Disclaimer This work is a product of the staff of the World Bank. The findings, interpretation, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement The material in this publication is copyrighted. Copying and/or transmitting portions of this work without permission may be a violation of applicable laws. For permission to photocopy or reprint any part of this work, please send a request with the complete information to the World Bank Group Romania (Vasile Lascăr Street, Nr. 31, Sector 2, Bucharest, Romania). Acknowledgements This report is the result of the work performed by a team of World Bank staff and experts led by Chris-Philip Fischer (Task Team Leader, Senior Water Resources Management Specialist), Elena Daniela Ghiță (co-Task Team Leader, Water Resources Management Specialist) and Natalia Limones Rodriguez (professor at University of Seville), including Anca (Borș) Mitrache, Cosmin Feodorov, Crystal Fenwick, Gabriel Ioniță, Ioan Bica, Iozefina Lipan, Nicolas Gerber, and Teodor Popa. The team also benefited from the solid support provided by Raimund Mair (Senior Water Resources Management Specialist) and Nathan Engle (Senior Water Resources Management Specialist), as well as from the feedback of Alanna Simpson (Lead Disaster Risk Management Specialist), Gabriel Seth Sidman (Climate Change Specialist), Marc Sadler (Program Leader), Mariano Gonzalez Serrano (Senior Energy Specialist). The authors would like to give special thanks to Winston Yu (Practice Manager, Water Global Practice in Europe and Central Asia, The World Bank) and Anna Akhalkatsi (Country Manager for Romania and Hungary, The World Bank), for the overall coordination, as well as for the guidance and valuable advice. In the same manner, the team would like to express its gratitude towards the counterparts in the Ministry of Environment, Waters and Forests (MEWF), National Administration “Romanian Waters” (ANAR), National Institute of Hydrology and Water Management (INHGA), National Regulatory Authority for Public Utilities Community Services (ANRSC), Romanian Water Association (ARA), General Inspectorate for Emergency Situations (IGSU), National Directorate of Forests – Romsilva, for their support and excellent collaboration (including data sharing) for the development of this document. Furthermore, the team would like to thank Ilie Vlaicu, president of ARA, for making possible the assessment of the benchmarking data for water supply and sanitation sector. The team would like to thank the experts involved in the EDORA Project of the European Commission’s Joint Research Centre (JRC) for providing access to the EDORA methodology for data-driven risk assessment, and to Dor Fridman, Peter Burek, Emilio Politti, Reetik Sahu, Barbara Willaarts and Taher Kahil from International Institute for Applied Systems Analysis (IIASA) and Marthe Wens from the Vrije Universiteit Amsterdam for its application to Romania. The team benefitted as well from the support provided by the Danube Water Program (DWP). i Table of contents Abbreviations ......................................................................................................................... i Executive Summary ............................................................................................................... 1 1 Introduction ................................................................................................................ 4 1.1 Background and objective................................................................................................................. 4 2 Methodological Approach ........................................................................................... 8 2.1 Key definitions and concepts ............................................................................................................ 8 2.2 Overview of the EDORA methodology ............................................................................................ 10 2.3 Zoom-in into the water supply and sanitation sector ..................................................................... 14 3 Drought Hazard in Romania ....................................................................................... 16 3.1 An overview of historical drought conditions ................................................................................. 16 3.2 Climate projections ......................................................................................................................... 19 4 Socioeconomic and environmental impacts and current and future drought risk in Romania .............................................................................................................................. 21 4.1 Agriculture, livestock and fisheries ................................................................................................. 21 4.2 Energy production ........................................................................................................................... 28 4.3 Inland fluvial transportation ........................................................................................................... 33 4.4 Industrial productivity ..................................................................................................................... 35 4.5 Natural ecosystems ......................................................................................................................... 37 4.6 Water supply ................................................................................................................................... 41 5 Opportunities to strengthen drought risk management ............................................. 45 5.1 Recommendation 1: Conduct a comprehensive DRRA and introduce proactive drought risk management practices................................................................................................................................ 47 5.2 Recommendation 2: Develop drought risk management plans at river basin scale....................... 47 5.3 Recommendation 3: Introduce drought risk mitigation and management in the water supply and sanitation sector ......................................................................................................................................... 49 5.4 Recommendation 4: Strengthen data sharing, collection and dissemination activities related to drought........................................................................................................................................................ 49 5.5 Recommendation 5: Earmark additional investments for drought risk management ................... 50 References .......................................................................................................................... 51 Annex 1: List of local and regional operating companies (LOCs/ROCs) interviewed ............... 59 Annex 2: Questionnaire for the qualitative interview with ROCs and LOCs ............................ 60 Annex 3: EDORA methodology ............................................................................................. 62 List of boxes Box 1: Standardized drought indices used to calculate drought hazard in Romania .................................. 11 Box 2: Summary of interviews conducted with ROCs and LOCs ................................................................. 14 ii List of figures Figure 1: Estimated drought risk across Romania (2011-12) ........................................................................ 4 Figure 2: Capture from Combined Drought Indicator (CDI) map of Europe at the beginning of August, 2022 .............................................................................................................................................................. 6 Figure 3: Sequence of commonly-accepted drought categories .................................................................. 8 Figure 4: Graphical definition of risk ............................................................................................................. 9 Figure 5: Location of the operating companies participating in the long interview - drought impacts on the water sector .......................................................................................................................................... 15 Figure 6: Medium-term drought index (SPI12) for the period 1951-2022 by RBA ..................................... 16 Figure 7: Long-term drought index (SPI48) for the period 1951-2022 by RBA ........................................... 16 Figure 8: Minimum values for medium- (SPI12) and long-term (SPI48) drought indices for the period 2012-2022 by RBA ....................................................................................................................................... 17 Figure 9: Medium-term (SPEI12) drought index for the period 1951-2022 by RBA ................................... 17 Figure 10: Long-term (SPEI48) drought index for the period 1951-2022 by RBA ....................................... 18 Figure 11: Minimum values for medium- (SPEI12) and long-term (SPEI48) drought indices for the period 2012-22 by RBA ........................................................................................................................................... 18 Figure 12: Standardized yearly river discharge in Romania (orange) compared with short-term drought index (SPEI12) (blue) (1990 to 2022) .......................................................................................................... 18 Figure 13: Standardized yearly river discharge (orange) compared with short-term drought conditions (SPEI12) (blue) for 2010-2021 for each RBA ............................................................................................... 19 Figure 14: Historical and projected mean temperatures across Romania under different climate scenarios (2000-2100)................................................................................................................................................. 20 Figure 15: Historical and projected changes in mean temperatures across Romania (1951-2100) by month under different climate scenarios (2000-2100) ............................................................................... 20 Figure 16: Future drought indices in Romania under climate change (RCP 2.6) with a three-month accumulation period ................................................................................................................................... 20 Figure 17: Future drought indices in Romania under climate change (RCP 8.5) with a three-month accumulation period ................................................................................................................................... 20 Figure 18: Annual agricultural output (millions of Euros) by development region (primary axis) compared with extreme drought (blue dashed line on secondary axis) over time ..................................................... 22 Figure 19: Number of locations within RBAs that experienced consistent restrictions in monthly water allocations for irrigation .............................................................................................................................. 22 Figure 20: Maize production (tons) compared with county average (April to September 2022) ............... 23 Figure 21: Maize yield (kg/ha) compared with county average (April to September 2022) ....................... 23 Figure 22: Average annual livestock production (million Euros) during non-drought (blue) versus mild drought conditions (brown) by region (2005 - 2020).................................................................................. 23 Figure 23: Average annual veterinary costs (million euros) during non-drought (blue) versus mild drought conditions (brown) conditions by region (2005 - 2020).............................................................................. 24 Figure 24: Distribution of farm fish facilities surface per development region .......................................... 24 Figure 25: Distribution of farm fish facilities number per development region ......................................... 24 Figure 26: Number of locations within RBAs that experienced consistent restrictions on water allocated to fisheries................................................................................................................................................... 25 Figure 27: Average annual loss for major crop groups associated with current drought risk by RBA ........ 26 iii Figure 28: Average annual loss of maize crops associated with current drought risk by RBA .................... 27 Figure 29: Predicted change in average annual loss of major crop groups associated with future drought risk (relative to 1990-2019 baseline) by RBA .............................................................................................. 27 Figure 30: Predicted change in average annual loss of maize crops associated with future drought risk (relative to 1990-2019 baseline) by RBA and climate change scenario ...................................................... 28 Figure 31: Distribution of electricity production in Romania in 2023, by source ....................................... 29 Figure 32: Monthly SPEI6 series and standardized total hydropower production in Romania, 2000-202229 Figure 33: Total hydropower production in Romania during different drought conditions (boxplots), compared with average monthly production of the entire series (black line) ........................................... 30 Figure 34: Current drought risk (average annual reduction in water use) for hydroelectricity generation by RBA 31 Figure 35: Drought loss exceedance curves for the energy sector in four selected river basin administrations in Romania ........................................................................................................................ 31 Figure 36: Electricity generation at the Iron Gate Hydropower Plant from 2000-2022.............................. 32 Figure 37: Future drought risk (changes in AAL in water use) for hydroelectricity production by RBA (relative to baseline) ................................................................................................................................... 32 Figure 38: Fluvial transport (standardized tons by quarter) (green lines) compared with drought conditions (bars) in Romania 2005 - 2022 .................................................................................................. 33 Figure 39: Total transport in thousands of tons per trimester by vessel (green boxplot) compared with mild drought conditions (brown boxplot) in Romania during 2005 - 2022 ................................................ 34 Figure 40: Tonnage (million tons) transported via inland fluvial systems ................................................... 34 Figure 41: Future drought risk on inland fluvial transportation (AAL) under short- and long-term scenarios ..................................................................................................................................................... 