats from natural hazards. Washington, DC: World Bank. Burke EJ, Brown SJ, Christidis N, (2006). Modelling the recent evolution of global drought and projections for the twenty-first century with the Hadley centre climate model, Journal of Hydrometeorology, Vol: 7, Issue: 5, 1113- 1125 Burke EJ, Brown SJ, (2008). Evaluating uncertainties in the projection of future drought, Journal of Hydrometeorology, Vol: 9, Issue: 2, 292-299 Burton, I., Kates, R. & White, G. (1993). Environment as hazard (2nd edition). New York: The Guilford Press. Cardona, O. D. (2005). Indicators of disaster risk and risk management. Summary report. IDB/IDEA Program on Indicators for Disaster Risk Management. Washington, DC: Inter-American Development Bank, Sustainable Development Department Environment Division. Carreno, M. L., Cardona, O. D. & Barbat, A. H. (2007). A disaster risk management performance index. Natural Hazards 41, 1–20. Carreno, M. L., Cardona, O. D. & Barbat, A. H. (2008). Application and robustness of the holistic approach for the 137 Methodology Report: Calculating Multi-hazard City Risk seismic risk evaluation of megacities. The 14th World Conference on Earthquake Engineering, Beijing, October 12–17, 2008. Chavas, D.R. and Emanuel, D.A. (2010). A QUikSCAT climatology of tropical cyclone size. Geophysical Research Letters, Vol. 37 Chreod Ltd. (2010). Global Dataset on Metropolitan Regions. Cismaru, M. (2006). Protection motivation theory – An additive or a multiplicative model? Advances in Consumer Research, 33, 271–272. Collins, M., An, S-I, Cai, W., Ganachaud, A., Guilyardi, E., Jin, F.-F., Jochum, M., Lengaigne, M., Power, S., Timmermann, A., Vecchi, G., and Wittenberg, A. (2010). The impact of global warming on the tropical Pacific Ocean and El Niño. Nature Geoscience, Vol. 3, 391-397. Columbo, A.F., Etkin, D. and Karney, B.W. (1999). Climate variability and the frequency of extreme temperature events for nine sites across Canada: Implications for power usage. Journal of Climate, vol. 12(8), 2490-2502. Conde, C.H. ( 2010). Power Cuts Will Hit Manila, New York Times, March 4, 2010 Cutter, S. L., Boruff, B. J. & Shirley, W. L. (2003). Social vulnerability to environmental hazards. Social Science Quarterly, 84 (2), 242–261. Dai, A., K. E. Trenberth, and T. Qian, 2004: A global data set of Palmer Drought Severity Index for 1870-2002: Relationship with soil moisture and effects of surface warming. J. Hydrometeorology, 5, 1117-1130. Davidson, R. (1997). An urban earthquake disaster risk index. Report no. 21. Stanford: The John A. Blume Earthquake Engineering Center, University of Stanford. Dickson, Eric, Tiwari, A., Baker & J., Hoornweg, D. (2010). Understanding urban risk: An approach for assessing disaster and climate risk in cities. Washington, DC: World Bank. Dilley, M., Chen, R. S., Deichmann, U., Lerner-Lam, A. L. & Arnold, M. (2005). Natural disaster hotspots: A global risk analysis: Synthesis report. Washington, DC: World Bank. Etkin, D. (2009). Patterns of risk: Spatial planning as a strategy for the mitigation of risk from natural hazards. In U. Fra Paleo (Ed.), Building safer communities. Risk governance, spatial planning, and responses to natural hazards. Amsterdam: IOS Press. Flanagan, B. E., Gregory, E. W., Hallisey, E. J., Heitgerd, J. L. & Lewis, B. (2011). A social vulnerability index for disaster management. Journal of Homeland Security and Emergency Management, 8 (1). Fleming, K., Johnston, P., Zwartz, D., Yokoyama, Y., Lambeckm K. and Chappell, J. (1998). "Refining the eustatic sea-level curve since the Last Glacial Maximum using far- and intermediate-field sites". Earth and Planetary Science Letters 163 (1-4): 327-342. Franck, T., (2009). Coastal Adaptation and Economic Tipping Points. Management of Environmental Quality, Vol. 20(4), 434-450 Frerks, G. and Bender, S. (2004). Conclusion: Vulnerability analysis as a means of strengthening policy formulation and policy practice. In G. Bankoff, G. Frerks, & D. Hilhorst (Eds.), Mapping vulnerability: disasters, development and people (pp. 194–205). London: Earthscan. Frich, P., Alexander, L.V., Della-Marta, P., Gleason, B., Haylock, M., Tank, A.M.G.K. and Peterson, T. (2002). Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate 138 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Research 19, 193–212. Füssel, H-M. (2009). An updated assessment of the risks from climate change based on research published since the IPCC Fourth Assessment Report. Climatic Change, Vol. 97, 469–482 GC Briefing (2011). Catastrophe Model Change: Issues and Challenges. Guy Carpenter, March 2011. http://www.guycarp.com/portal/extranet/insights/briefingsPDF/2011/Catastrophe%20Model%20Change%20Issues %20and%20Challenges.pdf;jsessionid=NS3Kdh69zjY3zH7kp2kccpfGz5c4PTDyLQJdyGvQQ8Q2RNJnCMTG!- 783285485?vid=1 Grossi, P and Kunreuther, H. (2005), Catastrophe Modeling: A New Approach to Managing Risk. Springer. Guha-Sapir, B., & Below, R. (2005) The quality and accuracy of disaster data: A comparative analyses of three global data sets. Brussels: The ProVention Consortium, The Disaster Management Facility, World Bank. Gumbel, E.J., 1958: Statistics of Extremes. Columbia University Press, New York. Hahn, M. B., Riederer, A. M. & Foster, S. (2009). The livelihood vulnerability index: A pragmatic approach to assessing risks from climate variability and change – A case study in Mozambique. Global Environmental Change, 19 (1), 74–88. Hayes, M.J. (n.d.). What is Drought? http://www.drought.unl.edu/whatis/indices.htm accessed June 2, 2011. Palmer, W. C, 1965: Meteorological Drought. Res. Paper No.45, 58pp., Dept. of Commerce, Washington, D.C. Heijmans, A. (2004). From vulnerability to empowerment. In G. Bankoof, G. Frerks & D. Hilhorst (Eds.), Mapping vulnerability: Disasters, development and people (pp. 115–127). London: Earthscan. Hewitt, K. (1997). Regions of risk – A geographical introduction to disasters. London: Longman. Hovden, J. (2004). Risk and uncertainty management strategies. Paper presented at the 6th International CRN Expert Workshop, Stockholm, April 22–24, 2004. Huang, Y. & Bocchi, A. (Eds.) (2009). Reshaping economic geography in East Asia. Washington, DC: World Bank. Jain, V. (2010). Anatomy of a Damage Function: Dispelling the Myths. March 16, 2010. AIR Worldwide. http://www.air-worldwide.com/PublicationsItem.aspx?id=19052. Jevrejeva, S., Moore, J. C. and Grinsted, A. (2010). How will sea level respond to changes in natural and anthropogenic forcings by 2100? GEOPHYSICAL RESEARCH LETTERS, VOL. 37, L07703, 5 PP., 2010, doi:10.1029/2010GL042947 Kaly, U. L., Pratt, C. & Mitchell, J. (2005). Building resilience in SIDS: The environmental vulnerability index. Final report. SOPAC and UNEP. Retrieved from http://www.vulnerabilityindex.net/EVI_Library.htm) Knutson, T. R., McBride, J. L., Chan, J., Emanuel, K., Holland, G., Landsea, C., Held, I, Kossin, J. P., Srivastava, A. K. & Sugi, M. (2010). Tropical cyclones and climate change. Nature Geoscience, 3, 157–163. Knutson, T. R., Sirutis, J. J., Garner, S. T., Vecchi, G. A. & Held, I. M. (2008). Simulated reduction in Atlantic hurricane frequency under twenty-first-century warming conditions. Nature Geoscience. Retrieved from www.nature.com/naturegeoscience Kripalani, R.H., Oh, J.H., Kulkarni, A., Sabade, S.S. and Chaudhari, H.S. (2007). South Asian summer monsoon precipitation variability: Coupled climate model simulations and projections under IPCC AR4, Theoretical and Applied Climatology, Volume 90, Numbers 3-4, 133-159 139 Methodology Report: Calculating Multi-hazard City Risk Leman, E. (2009). Adapting to climate change in Shanghai and the Yangtze Delta Region. Paper presented at the World Bank Urban Research Symposium V, Marseille, June 28, 2009. Menzefricke, U., (1979). Choosing between an additive and a multiplicative model of experimental effects. Decision Sciences, 10, 196–204. Maine, E. M., Shapiro, D. M. & Vining, A. R. (2008). The role of clustering in the growth of new technology-based firms. Small Business Economics 34:127–146. Mitchell, J. K. (1999). Natural disasters in the context of mega-cities. In J. K. Mitchell (Ed.), Crucibles of hazard: Mega-cities and disasters in transition. New York: United Nations University Press. Murakami,H., Wang, B. and Kitoh, A. (2011). Future