35 Figure 42: Number of locations that suffered consistent, monthly industrial water restrictions by RBA during the 2022 drought ............................................................................................................................. 36 Figure 43: Standardized monthly industrial production and compared with drought conditions in Romania, 2017-2022 ................................................................................................................................... 36 Figure 44: Standardized burned area per month and drought conditions across Romania, 2000-2023 .... 37 Figure 45: Number of wildfires extinguished in Romania (2021/22) .......................................................... 38 Figure 46: Number of fires extinguished in 2021 and 2022 compared with March 2022 drought conditions across Romania ......................................................................................................................... 38 Figure 47: Change in NPP of forest ecosystems .......................................................................................... 39 Figure 48: Change in NPP of wetland ecosystems ...................................................................................... 39 Figure 49: Future drought risk on forest ecosystems by RBA, compared to baseline ................................ 40 Figure 50: Future drought risks on wetland ecosystems by RBA, compared to baseline ........................... 40 Figure 51: Number of locations that suffered consistent restrictions on public water supplies by RBA and month .......................................................................................................................................................... 41 Figure 52: Bulk water sales for drinking water compared with drought risk (secondary axis) by RBA 2015-2022 42 Figure 53: Drought risk for bulk water deliveries to water utilities in current climate conditions by RBA . 44 Figure 54: Future drought risk for bulk water deliveries to water utilities by RBA, compared to baseline 44 List of tables Table 1: Data used to assess drought impacts by sector ............................................................................ 12 Table 2: Outcome indicator of drought risk by sector................................................................................. 13 Table 3: Current and future average annual crop losses associated with drought risk in Romania ........... 25 iv Abbreviations AAL Annual Average Loss AEZ Agro-Ecological Zones ANAR National Administration ‘Romanian Waters’ ANM National Meteorological Administration ANRSC National Regulatory Authority for Public Utilities Community Services ARA Romanian Water Association BACC Balanced Accuracy CCDR Country Climate and Development Report CMIP Coupled Model Inter-comparison Project CWatM Community Water Model DMP Dry Matter Production DRRA Drought Risk and Resilience Assessment EDORA European Drought Observatory for Resilience and Adaptation, a Project hosted by the Joint Research Centre EDO European Drought Observatory FRMP Flood Risk Management Plan GCM Global Climate Models GDP Gross Domestic Product GWP Global Water Partnership IDRMP Integrated Drought Risk Management Plan IGSU General Inspectorate for Emergency Situations INHGA National Institute of Hydrology and Water Management JRC Joint Research Centre LOC Local Operating Companies MEWF Ministry of Environment, Waters and Forests ML Machine Learning NGO Non-Governmental Organization NPP Net Primary Productivity NRW Non-Revenue Water NUTS Nomenclature of Territorial Units for Statistics RBA River Basin Administration RBMP River Basin Management Plans RCP Representative Concentration Pathways ROC Regional Operating Companies ROR Run-of-the river SCD Systematic Country Diagnostic i SPEI Standardized Precipitation-Evaporation Index SPI Standardized Precipitation Index SSFI Standardized Streamflow Index WEI+ Water Exploitation Index Plus WMO World Meteorological Organization ii Executive Summary 1. The number of drought events, as well as their associated losses, have been increasing in severity in Romania in the past 50 years, accelerated by climate change. Such events have a profound impact on population, as well as on the economic activities and environment. 2. In 2022, Romania faced the impacts of the most severe drought in the past 500 years that affected the entire Europe. 220 localities suffered water restrictions, low water levels in the Danube caused disruptions to hundreds of ships, while hydro-electric power production decreased 30 percent compared with previous years. Over 160,000 hectares were affected by a soil moisture deficit causing a 25-30 percent reduction in agricultural output, an increase in animal food prices and a reduction in livestock numbers. Wildfires increased at least seven-fold, damaging more than 10,000 hectares, while aquatic ecosystem have been impacted by the drying lakes. 3. As an European Union (EU) Member State, Romania deployed efforts to overcome the challenges and the weaknesses in addressing the drought problems, but more needs to be done for a proactive integrated drought risk management to increase the resilience of the population, the dependent water sectors and the environment. At international level, more emphasize is put on taking actions to reduce the climate change impacts (including the drought ones), and Romania joined these efforts. 4. This was the context that led to the necessity of a report on drought risk assessment for Romania, and the World Bank brought its global knowledge and experience in addressing this thematic, as a support to Ministry of Environment, Water and Forests (MEWF) and the National Administration “Romanian Waters” (ANAR) – key national institutions in the water sector in Romania. To be mentioned that other key institutions for the water sector are the Ministry of Development, Public Works and Administration and its subordinated institution - the National Regulatory Authority for Public Utilities Community Services (ANRSC), which regulates the water services sector. This support is in line with the WB efforts to strengthen the country’s capacities for water resources management and further invest in water infrastructure to adapt to the changing climate, as these were identified as weaknesses in the updated Systematic Country Diagnostic (SCD)1 and the recently developed Country Climate and Development Report (CCDR)2 for Romania. 5. The information on current and drought risk is crucial for Romania’s authorities in their efforts to achieve the objectives of the various strategies aiming for sustainable development (e.g., National Strategy for Water Management 2023 – 2030, the National Strategy regarding Adaptation to Climate Change for the period 2022 - 2030, with the perspective of 2050, the National Strategy of Disaster Risks Reduction 2024 – 2030, the National Strategy for the Sustainable Development of Romania 2030, National Circular Economy Strategy, Romania's Energy Strategy 2020 - 2030, with the perspective of 2050, Strategy for the Development of the Agri-Food sector in the medium and long term horizon 2020-2030, the National Strategy for the Rehabilitation and Expansion of the Irrigation Infrastructure from Romania, etc.). 6. This study provides a comprehensive analysis of current and future drought risks in Romania and points to the urgent need for enhanced integrated drought risk management. 7. An innovative, data-driven methodology, developed by the Joint Research Centre of the European Commission to create a pan-European Drought Atlas in the context of the European Drought Observatory for Resilience and Adaptation (EDORA) Project, was used for a more systematic assessment of current and future drought risks. The novelty of the approach is the use of machine learning algorithms to link drivers of drought risk with impacts, which contributes to increased accuracy of the results. For this work, the WB Water team benefitted from the support provided by the Water Danube Program and worked closely with the team that did a similar assessment for other non-EU countries in the Danube region (in Western Balkans and Eastern Europe), to complement EDORA project assessments. 1 World Bank, 2023a 2 World Bank, 2023b 1 8. This report is structured into five chapters. More details regarding the context of the drought thematic in Romania, the objectives of this report, the World Bank support to Government of Romania, etc. are provided in the Introduction. The second chapter is focused on the methodological approach, introducing the key concepts and definitions and presenting the EDORA methodology to calculate the drought risk. Drought hazard assessment has a dedicated chapter, highlighting the historical conditions and as well the future projections. The fourth chapter presents the results of the socioeconomic impacts assessment and the risk (under current and future climate conditions) for relevant economic sectors and for environment. The last part summarizes the findings and dives into the opportunities to strengthen the risk management. 9. The drought hazard assessment revealed the increasing severity of the drought, especially during the last decade, and the trend of intensified drier conditions for future. Summary of key drought hazards: • Romania has grown increasingly dryer since 1980, caused primarily by rising temperatures. • River flow in Romania has been consistently decreasing since 2011. • Under climate change, temperatures in Romania are expected to increase further, especially during the summer months. • Rising temperatures and more frequent and prolonged heat waves are predicted to intensify persistent drought conditions across the country. 10. The results of the drought socioeconomic impacts and the current and future risk assessment for relevant sectors indicate an increased trend. The risk is expressed as Annual Average Loss (AAL) due to drought, to show the magnitude of the potential impacts. 11. In general, the scores of the risk calculation models’ performance evaluation were high. 12. As a general remark, data on drought impacts are not systematically collected in Romania while data on critical hydrological metrics, such as groundwater, are also lacking or unavailable. Updated and more granular data would provide a more detailed risk assessment and additional insights into potential impacts that when combined with a comprehensive assessment of existing drought management measures and practices would support the development of tailored drought risk management strategies. 13. The results of this analysis demonstrate the urgent need for proactive drought risk management and to identify opportunities for strengthening resilience across Romania’s socioeconomic sectors and ecosystems in the face of future drought events. This drought risk assessment could be considered as the first step towards completing a full Drought Risk and Resilience Assessment (DRRA), which can support the country’s transition to a more proactive drought risk management response and prepare for a water-, food, and energy-secure future. 14. Although the primary purpose of this study was to identify and analyze current and future drought risk in Romania, and an assessment of existing drought risk management practices was not formerly undertaken, several opportunities to strengthen Romania’s capacity for drought risk management have been identified: • Conduct a comprehensive DRRA and introduce proactive drought risk management practices; • Develop drought risk management plans at river basin scale; • Introduce drought risk mitigation and management in the water supply and sanitation sector; • Strengthen data collection and dissemination activities related to drought; and • Earmark additional investments for drought risk management. 15. In face of increasing drought risks, Romania should act now towards an integrated drought risk management to increase the resilience of the population, economic sectors and environment, and to better use and secure the water resources. 2 Summary of the drought impacts and risks: Agriculture Impact: in 2022 over 160,000 hectares were affected by a soil moisture deficit causing a 25-30 percent reduction in agricultural output, an increase in animal food prices and a reduction in livestock numbers. Risk: ▪ Present conditions: the AAL in crop yields ranges between 3.9% for fodder, 4.5% for fruit and vegetables, 4.7% for potatoes, to 7.4% for cereals and 7.7% for oil crops. ▪ Future conditions: the AAL in crop yields due to drought ranges between 8.6 – 9.4% for fodder, 11.0 – 11.4% for fruit and vegetables, 13.5 - 14% for potatoes, to 12.8 – 13.7% for cereals and 17.4 – 18.1% for oil crops. Energy production Impact: in 2022 only 85% of total yearly energy production was reached. Risk: ▪ Present conditions: Water-use for hydro-electricity production shows an AAL of about 6.3% at the national level. ▪ Future conditions: Changes in AAL in the future under different RCPs mostly worsen drought impacts on energy production, where all currently high impact regions expect significant increases, reaching AALs due to drought of 16.2% -19.9%. Fluvial transport Impact: in 2022 a 4.6% reduction in transport was observed compared to normal yearly values, with only 81% of the average transportation expected in Q3. Risk: ▪ Present conditions: the reduction in goods loadings/unloading in Romania associated with drought is estimated almost at 1.4%. ▪ Future conditions: Changes in AAL due to drought in the future for most RCPs are expected to worsen the drought impacts on inland water transport, in RCP 7.0 and 8.5 are expected to reach 3.05% and 3.25%, respectively by the end of the century. Industrial productivity Impact: Although the industrial disruption cannot be linked entirely with the drought, it should be noted that in 2022 the industrial productivity is systematically lower and correlate with drought conditions. Risk: The lack of detailed data didn’t allow the construction of a drought risk model for this sector, and the risk was not calculated. Natural ecosystems Impact: 2022 has been the worst year since recorded data on wildfire exists: not only their number increased, but as well the interval for extinguish them and the burned surface per fire doubled (more than 10,000 hectares were affected). Risk: ▪ Present conditions: The AAL due to drought for the NPP of forest ecosystem shows an AAL of about 2.9% at the national level, reaching the maximum value in Prut-Barlad RBA (6%). The AAL for the NPP of the aquatic ecosystems is about 2.8% at the national level and has increased values in Banat RBA, Jiu RBA and Crisuri RBA (> 3.5%). ▪ Future conditions: The AAL for NPP of the forest systems is expected to increase relative to the baseline only in 4 RBAs: Jiu RBA (4.1% -4.5%), Buzau-Ialomita RBA (5.1% - 6.7%), Dobrogea-Litoral RBA (6.7% - 8.1%) and Prut-Barlad RBA (13.8% - 15.1%). The AAL for NPP of the wetland systems is expected to further increase mainly in the southern and southeastern river basins, where local tributaries join the Danube, and in the delta area: Jiu RBA (11.3% - 12.7%), Arges-Vedea RBA (11% - 12.2%) and Dobrogea-Litoral RBA (6.2 - 7.4%). Water supply Impact: in 2022 reduced river flows disrupted public water supplies in 220 localities, restricting non- essential water use in many parts of the country. Risk: ▪ Present conditions: The AAL due to drought of the volumes for public water supply is estimated 3.4% loss at national level, reaching over 7% in the Arges-Vedea RBA. The impacts are highest in the most populated river basin administrations, and over 55% of Romania’s population live in river basins w ith AAL larger than 4%. ▪ Future conditions: The river basin administrations with the highest impact at current conditions, Arges-Vedea RBA and Buzau-Ialomita RBA also show the highest increases in the AAL in the future RCP scenarios, reaching average losses of 12.8% - 20.7%. 3 1 Introduction 1.1 Background and objective 16. Drought is hydro-meteorological phenomenon with complex manifestation, characterized by lack of precipitation which leads to water shortages. Unlike other natural phenomena, it impacts larger areas, lasts longer periods of time (from months to years) and occurs in all climate regimes. 17. Drought was a major cause to important changes throughout the history of humankind (e.g., the fall of the pharaohs, the Mayan empire destruction in Mesoamerica etc.)3 due to its severe impacts. It affects the population, various economic sectors and the environment. 18. The major droughts that affected Europe did not bypass Romania due to its geographical position in the South-Eastern Central part, in the Lower Danube basin. Thus, the terrifying consequences of the droughts of 19174 and 19465 are still fresh in the collective memory of Romanians. 19. The National Risk Assessment coordinated by the General Inspectorate for Emergency Situations (IGSU) concluded that, even if, droughts are not not directly causing deaths in Romania, the phenomenon is one of the most severe natural hazards with major socio-economic and environmental impacts. For example, according to this assessment, almost half (48 percent) the country’s agricultural land was vulnerable to drought (figure 1)6. Figure 1: Estimated drought risk across Romania (2011-12) Source: General Inspectorate for Emergency Situations, 2016. 20. Furthermore, data shows that the drought conditions have increased substantially across the country and are expected to continue. The country has experienced several prolonged droughts since 1950, and most parts of the country have experienced persistent drought conditions since 2010. Rising temperatures combined with decreasing precipitation and extreme heat waves are expected to increase the frequency, intensity and duration of drought conditions with the potential to cause substantial adverse socioeconomic impacts. Over the years, Romania proves to 3 Broom, D., 2019 4 Crângan, C., 2022 5 Stănilă, I., 2022 6 General Inspectorate for Emergency Situations, 2016 4 be especially vulnerable to natural hazards induced or exacerbated by climate change, such as droughts and floods7. 21. A 2022 European-wide drought proved to be one of the most severe droughts ever recorded in Romania. In 2022, the lack of precipitation that led to historically low soil moisture combined with higher-than-average temperatures and a series of heatwaves triggering a severe to extreme drought across Europe (figure 2), severely impacted key socioeconomic sectors including agriculture, water supply, energy, and river transport, as well as the environment8. In Romania, reduced river flows disrupted public water supplies in 220 localities, restricting non-essential water use in many parts of the country. Low water levels in the Danube hampered river navigation, causing disruptions to hundreds of ships, while hydro-electric power production decreased 30 percent compared with previous years. Meanwhile, over 160,000 hectares were affected by a soil moisture deficit causing a 25-30 percent reduction in agricultural output, an increase in animal food prices and a reduction in livestock numbers. Finally, wildfires increased at least seven-fold, damaging more than 10,000 hectares9. With less than 200 mm of rainfall in 10 months and temperatures in excess of 36ºC for extended periods, the eastern half of the country was hit the hardest10. 22. At the international level, the United Nations (UN) has a long history in addressing the water issues and calling for actions, the most recent agreements including the Water Action Agenda (2023), the 2030 Agenda for Sustainable Development, the 2015-2030 Sendai Framework for Disaster Risk Reduction, the 2015 Addis Ababa Action Agenda on Financing for Development, and the 2015 Paris Agreement within the UN Convention Framework on Climate Change are a step forward towards achieving the Sustainable Development Goal 611. At European Union (EU) level, since 2000, water became a priority for the policy makers, once the Water Framework Directive 2000/60/EC was adopted, and since then, more laws and actions are contributing to water resources security. Furthermore, drought risk management is a priority area in the European Green Deal and is reflected as such in the following European strategies (e.g., EU Strategy on Adaptation to Climate Change, 2020 Circular Economy Action Plan etc.)12. As a Member State of both international bodies, Romania has committed to tackle the water issues and to overcome the challenges. 23. Despite the progress made towards addressing the water related issues, more needs to be done across the board to help key sectors mitigate and manage drought risk. Given its heightened susceptibility to natural hazards, Romania has made significant efforts to strengthen its institutional framework for disaster response13, but more needs to be done to help key sectors mitigate and adapt to drought risk. Water-dependent sectors, such as water supply, energy production, and agriculture, are especially vulnerable, but drought risk also threatens other sectors, such as transport and forestry. Given the potential for wide-reaching, catastrophic impacts, the imperative to develop and implement a comprehensive and integrated drought risk management program and ensure a water-, food-, and energy-secure future is urgent. 7 World Bank, 2023b 8 Copernicus Climate Change Service, 2022 9 Dumitrescu, R., 2022 10 National Meteorological Administration (ANM): https://www.meteoromania.ro/clim/caracterizare- anuala/cc_2022.html 11 United Nations, 2024 12 European Commission, 2023 13 World Bank, 2023b 5 Figure 2: Capture from Combined Drought Indicator (CDI)14 map of Europe at the beginning of August, 2022 Level Interpretation Watch Precipitation deficit Warning Negative soil moisture anomaly, usually linked with precipitation deficit Negative anomaly of vegetation growth, usually linked with precipitation deficit and negative soil Alert moisture anomaly Recovery After drought episode, both meteorological conditions and vegetation growth return to normal After a drought episode, soil moisture conditions are above the drought threshold, but not enough to Temporary Soil Moisture recovery consider the episode closed. After a drought episode, vegetation conditiond are above the drought threshold, but not enough to Temporary fAPAR recovery consider the episode closed. Source: Joint Research Centre (JRC) - Global Drought Observatory (GDO) of the Copernicus Emergency Management Service (CEMS), 2022 24. This was the context that led to the necessity of a report on drought risk assessment, and as a long-standing partner for sustainable development in Romania, Water WB team provided the needed support to the Ministry of Environment, Water and Forests (MEWF) and the National Administration “Romanian Waters” (ANAR) – key national institutions in the water sector in Romania – to address this issue. 25. This report lays the foundations for future work on drought risk management in Romania, as it presents a comprehensive analysis of current and future drought risks in Romania. In addition, the assessment moves beyond traditionally studied sectors, such as agriculture, to analyze potential risks and impacts of drought on other key socioeconomic sectors, such as water supply, energy, fluvial transport and natural systems. 26. An innovative methodology to calculate drought risk was used, which was recently developed by the European Commission’s Joint Research Centre (EC-JRC) under the European Drought Observatory for Resilience and Adaptation (EDORA) Project to develop the new European Drought Risk Atlas15. This data-driven approach was supplemented with key stakeholder interviews with water service providers, facilitated by ANAR and the Romanian Water Association (ARA). In 14 CDI is a drought indicator that combines the Standardized Precipitation Index, the anomalies of soil moisture and the anomalies of the fraction of Absorbed Photosynthetically Active Radiation (fAPAR) to indicate areas affected by agricultural drought, and areas with the potential to be affected. (source: Cammalleri, C., et al. (2021)) 15 Rossi et al., 2023. 6 addition, the assessment of the 2022 drought impacts in Romania led to improvements of the European Drought Risk Atlas. 27. For this work, the WB Water team benefitted from the support provided by the Water Danube Program and worked closely with the team that did a similar assessment for other non-EU countries in the Danube region (in Western Balkans and Eastern Europe), to complement EDORA project assessments. 28. The report’s findings underpin the results of the recent Romania Climate Change Development Report (CCDR) and highlight the urgent need for Romania to develop a comprehensive drought risk management program. The potential impacts of drought in Romania are extensive, making drought risk management all the more critical in the current era of climate change and growing water insecurity. This report focuses on identifying and estimating current and future drought risk across multiple sectors bolstering findings from the Romania CCDR that identified the need to strengthen water resources management in Romania, including the management of drought and flood risks to build greater resilience to climate change. Though an important first step, a more comprehensive instrument, such as the Drought Risk and Resilience Assessment (DRRA), that brings together already-established tools and frameworks to support an evaluation of the risks and effects of drought in a given country or area and develop effective interventions to mitigate impacts and increase resilience, would help the Government of Romania further this process. 29. The content of the report is structured into five chapters and more technical information can be found in the Annexes. Chapter 1 briefly introduces the impetus for a Drought Risk, followed by important definitions and drought-related concepts and an overview of the methodological approach in Chapter 2. Chapter 3 outlines the hazard conditions for Romania. Chapter 4 delves into the socioeconomic case for improved drought risk management through an analysis of historical drought impacts on key sectors and natural systems observed during the 2022 drought, in parallel with presenting the key results detailing current and future drought risks in Romania. It is followed by a summary of potential next steps in Chapter 5. 7 2 Methodological Approach 30. This Chapter provides an overview of the main concepts related to drought: hazard, exposure, vulnerability and risk and the links between the concepts (in section 2.1) and the methodology that was used to calculate the current and future drought risk (in section 2.2). 2.1 Key definitions and concepts 31. The terms ‘drought’ and ‘drought risk management’ are widely used, but the lack of clarity concerning certain drought-related characteristics (e.g., period, severity) hinders the drought risk management. This is the reason why, this section is dedicated to clarifying some of the concepts used in the report and their links. 32. A Drought event is generally defined as a temporary reduction in average water availability, frequently caused by a shortage of rainfall combined with high temperatures16. There are five commonly-accepted—often sequential—categories of drought17: 1. Meteorological drought: when dry weather dominates an area; 2. Hydrological drought: when water supplies become low; 3. Agricultural drought: when crops become affected; 4. Socioeconomic drought: when the supply and demand of goods become affected; 5. Ecological drought: when ecosystems become affected. Although these phenomena can overlap, their onsets and ends are staggered in time, and the complete sequence generally only occurs under prolonged or intense rainfall deficits (figure 3). Figure 3: Sequence of commonly-accepted drought categories Note: All droughts originate from a deficiency of precipitation or meteorological drought but other types of drought and impacts cascade from this deficiency. Source: US NDMC website: https://drought.unl.edu/ 16 Directorate-General for Environment, 2024. 17 NIDIS, 2024. 8 33. Drought Hazard can be defined as the probability of a drought phenomenon to occur (characterized by its duration, severity, extension, return period), thus a potential phenomenon that can cause losses and damages to exposed elements. 34. Drought Exposure represent the set of elements (population, communities, environment etc.) which could be exposed to drought phenomenon. 35. Drought Vulnerability represent the set of characteristics of the exposed elements to droughts (population, communities and environment) that increase the susceptibility of impacts caused by a drought event. 36. Drought Risk is the combination of the probability of a drought event to occur and its potential negative consequences on the exposed elements, expressed as potential drought damages and losses. Thus, the drought risk depends on three key dimensions: 1) the hazard, including the magnitude, timing, and frequency of different possible drought events; 2) exposed assets, productivity, or ecosystems; and 3) their intrinsic vulnerability to drought events (Figure 4). Figure 4: Graphical definition of risk Source: WB Water team 37. Drought impacts represent the damages and losses caused by a drought event. The evaluation of the drought impacts is not a simple process, as some of the drought impacts appear long after the drought passed or are hard to be detected due to the complexity of the affected systems. 38. Drought risk management is the process of planning and implementing measures to mitigate and manage the drought risks, with the aim of reducing the negative impacts and increasing the resilience of the exposed elements to this phenomenon. Drought risk management relies on identifying and developing a thorough understanding of drought risks either independently or as part of the drought risk management process. 