Change of Western North Pacific Typhoons: Projections by a 20-km-Mesh Global Atmospheric Model, Journal of Climate. Vol. 24 (4), 1154-170. Nicholls, R. J., Hanson, S., Herweiger, C., Patmore, N. Hallegatte, S., Corfee-Morlot, J., Chateau, J. & Muir-Woord, R. (2008). Ranking port cities with with exposure and vulnerability to climate extremes: Exposure estimates. OECD Environment Working Papers, No. 1. Paris: OECD. NIST/SEMATECH e-Handbook of Statistical Methods, http://www.itl.nist.gov/div898/handbook/, 20 June 2011. Peduzzi, P., Dao, H., Herold, C. & Mouton, F. (2009). Assessing global exposure and vulnerability towards natural hazards: the Disaster Risk Index. Natural Hazards Earth Systems Science, 9, 1149–1159. Pelling M. (2004). Visions of risk: A review of international indicators of disaster risk and its management. A Report for the ISDR Inter-Agency Task force on Disaster Reduction, Working Group 3: Risk, Vulnerability and Disaster Impact Assessment. Perry, R. W. & Qarantelli, E. L. (2005). What is a disaster? New answers to old questions. Bloomington, IN: Xlibris. Rahmstorf, S. (2007). "Sea-Level Rise: A Semi-Empirical Approach to Projecting Future." Science 315 368–370. Sheridan, S. (2004). Developing heat-warning systems for cities worldwide. In Janelle, D.G.,Warf, B. and Hansen, K., editors, WorldMinds: geographical perspectives on 100 problems. Dordrecht: Kluwer Academic Publishers, 487–92. Rygel, L., O’ Sullivan, D. & Yarnal, B. (2006). A method of constructing a social vulnerability index: An application to hurricane storm surges in a developed country. Mitigation and Adaptation Strategies for Global Change, 11, 741–764. Ravallion, M. (2010). Mashup indices of development. Policy Research Working Paper 5432. Washington, DC: World Bank. Shah, F. (2010). Can Asian cities be both green and resilient? Global Asia, 4 (4). Simpson, A.G. (2011). P/C Industry Depends Too Much on Catastrophe Models, Says Pioneer Clark. Insurance Journal, April 14, 2011. http://www.insurancejournal.com/news/national/2011/04/14/194464.htm Simpson, D. M. (2006). Indicator issues and proposed framework for a disaster preparedness index (DPI). Draft Report to the Fritz Institute Disaster Preparedness Assessment Project by the Center for Hazards Research and Policy Development, University of Louisville. Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.) (2007). Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate 140 Methodology Report: Calculating Multi-hazard City Risk Resilient Cities: Multi-hazard City Risk Index Project Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Souch, C. and Grimmond, C.S.B. (2004). Applied climatology: ‘heat waves’. Progress in Physical Geography, vol. 28(4), 599-606. Slovic, P. (2000). The perception of risk. London: EarthScan Publications. Tol, R.S. (2002). Estimates of the damage cost of climate change. Part I: benchmark estimates. Environmental and Resource Economics 21, 42–73. United Nations - Department of Economic and Social Affairs (2007). World Urbanization Prospects: The 2007 Revision (CD ROM) Vongvisessomjai, S. (2009). Tropical cyclone disasters in the Gulf of Thailand. Songklanakarin Journal of Science and Technology, 31 (2), 213–227. Wang, J.L. and Gaffen, D.J. (2001). Trends in extremes of surface humidity, temperature, and summertime heat stress in China. Advances in Atmospheric Sciences 18, 742–51. Wood, R. M., Drayton, M., Berger, A., Burgess, P. & Wright, T. (2005). Catastrophe loss modelling of storm-surge flood risk in eastern England. Philosophical Transactions of The Royal Society, 363 (1831), 1407–1422. World Bank (2009). Reshaping Econoic Geography: World Development Report 2009. Washington, DC, World Bank. World Bank (2010). Economics of adaptation to climate change: Synthesis report. Washington, DC, World Bank. World Wide Fund for Nature (WWF) (2009). Mega-stress for mega-cities A climate vulnerability ranking of major coastal cities in Asia. World Wide Fund. Yusuf, A.A. and Francisco, H. (2009). Climate Change Vulnerability Mapping for Southeast Asia, Economy and Environmental Program for Southeast Asia, Singapore, http://www.eepsea.org 141