39. A comprehensive approach aims to prepare for, mitigate and respond to drought risks before they arise, while, an emergency response takes action only after the onset of a drought event. 1. Emergency response: This approach is based on crisis management whereby measures and actions are implemented only after a drought event has started and the effects are perceived. This approach may result in less efficient technical and economic solutions, given time constraints often prevent a thorough situational assessment and limit stakeholder participation. Crisis management focuses on limiting rather than preventing damages. 2. Integrated response: A comprehensive approach to drought risk management includes designing measures in advance, with related planning tools and stakeholder participation. This proactive approach is based on short- and long-term measures and includes monitoring and early-warning systems, risk and impact assessment i.e., identifying vulnerable populations, and tailored measures for risk mitigation, preparedness and response, reflected in the three pillars approach for drought management.18 18 World Bank, 2024a. 9 2.2 Overview of the EDORA methodology 40. As mentioned before, for the assessment of the drought risk in Romania, the European Drought Risk Assessment (EDORA) methodology was used. It introduces an innovative hybrid approach to calculate the risk, using a factor approach on one hand (it combines the hazard, exposure, and vulnerability of specific sectors or systems to estimate the drought risk), and an outcome approach on the other hand (impact information is used to determine the importance of socio- economic and environmental factors in the negative impacts registered). This hybrid approach allows for a holistic and a more accurate risk assessment, as it allows the integration of those impacts that appear long after the drought is passed, and the correct estimation of the impacts which are often amplified by other climatic and non-climatic shocks. However, the method's strong dependence on data availability should be flagged, as it can cause biases, gaps, and inaccuracies in the risk system and estimations if the calibration data or information is flawed. 41. Following the hybrid drought risk factors – outcomes methodology, data on drought hazard, exposure, vulnerability, and impacts was collected from various sources (from national authorities, global models etc.) to estimate quantitatively the drought risk. In some of the cases, in the absence of data sets from national sources, global databases were used. Once the data was collected, extensive efforts were deployed to harmonize the datasets, to process and to bring them to the format required by the tools used for calculating the drought risk. 42. A complex model was developed to assess the Romania’s current and future drought risk for several sectors (agriculture, energy production, inland fluvial transportation, natural ecosystems and water supply and sanitation). It implied Machine Learning algorithms to assess the link between the hydro-meteorological conditions and the various impacts, decision trees and decision forests to optimize the hazard-impact link for each sector and the vulnerability, qualitative and quantitative analysis to build detailed drought impact chains etc. 43. The risk was assed as a relative Average Annual Loss (AAL) due to drought in current and future conditions for different systems-at-risk. The AAL is a measure of the mean loss for a defined period of time (in years) with some years of no (or little) losses. The advantage of using the AAL is the indication of the magnitude of the drought impacts. If data is available, the AAL can be easily translated into Monetary Losses. Moreover, the reason for using this expression of impacts resides in: (i) the high variability of the monetary values for many sectors and its dependency on a variety of factors, beyond drought conditions (e.g., war in Ukraine etc.), which makes comparison from year to year difficult, (ii) the lack of unbiased databases with estimated damages in monetary values for each sector, (iii) some impacts, such as damages to forestry and wetlands, can be quantified empirically, but are difficult to quantify economically, (iv) other economic indicators mask more nuanced impacts that are better captured using non-economic indicator (for example, a change in a farmer’s annual income is meaningless without knowing the volume of crops produced and sold). 44. More information regarding the methodology and its application for Romania’s drought risk assessment can be found in Annex 3. Drought hazard assessment 45. Understanding drought hazard is critical to calculating drought risks and estimating the potential damages caused by droughts. Standardized drought indices are widely used to describe the drought hazard because they provide a consistent and quantitative way to reliably assess the intensity, duration, frequency, and spatial distribution of drought. 46. To understand the drought hazards in Romania, the WB team used hydrometeorological data to calculate drought indices for the period 1951-2022 (box 1): the Standardized Precipitation Index (SPI), the Standardized Precipitation-Evaporation Index (SPEI) and the Standardized Streamflow Index (SSFI) etc. To estimate future drought hazard, widely-accepted climate change models were used to compute the same drought indices projected through the year 2100. 10 47. Standardized drought hazard indices were then compared with different historical, sector- specific data series on exposure to assess the impact of droughts on different sectors and systems. 48. For a comprehensive description of drought indices and methodological approach, see Annex 3. Box 1: Standardized drought indices used to calculate drought hazard in Romania Three relevant standardized drought hazard indices were calculated for this assessment: • The Standardized Precipitation Index (SPI)19 is a widely-used index to characterize the meteorological drought. This report presents results from the SPI12 which measures deviations over a 12-month (short-term) period and the SPI48 which represents a 48-month (long-term) period. • The Standardized Precipitation-Evaporation Index (SPEI)20. High temperatures increase the rate of evapotranspiration which may exacerbate drought conditions even if precipitation levels remain the same. This effect is captured by the SPEI which computes a simple monthly water balance by comparing monthly precipitation and monthly potential evapotranspiration. The calculated values are then aggregated at different time scales (e.g., SPEI12 and SPEI48) following the same procedure as the SPI. • Standardized Streamflow Index (SSFI).21 Prolonged meteorological drought can lead to reduced streamflow in rivers, lower water levels in lakes and reservoirs, and lower groundwater tables, generating hydrological drought. To assess hydrological drought in Romania the Standardized Streamflow Index (SSFI) was calculated for the period 1990 to 2022. • The result is an expression of how much the given indicator (e.g., precipitation and evapotranspiration or river flow) deviates from its long-term average for a given period in a given location. Negative values of drought indices indicate drier conditions (more drought), while positive values suggest wetter conditions (less drought). Drought exposure assessment 49. This step involved mapping the physical areas and systems that are exposed to drought risks. Spatial datasets of assets and production related to the various sectors mentioned in the report, were used to estimate the drought exposure. Drought vulnerability assessment 50. For each analyzed sector, vulnerability indicators were identified and combined into vulnerability classes/ clusters to understand how susceptible different systems are to drought or what is the level of their capacity to cope with and adapt to droughts. The vulnerability classes have different relevance from one sector to another. 51. To address the challenges of data gaps and better characterize vulnerability in drought risk analysis, regions with similar hazard-impact responses are grouped together, using both general and sector-specific proxies for vulnerability. This procedure assumes that each group has a similar likelihood of experiencing negative direct impacts. Drought impacts evaluation 52. Impact time-series, for a period covering at least 5 years, for each analyzed sector, were developed based on the data collected by WB team from various data sources (from national authorities, global models etc.). 19 McKee et al., 1993. 20 Vicente-Serrano et al., 2010. 21 Modarres, 2007; Telesca et al., 2012 11 53. For this assessment, the data sets were selected based on their availability, reliability, and completeness e.g., for the requisite time period. Where possible, the most comprehensive and reliable indicators were used to estimate drought impacts for each sector; however, in cases where data were unavailable or incomplete, proxies were used instead. For example, in the case of agriculture, restrictions in water allocated to irrigation are a good proxy for irrigated area. 54. The sector-specific data series that were used to assess the impact of droughts on different sectors and systems are presented in Table 1. Table 1: Data used to assess drought impacts by sector Sector Data Area with crops (winter and spring) affected - in total, by season Crops affected by drought (%) Territorial distribution of area affected by drought, by season Size of drought effects (% of expected production) Agriculture Compensations to farmers - impact on budget Economic/financial impact of output drop in 2022 Historical data regarding crop yields (2010-2022) by region/county Area irrigated during 2022, by month and county Area irrigated during 2010-2022, annually, by county Drought effects on forest (new and consolidated) vegetation status in 2022, by region Observed effects on forests during 2010-2021 under recurrent droughts Forestry Forest fires (number and burned surface) in 2022, by region Forest fires (number and burned surface) in 2010-2021, by region (for comparison) Variation of storage in hydropower reservoirs, monthly, in 2022 Energy production monthly in 2022 Energy Energy production annually in 2010-2021 Drought impact on Nuclear Plant Number of days (period) with navigation restrictions on Danube River, by sector Freight / weight restrictions enforced on Danube River Transport Rail transport restrictions enforced on main rail lines (passengers and freight) - number of hours/day, number of days, lines Number of passengers affected by rail restrictions, by region Number of localities affected by supply restrictions by region 12 Sector Data Water Supply & Number of populations affected by supply restrictions, by region Sanitation Number of days/hours per day of supply restrictions, by region Financial Impact of restrictions on water operators (regional & local) Localities / population with own supply affected by groundwater depletion 55. As mentioned before, the link between drought hazard and socioeconomic impacts was studied to determine the strength of their relationship, based on comparison for a specific timeline and location or based on samples (baseline vs. periods of drought). These data compilation and analysis fed the data-driven drought risk assessment that aimed to complement the understanding of current and future drought impacts and vulnerabilities in Romania. More data on impacts beyond those presented in the table, for example industrial production figures, were also compiled and explored, but were not included in the machine learning modelling because their connection with drought was not evident enough. Current and future drought risk assessment 56. The drought risk assessment is a 5-steps process, which implies: (i) identification of relevant proxies for drought impacts for each sector and system at risk, (ii) development of system-specific conceptual models leading to the formation of the system-specific impact chains, (iii) quantitative analysis of the hazard, exposure and vulnerability, (iv) calibration of the models using the machine learning algorithms, (v) calculation of the current and future drought risk metrics. 57. As mentioned before, the drought risk was expressed as the change in the socioeconomic variable analyzed, characterized as relative AAL due to drought. The advantages of selecting this parameter to express the risk were explained before. For these reasons, this study identified five alternative outcome indicators (table 2) that best describe drought risk by sector in Romania. Table 2: Outcome indicator of drought risk by sector Sector Outcome indicator Agriculture Change in crop yield Energy production Change in volume of water used to produce hydroelectricity Inland river transport Change in tonnage per vessel Forests and wetlands Change in Net Primary Productivity (NPP) Water supply Change in volume of bulk water delivered to water utilities 58. An increase in AAL relative to the average indicates a negative or detrimental impact generally associated with drier conditions, whereas a decrease in AAL relative to the average indicates a positive or beneficial impact generally associated with wetter conditions. 59. For the Romania drought risk assessment, the EDORA project methodology was applied using more granular data, collected for an inventory of drought impacts to provide a more detailed assessment of sectoral drought risks at the level of River Basin Administration (RBA) under current and future climate conditions. As the model is highly dependent on detailed spatiotemporal data, improving the input data would further enhance the impact assessment, particularly for energy and inland water transportation, and terrestrial and aquatic ecosystems. Other demographic and socioeconomic data, including crop prices, irrigation coverage, electricity demand, water demand, and water abstraction costs, can support the expansion of the modelling exercise and the interpretation of results, providing a clearer context for the assessment of drought impacts. These additional granular data contributed to the improvements of the European Drought Risk Atlas. 13 2.3 Zoom-in into the water supply and sanitation sector 60. Water service providers in Romania are aware of increasing drought risks but lack the preparedness and capacity to manage drought. During the 2022 drought, the Romanian Water Association (ARA), representing Romania’s 44 Regional Operating Companies (ROCs), and ANRSC, registered increasing concerns and reports from Romania’s water service providers about the challenges they faced to maintain water supplies. In a quick on-line survey realized by the World Bank in summer 2023 with all ARA members, most ROCs expressed concern about increasing drought hazards and many confirmed they were affected in some way by the drought. 61. Increased operational and maintenance challenges resulting from droughts can increase costs for providers. In some cases, scarcity necessitates more intricate water extraction processes, raising resource requirements. For example, low flow conditions lead to higher concentrations of impurities requiring additional treatment and increasing operational costs. Increased pumping and treatment activities during droughts lead to higher energy consumption while the added strain on water infrastructure accelerates wear and tear on equipment and facilities. The overall result is an increase in the frequency of modified operations and the intensity of maintenance activities. Officials suggested ensuring the provision of safe drinking water is more labor-intensive during drought conditions, but that hiring additional staff is not feasible, meaning financial data may not capture these impacts. On the other hand, some costs may decrease with reduced bulk water flows and reserves (e.g., reduced energy costs and reduced water costs due to lower volumes of water), but this reduction leads to non-fulfillment of its responsibilities as water supplier. Although officials expressed direct knowledge of these exceptional circumstances, the financial data available from 2019 did not reflect these conditions. Other operational costs that could be attributed to droughts but were not mentioned are likely considered as part of normal operating costs. For example, monitoring water flows and levels are part of routine activities even though droughts may increase their frequency. 62. Despite a thorough assessment of benchmarking data from ROCs, no clear relationship between costs and revenues, and drought conditions was observed. Potential explanations for a lack of correlation include: 1. Insufficient longitudinal data: Benchmarking data covered the period 2013-2022, which could all be considered relatively dry years. A longer time series would therefore be needed to compare data with non-drought conditions. 2. Different temporal scales: The drought indices used express meteorological drought (i.e., precipitation and evapotranspiration), which affect surface and groundwater differently at different times. As each water service provider relies on different water sources, a case-by- case impact analysis would be needed to identify the impact of drought on costs and revenues. 3. Different spatial scales: Drought indices were expressed at RBA scale, which might not be sensitive to ROC data that are collected at a smaller i.e., location- or utility-specific, scale. 4. Competing impacts: The war in the Ukraine severely impacted energy costs in Europe and Romania making it difficult to disaggregate increased energy costs related to droughts. 63. To better quantify drought impacts for water supply and sanitation sector, longer time series than the existing ones (the benchmarking data are available from 2013) and eventually different metrics should be analyzed. Nonetheless, the lack of correlation between drought conditions and changes in costs and revenues, does not mean droughts do not impact the operation of ROCs and LoCs. Thus, WB team conducted interviews with ANAR, INHGA, RBAs, IGSU and selected ROCs and LOCs to zoom-in into the sector challenges related to drought impact assessment and drought risk management and the qualitative interviews provided different results (box 2). Additional details about the selected ROCs and LOCs and the content of the interviews can be found in the annexes (Annex 1 and Annex 2). Box 2: Summary of interviews conducted with ROCs and LOCs 14 Based on their responses to a quick online survey, eight ROCs and nine Local Operating Companies (LOCs) were selected to participate; responses from eight ROCs and six LOCs were included for analysis (figure 5). The questionnaire focused on the following aspects: • Drought impacts: Interviews delved into the direct (economic) impacts experienced by the companies during drought conditions, providing a nuanced understanding of the challenges faced. • Risk assessment: Interviews explored perceived risks associated with drought events, helping to identify vulnerabilities and gaps in current strategies and infrastructure. The capacity for risk self-assessment of the companies was analyzed. • Mitigation measures: Detailed discussions were held on the various measures implemented by companies to mitigate the effects of droughts. This included strategies to address immediate challenges and long-term resilience-building initiatives. • Resilience enhancement needs: Participants were asked about improvements needed to enhance their resilience to drought events. This encompassed both operational and strategic considerations. Figure 5: Location of the operating companies participating in the long interview - drought impacts on the water sector Source: WB water team elaboration 15 3 Drought Hazard in Romania 64. This chapter presents the results of the assessment of the drought indices for different periods of time in the past (in section 3.1) and the results of the climate projection (in section 3.2). 3.1 An overview of historical drought conditions 65. As Romania faced severe drought events in the past, an assessment of the most relevant drought indices was done to better understand the drought conditions. The results are presented below. 66. Since 1951, Romania has experienced several extended dry periods and two countrywide droughts (2012 and 2022). The medium- (SPI12) and long-term (SPI48) drought indices show that Romania experienced longer dry periods at least three times from 1951 to 2022 (Figures 6 and 7). Dry periods (below average precipitation) can be observed from 1956-65, 1982-1996 and again at the beginning of the 21st century for each index. These periods were followed by years with positive SPI values, indicating years of above average precipitation. Dry periods do not always affect the entire country; however, medium- (SPI12) and long-term (SPI48) variations in rainfall at the RBA level for the ten-year period (2012-22) indicate most of the country was affected by the 2012 and 2022 droughts (Figure 8). Figure 6: Medium-term drought index (SPI12) for the period 1951-2022 by RBA Source: WB water team’s analysis using the CRU dataset Figure 7: Long-term drought index (SPI48) for the period 1951-2022 by RBA Source: WB water team’s analysis using the CRU dataset 16 Figure 8: Minimum values for medium- (SPI12) and long-term (SPI48) drought indices for the period 2012-2022 by RBA Note: Interpretation of the SPI values: extremely dry (SPI < -2.0), very dry (-2.0 < SPI < -1.5), moderately dry (-1.5 < SPI < -1.0), normal (-1 < SPI < 1), moderately wet (1.0 < SPI < 1.5), very wet (1.5 < SPI < 2.0), extremely wet (2.0 < SPI)22. Source: WB water team’s analysis using the CRU dataset 67. High temperatures are leading to more severe and persistent drought conditions in Romania. The SPEI, more representative of the resulting water balance, indices indicate Romania has gotten gradually dryer since 1980, with only sporadic recoveries. In more recent years, the SPEI12 shows Romania experienced consistently dry conditions from 2010 onwards, with only 2014-15 and 2021 exhibiting “normal” water balances (Figure 9). This is even more pronounced when considering the SPEI48, which shows long-term drought conditions that progressively worsen from 2011 onwards (Figure 10). These results indicate that even if precipitation levels occasionally remained normal, increased temperatures in Romanian and higher potential evapotranspiration levels are leading to more severe and persistent droughts. Mapping minimum SPEI values for medium- and long-term drought conditions at the river basin-level confirms overall dry conditions across most of the country over the last years, with only limited medium-term improvement in 2021 for some parts of central Romania (Figure 11). Figure 9: Medium-term (SPEI12) drought index for the period 1951-2022 by RBA Source: WB water team’s analysis using SPEI Global Drought Monitor and CRU datasets 22 https://drought.emergency.copernicus.eu/data/factsheets/factsheet_spi.pdf 17 Figure 10: Long-term (SPEI48) drought index for the period 1951-2022 by RBA Source: WB water team’ analysis using SPEI Global Drought Monitor and CRU datasets Figure 11: Minimum values for medium- (SPEI12) and long-term (SPEI48) drought indices for the period 2012-22 by RBA Source: WB water team’s analysis using SPEI Global Drought Monitor and CRU datasets 68. River flow has been consistently decreasing in Romania since 2011 reflecting the SPI12 and, especially, SPEI12 trends. The meteorological drought conditions identified using the SPI and SPEI clearly translate into hydrological drought, with reduced water levels in Romania’s rivers. Standardized yearly river discharge is strongly correlated with the short-term (SPEI12) drought (Figure 12). The graph shows an almost perfect match between river outflow in Romania and short-term drought conditions and reveals dwindling water resources during the last decade. Yearly river discharge is similarly correlated with short-term drought at the RBA-level (Figure 13). Figure 12: Standardized yearly river discharge in Romania (orange) compared with short-term drought index (SPEI12) (blue) (1990 to 2022) Source: WB water team’s analysis using Eurostat, ANAR and SPEI Global Drought Monitor data 18 Figure 13: Standardized yearly river discharge (orange) compared with short-term drought conditions (SPEI12) (blue) for 2010-2021 for each RBA Note: Interpretation of the SSFI values: extremely dry (SSFI < -2.00), severely dry (-1.99 < SSFI < -1.50), moderately dry (-1.49 < SSFI < -1.00), normal (-0.99 < SSFI < 0.99), moderately wet (1.00 < SSFI < 1.49), very wet (1.50 < SSFI < 1.99), extremely wet (2.00 < SSFI)23. Source: WB water team’s analysis using Eurostat, ANAR and SPEI Global Drought Monitor data 3.2 Climate projections 69. This section provides an overview of the future drought conditions relying on how climate change impacts the drought hazard. The socio-economic developments/changes were not considered for this assessment. 70. Temperatures in Romania are expected to increase, especially during the summer months. According to the Coupled Model Inter-comparison Project (CMIP6)24, trends in future precipitation for Romania do not concur across climate scenarios. However, all climate scenarios indicate temperatures are expected to increase (Figure 14). This increase will be especially noticeable during summer months (Figure 15). 71. Given the potential for rising temperatures and heat waves to exacerbate drought and water scarcity in Romania, future projections of drought indices under different climate change scenarios for the period 2021–2100 were made under four different Representative Concentration Pathways (RCPs): 2.6 (very stringent mitigation scenario), 4.5 (intermediate mitigation scenario), 7.0 (baseline scenario), and 8.5 (business as usual scenario), using five Global Climate Models (GCMs): GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI-RSM2-0, UKESM1-0LL. Therefore, water demand is fixed to a historical representative level (i.e., the year 2015). 72. Temperature increases are predicted to drive evapotranspiration losses and intensify ongoing dry conditions across the country. 23 Yildirim et al. (2022) 24 Viewer available in https://climateknowledgeportal.worldbank.org/country/romania/climate-data- projections 19 Figure 15: Historical and projected changes in mean Figure 14: Historical and projected mean temperatures across temperatures across Romania (1951-2100) by month under Romania under different climate scenarios (2000-2100) different climate scenarios (2000-2100) w Note: Reference period = 1995-2014, Multi-Model Ensemble. Source: Coupled Model Inter-comparison Project 6, World Climate Research Program 73. Overall, Romania is predicted to experience more frequent and/or more intense drought conditions under climate change. The predicted mean and interquartile range (IQR)25 for each standardized drought index (SPI, SPEI, and SSFI) under RCP 2.6 (“very stringent”) and RCP 8.5 (“business as usual”) climate scenarios were plotted for Romania (figures 16 and 17). Under RCP 2.6, the country becomes a bit wetter, with average and minimum SPI values increasing relative to the baseline, as well as minimum values. However, the SPEI shows a decreasing trend, implying water deficits will become problematic overtime. Under RCP 8.5, all three drought indices indicate worsening drought conditions, most significantly between 2050-2060. An overall agreement between GCMs is observed for both RCPs. Figure 16: Future drought indices in Romania under climate Figure 17: Future drought indices in Romania under climate change (RCP 2.6) with a three-month accumulation period change (RCP 8.5) with a three-month accumulation period Note: Anomalies are relative to the historical baseline (1990 -2019) Source: WB water team’s analysis using data from five Global Climate Models (GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI- RSM2-0, UKESM1-0LL) 25 The interquartile range (IQR) shows the statistical dispersion of the values and could be used to detect outlier values. 20 4 Socioeconomic and environmental impacts and current and future drought risk in Romania 74. Severe drought conditions have caused important damages in Romania over the time. For example, multiple sectors across the country, including agriculture, water supply, energy and transport, reported experiencing drought-induced losses during the 2022 drought. However, drought impact data were not systematically collected at the time, making it difficult to quantify the magnitude of damages. Moreover, while the impacts of drought on agriculture are well- known, the impacts on other sectors, such as water, energy and transport, are less understood. To address this data gap and better understand drought-induced socioeconomic impacts in Romania, an ex-post impact assessment of historical droughts in Romania (with an emphasis on the 2022 drought where possible) was undertaken, followed by a comprehensive assessment of current and future drought risk. 75. This chapter presents the results of these assessments (drought impacts and current and future drought risks) focusing on agriculture, energy production, inland fluvial transportation, industrial productivity (only the impact was assessed), natural ecosystems, water supply sectors. 76. While the impacts assessment process was driven by the source and type of information that could be collected to cover a longer period of time and the entire territory of the country, the risk assessment process was a rigorous one, more detailed information being provided in Chapter 2 and Annex 3. 77. As mentioned before, the future projections were made for the period 2021 - 2100 under four different scenarios corresponding to Representative Concentration Pathways (RCPs): 2.6 (very stringent mitigation scenario), 4.5 (intermediate mitigation scenario), 7.0 (baseline scenario), and 8.5 (business as usual scenario), using five Global Climate Models (GCMs). The period 1990 – 2019 was considered as baseline period for these scenarios. 78. The performance of the ML models to calculate the drought risk in current and future climate conditions was assessed using precision and balanced accuracy (BACC) metrics26 and the scores were, in general, excellent. 79. As mentioned before, the current and future drought risks are expressed as an average annual loss, and not in monetary values. The advantages of using a relative metric for risk assessment were presented in section 2.2. 4.1 Agriculture, livestock and fisheries 4.1.1 Historical impacts of drought 80. Agriculture is an essential part of Romania’s economy and highly vulnerable to drought. More than half of Romania’s land area (57 percent) is dedicated to agriculture, which employs 18 percent of the country’s working population and contributes 4 percent of the national gross domestic product (GDP)27. Though Romania is the one of the largest producers of cereals in the European Union, most of Romania’s farmers are smallholders who rely on subsistence farming making them especially vulnerable to drought28. The cascading and compounding impacts of drought in the agricultural sector are far-reaching: reduced water can lead to increased irrigation costs and reduced crop yields, loss of livestock and pasture, soil degradation and, eventually, food 26 Precision quantifies the number of correct positive predictions made out of positive predictions made by the ML model. BACC is used to assess the performance of the ML model to calculate the drought risk. In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 27 World Bank, 2024. 28 Directorate-General for Agriculture and Rural Development, 2024. 21 insecurity. These impacts lower profits for farmers, raise prices for consumers, and can impact long-term agricultural productivity and economic growth. 81. This study assessed the impacts related to agricultural productivity, livestock productivity, expenses with veterinary services, aquaculture productivity and the main findings are presented below. 82. Drought negatively impacts agricultural productivity. Historically, annual agricultural output at the regional level has been negatively correlated with drought (figure 18). During the 2022 drought, several regions suffered from drought-imposed water restrictions, notably the Dobrogea-Litoral, Prut-Barlad and Buzau-Ialomita RBAs (figure 19). Although the severe drought lasted approximately six months, beginning in February/March and peaking in August, in Prut- Barlad RBA, water restrictions continued into 2023. In line with historical data, maize production decreased country-wide during the 2022 growing season. Total yield and yield per hectare were generally low across the country (figures 20 and 21), but some of the most impacted areas included those RBAs that had experienced substantial restrictions in water allocations. Figure 18: Annual agricultural output (millions of Euros) by development region (primary axis) compared with extreme drought (blue dashed line on secondary axis) over time Source: WB water team’s analysis using Eurostat, Tempo Online Romania and SPEI Global Drought Monitor data Figure 19: Number of locations within RBAs that experienced consistent restrictions in monthly water allocations for irrigation Source: WB water team’s analysis using ANAR data 22 Figure 20: Maize production (tons) compared with county Figure 21: Maize yield (kg/ha) compared with county average average (April to September 2022) (April to September 2022) Source: WB water team’s analysis using remotely sensed data 83. Livestock productivity decreases substantially during drought periods, while veterinary expenses increase. Around half of the country’s livestock production is concentrated in three development regions: Sud - Vest Oltenia (averaging EUR 367 million per year), Centru (averaging EUR 303 million per year), and Nord - Vest (averaging EUR 350 million per year)29. Productivity decreased substantially in all three regions during drought periods (figure 22), with EUR 36 million per year in Sud - Vest Oltenia, EUR 28 million per year in Centru and EUR 14 million per year in Nord - Vest. Meanwhile, the average cost of veterinary services increased by 10 million Euros during drought conditions in all three regions (figure 23). Figure 22: Average annual livestock production (million Euros) during non-drought (blue) versus mild drought conditions (brown) by region (2005 - 2020) Note: Boxplots represent the range of average annual livestock, expressed as monetary value, under different climate conditions (non-drought and mild drought). Source: WB water team’s analysis using Eurostat and SPEI Global Drought Monitor data 29 Eurostat dataset on animal production at basic price in million euros for the period 2005-2020. 23 Figure 23: Average annual veterinary costs (million euros) during non-drought (blue) versus mild drought conditions (brown) conditions by region (2005 - 2020) Note: Boxplots represent the range of average annual veterinary costs under different climate conditions (non-drought and mild drought). Source: WB water team’s analysis using Eurostat and SPEI Global Drought Monitor data 84. Aquaculture is a relatively small but important water-dependent industry in Romania with approximately 135,000 ha of freshwater dedicated to fish farming. There are approximately 1,000 fish farms in Romania. The Sud-Est region is home to the largest facilities with over 89,000 ha (representing 66 percent of the country’s facilities), while Sud-Muntenia is home to the largest number of farms, with 283 (figures 24 and 25). Fish farms are operated almost exclusively in freshwaters, and approximately 83,000 ha of existing farms overlap with protected natural areas illustrating the interconnectivity between sectors30. Figure 24: Distribution of farm fish facilities surface per Figure 25: Distribution of farm fish facilities number per development region development region Source: Ministry of Agriculture and Rural Development (2021) 85. Reduced water allocations to fisheries risk impacting productivity. In Romania, aquaculture production was 11,714 tons31 in 2021, slightly above that of Bulgaria (10,725 tons), but below that of the Czech Republic (20,991 tons), Hungary (17,995 tons) and Malta (16,410 tons)32. During the 2022 drought, fisheries in Dobrogea-Litoral, Prut-Barlad, Olt and Crisuri RBAs were impacted by water restrictions (figure 26), potentially affecting productivity and the livelihoods of Romania’s fishing communities. 30 Ministry of Agriculture and Rural Development (2021) 31 1.04 percent of aquaculture production in the EU. 32 https://www.agerpres.ro/economic-extern/2023/10/03/productia-romaniei-din-acvacultura-este-la- jumatate-fata-de-cea-a-cehiei--1179598 24 Figure 26: Number of locations within RBAs that experienced consistent restrictions on water allocated to fisheries Source: WB water team’s analysis using ANAR data 86. The impact expressed as annual yield of 18 different crops between the years 1990 –2019 provided at a NUTS 3 level from the National Institute of Statistics TEMPO Online statistics Romania was used as input to evaluate the drought risk for agriculture. 4.1.2 Current and future drought risk 87. In case of agriculture sector, the current and future drought risks were calculated for 18 crops and crop groups, accounting for 98 percent of Romania’s crop and fodder production in 2019 (Table 3). 88. Cereals and wheat crops face important potential reductions in crop yields around the country, showing an increased reduction trend from north to south. Current drought risk for cereals is high country-wide, with an AAL of 6.3 to 9.1 percent (figure 27). With the exception of Prut-Barlad, drought risk for wheat — a very important reference crop — is similarly high around the country. Overall, the current drought risk for cereals and oil-crops is significant with an AAL of 7.4 and 7.7 percent, respectively, followed by potatoes (4.7 percent) and fruit and vegetables (4.5 percent). The current drought risk for fodder crops is comparatively low (3.9 percent). Table 3: Current and future average annual crop losses associated with drought risk in Romania Crop AAL (Range) AAL (Mean) Future AAL (Mean*) Cereals Average 6.3% - 9.1% 7.4% 12.8% - 13.7% Barley 4.8% - 8.7% 6.6% 12.9% - 14.1% Maize 6% - 10.1% 8.2% 13.4% - 14.7% Wheat 3.2% - 8% 6.3% 11.4% - 12.3% Cereals other 5.5% - 9.9% 7.3% 12.5% - 14.6% Fruit and Vegetables 3.5% - 5.6% 4.5% 11% - 11.4% Average Cabbage 1.3% - 4.7% 3.2% 5.1% - 5.7% Garlic 1% - 6.1% 3.6% 7.5% - 7.6% Melons and watermelons 4.4% - 10% 7.1% 15.4% - 16.2% Onion 1.7% - 5.5% 3.5% 8.5% - 9.6% Green pepper 1.5% - 4.4% 2.8% 6.9% - 7.3% Tomatoes 2% - 5.8% 4.2% 16.1% - 16.7% Oil crops Average 6.1% - 10.2% 7.7% 17.4% - 18.1% Rapeseed 8.4% - 14.3% 11% 13.1% - 14.4% Soya 7.6% - 12.8% 9.4% 16.5% - 17% 25 Crop AAL (Range) AAL (Mean) Future AAL (Mean*) Sunflower seed 4.3% - 9.8% 6.8% 18.3% - 19.1% Linseed 10.2% - 21.1% 15.4% 20.7% - 20.9% Potatoes 3.4% - 6.7% 4.7% 13.5% - 14% Fodder Average 3% - 5.9% 3.9% 8.6% - 9.4% Note: *The future column provides the range of different RCPs average Source: WB water team’s analysis using the modelling results Figure 27: Average annual loss for major crop groups associated with current drought risk by RBA Source: WB water team’s analysis using the modelling results 89. Maize is particularly vulnerable to drought, and currently facing a potential reduction in crop yields of 8.2 percent. The most common cereal crop in Romania is maize, accounting for 59 percent of total cereal production. At the river basin scale, maize losses also increase from north to south (figure 28) and range from 6 percent in Somes-Tisa RBA to approximately 10.1 percent in the Southern basins. 90. Future conditions imply even greater drought-associated losses for all crop groups and most crops around the country. AAL is predicted to increase by a magnitude of two - to four-fold in the future (figure 29), rising as high as 13.7 percent for cereals, 18.1 percent for oil crops, and 14 percent for potatoes (table 3). Fruit and vegetables and fodder losses are predicted to increase two - to three- fold (figure 28), with future AAL rising as high as 11.4 and 9.4 percent, respectively (table 3). For specific crops, sunflower seeds and tomatoes are expected to experience the highest AALs, reaching up to 19.1 and 16.7 percent, respectively (table 3). Other main crops expected to experience high losses include wheat and barley (reaching as high as 12.3 and 14.1 percent, respectively), melons and watermelons (up to 16.2 percent) and onions (up to 9.6 percent) (table 3). 26 Figure 28: Average annual loss of maize crops associated with current drought risk by RBA Source: WB water team’s analysis using the modelling results 91. Future crop reductions of cereals and fruit, and vegetables, are predicted to be relatively high across all river basin administrations. Four river basin administrations with medium-sized oil crops are expected to experience an increase in magnitude of AAL of over 200 percent (figure 29), with actual losses predicted to range from 19.2 percent in Siret RBA (RO10) to 24.7 percent in Dobrogea-Litoral RBA (RO6) (table 3). Overall, AAL magnitude for fodder crops is predicted to increase more than 100 percent in five river basin administrations in Southern and Central Romania (figure 29) and more than 300 percent in Buzau-Ialomita RBA (RO5) and Dobrogea-Litoral RBA (RO6) where reductions in crop yields are predicted to reach as high as 18.5 and 19.2 percent, respectively (table 3). Figure 29: Predicted change in average annual loss of major crop groups associated with future drought risk (relative to 1990-2019 baseline) by RBA Note: Future risk is presented as relative change in AAL to assess the magnitude of differences between the risk in current conditions and the risk in future climate conditions. Source: WB water team’s analysis using the modelling results 27 92. Reductions in maize crop yields are predicted to be high, but variable, across all river basin administrations. Overall, AAL for maize crops at the river basin administration-level is expected to increase by at least 28 percent but as high as 127 percent (figure 30). Future drought risk is expected to hit maize crops in the Jiu RBA (RO2) the hardest (figure 30), where AAL is predicted to reach 23 percent. Figure 30: Predicted change in average annual loss of maize crops associated with future drought risk (relative to 1990- 2019 baseline) by RBA and climate change scenario Note: Future risk is presented as relative change in AAL to assess the magnitude of differences between the risk in current conditions and the risk in future climate conditions. Source: WB water team’s analysis using the modelling results 93. The results regarding the model performance evaluation, based on the precision and BACC metrics33, are relatively high for most crops and, in general, with precision values ≥ 0.95 and BACC values ≥ 0.85. 4.2 Energy production 4.2.1 Historical impacts of drought 94. Hydropower is the most important source of electricity in Romania, making the energy sector especially vulnerable to drought. Hydropower represents one-third of all electricity generated in Romania (figure 31)34. Drought can have a significant impact on energy production, particularly in areas where hydropower is a significant source of electricity. Hydropower plants rely on a steady supply of water to generate electricity; if water levels in reservoirs drop too low, the amount of electricity that can be generated may be reduced. Similarly, nuclear power plants, which generate 20 percent of Romania’s electricity, rely on water from nearby rivers or lakes for cooling, and may have to shut down (as it was the case in 200335) or operate at reduced capacity if water levels drop too low. 33 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 34 Statista (2024) 35 Zamfir, A. (2003) 28 95. This study assessed the drought impacts on hydropower production over two decades to highlight the link between droughts and reduction of hydropower production. Figure 31: Distribution of electricity production in Romania in 2023, by source Source: Statista, 2024. 96. Hydropower production is significantly reduced during drought months. Over the last decade, dry events have been particularly problematic for the energy sector and drought months yield significantly lower hydropower production (figure 32). In 2022, hydropower production in Romania was 15 percent less than the country’s annual average. Hydropower production was even lower in 2012 (75 percent of the annual average), as well as in 2017, 2019 and 2020. Monthly hydropower production was 1,328,645 MWh during the period 2000-2022. On average, production was 18 percent less during moderately dry months, and 25 percent less during severely dry months (figure 33). Figure 32: Monthly SPEI6 series and standardized36 total hydropower production in Romania, 2000-2022 Source: WB water team’s analysis using SPEEH Hidroelectrica SA Romania and SPEI Global Drought Monitor data 36 Standardization is a scaling method and it was used for comparison purposes. 29 Figure 33: Total hydropower production in Romania during different drought conditions (boxplots), compared with average monthly production of the entire series (black line) Note: Boxplots represent the range of hydropower production under different climate conditions. The horizontal dashed line is the average monthly production. Source: WB water team’s analysis using SPEEH Hidrolectrica SA Romania and SPEI Global Drought Monitor data 97. The impact expressed as volumes of water used for hydropower production between the years 2009 – 2020 provided by ANAR was used as input to evaluate the drought risk for energy production. 4.2.2 Current and future drought risk 98. Drought risk to energy production is increasing around the country with the potential to impact Romania’s installed hydropower capacity. About half (46 percent) of Romania’s hydropower plants are run-of-the river (ROR) facilities and rely on river flow to generate electricity. The other half (54 percent) have storage to offset intra-annual and annual changes in water availability. ROR facilities are directly impacted by changes in river flow, while impacts on hydropower plants with storage may be delayed. Because drought indicators rely on meteorological data and don’t capture storage capacity, drought risk presented in this study is more likely to be a function of drought risk faced by ROR plants, which in any case generate 63 percent of the country’s hydroelectricity. Furthermore, 25 percent of Romania’s installed hydropower capacity is located on the Jiu RBA, which is connected hydrologically to the Danube, which as a transboundary river and is not captured in this analysis37. Hydroelectricity production was shown to be negatively correlated with drought conditions in Romania (shown in figures 12 and 13), thus data limitations aside, the results of this analysis are intended to be indicative of risks facing Romania’s energy sector. A case-by-case analysis at either the RBA or hydropower plant-level would likely produce more definitive results. 99. Current and future drought risks to the energy sector are measured as a reduction in the amount of water used to produce hydroelectricity, currently 6.3 percent, nationally. At the RBA level, the reduction in water used to produce hydroelectricity due to droughts ranges from approximately 2 percent in Northern Somes-Tisa RBA (RO9) and Siret RBA (RO10) to 8.2 percent in Olt RBA (RO3) and 11.2 percent in Arges-Vedea RBA (RO4) (figure 34). 37 SPEEH Hidroelectrica SA: https://www.hidroelectrica.ro/article/20 30 Figure 34: Current drought risk (average annual reduction in water use) for hydroelectricity generation by RBA Source: WB water team’s analysis using the modelling results 100. Location and capacity of the hydropower plants along the transboundary rivers should be well assessed when calculating the drought risk. In case of the Jiu RBA (RO2), where approximately 25 percent of Romania’s installed hydropower capacity is located, the results show a lower reduction in water use (2.1 percent) than most other basins. This can be explained by the location and capacity of the Iron Gate hydropower plant on the Danube River, meaning its performance is linked more to hydrological conditions in the upper Danube basin than to the conditions in the Jiu RBA. This points to the importance of exploring such linkages in case of hydropower plants located on transboundary rivers in the future. Similarly, most river basins experience relatively high losses with relatively short return periods; however, in the Jiu RBA case, documented losses do not exceed 15 percent, observed in the loss exceedance curves (figure 35) and the observed hydropower generation in the Iron Gates hydropower plant (figure 36). Figure 35: Drought loss exceedance curves for the energy sector in four selected river basin administrations in Romania Source: WB water team’s analysis using the modelling results 31 Figure 36: Electricity generation at the Iron Gate Hydropower Plant from 2000-2022 Hydroelectricity production in Iron Gate Hydropower Plant Annual Electricity Generation 8.0 7.5 7.0 [TWh] 6.5 6.0 5.5 5.0 Year Note: dotted line = trend Source: SPEEH Hidroelectrica SA 101. Overall, the hydroelectricity production in Romania associated with droughts is expected to decrease in the future. The decrease of volume of water used to produce hydroelectricity is predicted to increase in magnitude by more than 100 percent in the future, suggesting heightened drought risks. Some RBAs, i.e., Arges-Vedea RBA (RO4) and Buzau-Ialomita RBA (RO5) are expected to experience relatively high changes in the magnitude of water used (figure 37), driving average annual losses of at least 16.2 percent. In other cases, future trajectories imply a slight reduction in AAL (meaning an increase in production) and/or mixed results under different climate change scenarios. Nevertheless, even RBAs facing a relatively lower drought risk at present, such as Jiu RBA (RO2), Somes-Tisa RBA (RO9), and Siret RBA (RO10), are still predicted to experience important losses in the future. Figure 37: Future drought risk (changes in AAL in water use) for hydroelectricity production by RBA (relative to baseline) Note: Future risk is presented as relative change in AAL to assess the magnitude of differences between the risk in current conditions and the risk in future climate conditions. Source: WB water team’s analysis using the modelling results 32 102. The results regarding the model performance evaluation, based on the precision and BACC metrics38, are high, the precision nearly reaches 100% and the balanced accuracies is approximately at 95%. 4.3 Inland fluvial transportation 4.3.1 Historical impacts of drought 103. Total fluvial transportation in Romania decreases during drought conditions. During a drought, water levels in rivers and other bodies of water can drop significantly, making them shallower and reducing the depth available for navigation. About 30 percent of the country’s freight transport relies on fluvial transport, playing a very important role in the national and international transport, especially in terms of Romania's trade with neighbors in Central and Eastern Europe39. 104. This study assessed the drought impacts on inland fluvial transportation over almost two decades to highlight the link between droughts and the reduction on freight transport. 105. Total annual transport was 4.6 percent less in 2022 compared with other years and almost 20 percent less during the third quarter (July-September), coinciding with peak drought conditions (figure 38) and government restrictions placed on fluvial transport during the summer of 2022, and especially in August. Figure 38: Fluvial transport (standardized40 tons by quarter) (green lines) compared with drought conditions (bars) in Romania 2005 - 2022 Source: WB water team’s analysis using Eurostat and SPEI Global Drought Monitor data 106. Average tonnage is negatively impacted by drought conditions. Although interannual data are highly variable, average tonnage is reduced by around 300 thousand tons per vessel during mild drought conditions (figure 39). 38 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 39 Comité National Routier, Romanian National Union of Road Hauliers (2020). 40 Standardization is a scaling method and it was used for comparison purposes. 33 Figure 39: Total transport in thousands of tons per trimester by vessel (green boxplot) compared with mild drought conditions (brown boxplot) in Romania during 2005 - 2022 Note: Boxplots represent the range of all transport by vessel by quarter under different climate conditions (non-drought and mild drought). Source: WB water team’s analysis using Eurostat and SPEI Global Drought Monitor data 107. Tonnage has generally fluctuated over time, reaching a high of slightly more than 33 million tons in 2019. Reductions in transport were observed in 2009, 2013, 2017, and 2018 (figure 40). These reductions may be associated with droughts but may also be driven by national and global economic conditions. Figure 40: Tonnage (million tons) transported via inland fluvial systems Source: WB water team’s analysis using EUROSTAT data 108. The impact expressed as tonnage of goods transported in vessels between the years 2005 – 2019 provided by National Institute of Statistics TEMPO Online Statistics Romania was used as input to evaluate the drought risk for fluvial transport. 34 4.3.2 Current and future drought risk 109. The AAL under the current conditions of drought for the volume of goods (loadings/unloadings) transported by fluvial system in Romania is estimated to be 1.4 percent. 110. Future fluvial transport is expected to decrease in the long-term as a function of drought in Romania. Reductions in tonnage vary but are expected to grow across all future climate scenarios (figure 41). Greater impacts are predicted to materialize in the long-term (RCP 7 and 8.5), increasing from 1.8 percent to 3.1 and 3.3 percent, respectively. Figure 41: Future drought risk on inland fluvial transportation (AAL) under short- and long-term scenarios Note: Boxplots represent the median and range of AAL for inland river transport under all climate models. The horizontal line represents the historical (baseline) value. Source: WB water team’s analysis using the modelling results 111. The results regarding the model performance evaluation, based on the precision and BACC metrics41, are high, both precision and balanced accuracy are very high (> 0.98). 4.4 Industrial productivity 4.4.1 Historical impacts of drought 112. Water restrictions were imposed on industries in several RBAs during the 2022 drought. Drought can have a significant impact on industry, particularly for sectors that either rely on water as a key input or depend on a reliable supply of water for operations. Reduced water availability, increased costs and supply chain disruptions are among the primary effects of drought on industry. In the last decade, the Romania’s industry contribution to GDP ranged between 26 to 33 percent42. 113. This study assessed the drought impacts on industrial productivity over five years (2017 – 2022) to highlight the link between droughts and reduction of productivity. 41 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 42 Statista, 2024 35 114. During the 2022 drought, but especially August and September, three RBAs in the east of the country (Buzau-Ialomita, Dobrogea-Litoral and Prut-Barlad RBAs) were forced to restrict industrial water allocations in dozens of locations (figure 42). Figure 42: Number of locations that suffered consistent, monthly industrial water restrictions by RBA during the 2022 drought Source: WB water team’s analysis using ANAR data 115. Industrial productivity in Romania is negatively correlated with drought conditions. Monthly industrial productivity indices were compared with the SPEI12 values. Although the industrial disruption cannot be linked entirely with the drought, it should be noted that the productivity figures from July 2022 are systematically low (and the lowest next to COVID) and correlate moderately strongly with drought conditions (figure 43). The strength of this relationship does not change significantly when the three months of disruption directly related to COVID are excluded, pointing to the reliability of these results. Figure 43: Standardized43 monthly industrial production and compared with drought conditions in Romania, 2017-2022 Source: WB water team’s analysis using Tempo Online and SPEI Global Drought Monitor data 116. Insufficient data sets were available at the moment of elaboration of this study to estimate current and future drought risk on industrial productivity in Romania. The statistical connection with the drought is not as evident as with other sectors and systems, but it is worth exploring it in the future with longer, more granular and more complete datasets. 43 Standardization is a scaling method and it was used for comparison purposes. 36 4.5 Natural ecosystems 4.5.1 Historical impacts of drought 117. While socioeconomic impacts tend to receive the most attention, the potential environmental and landscape impacts of drought are often disregarded. This is partly due to a lack of data but also because their effects may not be immediately apparent or clearly linked to drought. However, the degradation of natural resources and landscapes resulting from drought impacts the health of ecosystems and can have a significant impact on various economic sectors, such as agriculture, livestock farming, and tourism. It is crucial to examine some of the risks associated with environmental drought in order to mitigate potential negative effects. 118. This study assessed the drought impacts in terms of number of wildfires and burned surface of natural ecosystems, as well as the net primary production (NPP) of terrestrial and aquatic ecosystems to highlight the link between droughts and reduction of natural ecosystems. 119. Wildfires are correlated with historical drought conditions in Romania. The burned area in Romania tends to increase with moderate seasonal drought and the SPEI4 scale is the one that connects better with this variable. Monthly burned area indices were compared with the SPEI4 values. The increase of burned areas (see the red hashed periods) are correlated well with the drought conditions (figure 44). Figure 44: Standardized44 burned area per month and drought conditions across Romania, 2000-2023 Source: WB water team’s analysis using MODIS data 120. Poor drought risk management practices may exacerbate wildfires. Recorded wildfires increased from 262 in 2021 to 960 in 2022; of these, 686 (45 percent) were recorded in March 2022 (figure 45). February and March 2022 registered the sharpest decline in drought conditions (rapid drought intensification) meaning a combination of direct (e.g., meteorological conditions) and indirect drought factors (e.g., using fire in agricultural practices) might explain the radical proliferation of fires. Fires are generally clustered in the driest areas (figure 46). 121. 2022 has been the worst year since recorded data on wildfire exists. It’s not only the increased number of forest fires, but as well the burned surface which was almost double per wildfire compared with the situation from 2021, and this resulted in increased time periods to extinguish them. In 124 cases, the time interval to extinguish the wildfire was longer than 24 hours45. 44 Standardization is a scaling method and it was used for comparison purposes. 45 Joint Research Centre (JRC), 2023 37 122. The impact expressed as Net Primary Production (NPP46) between the years 2001 – 2022 collected by the MODIS sensor on the Terra satellite was used as input to evaluate the drought risk for natural ecosystems, as presented in the following sub-chapter. Figure 45: Number of wildfires extinguished in Romania (2021/22) Source: WB water team’s analysis MEWF and the National Directorate of Forests – Romsilva data Figure 46: Number of fires extinguished in 2021 and 2022 compared with March 2022 drought conditions across Romania Source: WB water team’s analysis using MEWF data 46 NPP is “the amount of biomass or carbon produced by primary producers per unit area and time”, “Obtained by subtracting plant respiratory costs (Rp) from gross primary productivity (GPP) or total photosynthesis. ” JRC, 2019. 38 4.5.2 Current and future drought risk 123. Drought is contributing to important losses in Romania’s ecosystems. Drought risk faced by Romania’s ecosystems is measured as a function of loss in net primary productivity (NPP). Changes in NPP within forest ecosystems associated with current drought risk show an average annual loss of about 2.9 percent nationally. At the RBA level, the reduction in NPP associated with droughts ranges from 2 percent (or below) in the interior of Romania to about 4 to 6 percent in the eastern most and western most regions of the country, e.g., Banat RBA (RO1), Crisuri RBA (RO8), Buzau- Ialomita RBA (RO5), and Dobrogea-Littoral RBA (R06). The largest impact is estimated in the Prut- Barlad RBA (RO11), located in the eastern most region (figure 47). 124. Overall, the NPP of aquatic ecosystems is reduced by about 2.8 percent at the national level. At RBA level the reduction in NPP due to droughts ranges between 2 and 2.5 percent in the eastern river basin administrations, e.g., Prut-Barlad RBA (RO11) or Siret RBA (RO10) to just less than 4 percent in the western river basin administrations e.g., Banat RBA (RO1), Jiu RBA (RO2), and Crisuri RBA (RO8). The most exposed river basin administrations, e.g., Dobrogea-Litoral RBA (RO6) and Buzau-Ialomita RBA (RO5), experience reductions in NPP of between roughly 2.5 and 3 percent (figure 48). Figure 47: Change in NPP of forest ecosystems Figure 48: Change in NPP of wetland ecosystems associated with current droughts conditions by RBA associated with current droughts conditions by RBA Source: WB water team’s analysis using the modelling results 125. Losses in the NPP of forest systems are expected to increase relative to baseline in four RBAs due to future drought risks. The decrease of NPP of forest ecosystems is predicted to intensify in magnitude in the Prut-Barlad RBA (RO11) and Dobrogea-Litoral RBA (RO6) by 120 percent across all four future climate scenarios (figure 49). These river basin administrations are anticipated to experience a reduction in NPP of between 13.8 and 15.1 percent and 6.7 and 8.1 percent, respectively. Additionally, NPP is expected to decrease in the Buzau-Ialomita RBA (RO5) and Jiu RBA (RO2), by between 5.1 and 6.7 percent and 4.1 and 4.5 percent, respectively. Conversely, the NPP of forest systems is expected to increase in the northern, central, and western RBAs the future. 126. Future drought risk is expected to substantially impact wetland ecosystems in the south and southeast of the country. The NPP of wetland systems is expected to decrease significantly in the southern and southeastern river basin administrations, where local tributaries join the Danube, and in the delta area (e.g., Dobrogea-Litoral RBA (RO6)), across all future climate change scenarios. In these RBAs average annual losses are expected to increase by a magnitude of between 180 and 270 percent (figure 50), resulting in future losses in NPP of between 11.3 and 12.7 percent in the Jiu RBA (RO2), 11 to 12.2 percent in the Arges-Vedea RBA (RO4), and from 6.2 to 7.4 percent in Dobrogea-Litoral RBA (RO6). The northern and western river basin administrations, which experienced the highest AALs historically are not expected to experience a change in NPP and in some cases may experience an improvement relative to baseline. 39 127. The results regarding the model performance evaluation, based on the precision and BACC metrics47, are high, both precision and balanced accuracy are very high (> 0.95). Figure 49: Future drought risk on forest ecosystems by RBA, compared to baseline Source: WB water team’s analysis using the modelling results Figure 50: Future drought risks on wetland ecosystems by RBA, compared to baseline Source: WB water team’s analysis using the modelling results 47 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 40 4.6 Water supply 4.6.1 Historical impacts of drought 128. Drinking water supplies around the country were severely interrupted by the 2022 drought. Interruptions/restrictions in water supply significantly impact household well-being and also challenge the functioning of social infrastructure and local businesses. Drinking water supply was interrupted in 220 communities forcing the Romanian General Inspectorate for Emergency Situations (IGSU) to deliver water in lorries to affected people. In many localities around the country the use of drinking water for non-essential activities (e.g., washing cars and watering gardens) was restricted for many months. The disruptions raised concerns surrounding the reliability of water supplies in Romania during droughts, especially as drought events are expected to increase in frequency and intensity due to climate change. 129. The National Administration of Romanian Waters (ANAR) manages (bulk) water resources as well as water storage (e.g., retention lakes, dams and reservoirs) across the country’s eleven RBAs. As the authority responsible for water resources management (WRM) and the supply and sale of bulk water, ANAR is required to monitor the availability of water resources and ensure the continuous supply of water to key sectors, including, for example, restricting non-essential water use when necessary. Imposing water restrictions can lead to reduced tariff revenues because of decreases in bulk water abstraction and sewage discharges. Meanwhile, lower water levels can lead to an increase in the concentration of contaminants requiring additional treatment and leading to increased operational costs. 130. This study assessed the drought impacts in terms of: the number of locations where water usage restrictions were applied, the volume of bulk water sold by ANAR, the impacts for ROCs and LOCs, and the abstracted volumes of water for water supply. 131. During the last six months of 2022, 342 restrictive measures were implemented across six river basin administrations, affecting public water supplies and water for fisheries, irrigation, and industry. Six RBAs were forced to restrict public drinking water supplies (figure 51). The affected locations were mainly in the eastern half of the country and included: Dobrogea-Litoral RBA (RO6) and Buzau-Ialomita RBA (RO5) (August and September 2022), Prut-Barlad RBA (RO11) (June 2022- January 2023), and Siret RBA (RO10) (July-September 2022). Two other river basin administrations were forced to implement restrictions: Crisuri RBA in the west (July and August 2022), and Olt RBA in the south (July 2022 to January 2023). While restrictions are essential to safeguard water resources, they impact the volume of bulk water sales. Further, controlling and enforcing restrictions, which fall to RBAs, result in additional costs. Figure 51: Number of locations that suffered consistent restrictions on public water supplies by RBA and month Source: WB water team’s analysis using ANAR data 41 132. The assessment of the raw volumes of water sold by ANAR to the various economic sectors starting with 2015, shows a correlation between the reduced water sales for energy production and the drought conditions. Each drought peak corresponds closely to a decrease in water sales to the energy sector, while wet years see an increase. This could represent a substantial decrease in revenues for ANAR and its subordinated institutions, as the contributions from the energy sector account for more than 50 percent of ANAR’s total revenues. 133. Water restrictions during drought periods appear to exert little influence on bulk water sales for drinking water. The volume of surface and groundwater sold to drinking water providers represents 8.8 and 4.2 percent of all bulk water sales, respectively, and remain quite stable over time. A graphical analysis of changes in bulk water sales for drinking water purposes by RBA compared with drought conditions shows there is limited correlation between the two (figure 52). 134. Operation and maintenance become more challenging during droughts, potentially leading to increased costs for ANAR. Overseeing bulk water and its delivery systems during droughts entails additional, drought-specific costs, such as clearing depleted river channels and monitoring water resources and demands more closely. Figure 52: Bulk water sales for drinking water compared with drought risk (secondary axis) by RBA 2015-2022 Source: WB water team’s analysis using ANAR and SPEI Global Drought Monitor Data 135. The findings of the interviews applied to the selected ROCs and LOCs revealed that: • Drought triggers an increase in water demand. Increasing demand is a known phenomenon that occurs regularly during the summer months e.g., June to September. The increase in demand was particularly sharp in 2022, when demand reportedly tripled for several ROCs/LOCs, especially amongst poorer, rural communities. This can generally be attributed to: ▪ Increased use of water for gardens and livestock. In periods of drought, these become unauthorized water uses. This points to a need to better monitor water use, as well as support structural development that addresses non-drinking water needs, especially given the reliance on subsistence farming in Romania. ▪ Increasing water demand during the drought paradoxically leads to increased volumes of non-revenue water (NRW). In several ROCs/LOCs NRW is as high as 60 percent meaning more than half the water being treated disappears before reaching the final consumer. Higher demand often requires high pressure in the system, leading to higher losses. In addition, service interruptions during drought months can also increase NRW, as the re-starting of the distribution system leads to higher ratios of NRW in the short-run. 42 • Significant decreases in groundwater levels in some wells were observed. This did not lead to impacts in terms of water scarcity/restrictions in all cases, but did in some. In the long run, this can have serious and even irreversible effects, with some wells becoming unusable, or bearing unsustainable production costs. Identifying the cause of the low levels is imperative, especially to ascertain whether this is a cumulative effect of repeated droughts in recent years, or the result of uncoordinated/unauthorized drillings. • Low water levels in rivers and reservoirs reportedly impact water quality during drought events. This can cause considerable impacts on operating costs and the wear and tear of equipment (e.g., pumps breaking down). • In several cases, respondents stated limitations in the delivery of bulk water through ANAR were not announced with sufficient time, making it difficult for the water service providers to plan and implement contingency measures. Further, it was noted that the regulation of water levels in reservoirs is not coordinated with the ROCs/LOCs. The water service provides requested a better exchange of information with ANAR to enable coordinated action on the management of the water reserves. • None of the ROCs or LOCs interviewed counted with a systematic monitoring of drought impacts or had undertaken a drought risk assessment. Increases in additional activities and operational expenses during times of drought are believed to strain the financial health of utilities, but little quantitative evidence is available. All water service providers interviewed would welcome support (e.g., through guidance) to systematically monitor drought impacts and assess drought risks. None of the interviewed water service providers have a drought risk management plan in place to better prepare for future drought events. During drought events, the ROCs and LOCs undertake the following activities: ▪ Increase water availability monitoring. In larger ROCs this is often automated with data coming in at very short time intervals and allowing “real-time” monitoring. Forecasting is less often practiced and then based on historical data. ▪ Awareness-raising campaigns to advocate for water conservation. The effectiveness of such campaigns has not been assessed yet. 136. The impact expressed as volumes of abstracted water for water supply between the years 2008 – 2020 from Eurostat was used as input to evaluate the drought risk for water supply, as presented in the following sub-chapter. 4.6.2 Current and future drought risk 137. A reduction of about 3.4 percent in water withdrawals nationally can be attributed to drought risks. Drought risk to public water supplies was measured as the change in bulk water delivered to utilities. At the river basin administration scale, the reduction in water withdrawals due to droughts ranges between 7 percent in the southern and eastern river basin administrations (figure 53) to roughly 1.5 percent in the northern and western river basin administrations. 138. More than half (55 percent) of Romania’s population lives in regions with average annual losses of 4 percent or higher. The greatest impact occurs in the most populated river basin administrations, including Arges-Vedea RBA (RO4) with a loss of 7.2 percent and Buzau-Ialomita RBA (RO5) with a loss of 6 percent. Together these basins account for over one-third (36 percent) of Romania’s population. Other river basin administrations with relatively high drought risks have populations ranging between 814,737 in Dobrogea-Litoral RBA (RO6) with an AAL of 4.2 percent to 1.8 million inhabitants in Prut-Barlad RBA (RO11) with an AAL of 4.9 percent. Reduced water withdrawals may lead to municipal water shortages, put further pressure on existing water sources, require the use of alternative sources, and may increase water abstraction and treatment costs or reduce water quality. 139. The AAL of overall water withdrawals is predicted to increase by up to three times across all climate scenarios (figure 54). RBAs that have already experienced the highest average annual 43 losses historically (southeastern and eastern) are expected to suffer the greatest increases, reaching up to 20.7 percent in Arges-Vedea RBA (RO4) and Buzau-Ialomita RBA (RO5), up to 19.3 percent in Prut-Barlad RBA (RO11), and 14 percent in Dobrogea-Litoral RBA (RO6). Meanwhile, the Somes-Tisa RBA (RO9) and Siret RBA (RO10) are predicted to experience significant improvements in water availability with losses smaller than 1 percent. However, less than 20 percent of the country’s population lives in these areas. Figure 53: Drought risk for bulk water deliveries to water utilities in current climate conditions by RBA Source: WB water team’s analysis using the modelling results Figure 54: Future drought risk for bulk water deliveries to water utilities by RBA, compared to baseline Source: WB water team’s analysis using the modelling results 140. The results regarding the model performance evaluation, based on the precision and BACC metrics48, are high, as the precision nearly reaches 100% and the balanced accuracies is higher than 95%. 48 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 44 5 Opportunities to strengthen drought risk management 141. The purpose of this study was to assess the impacts of the 2022 drought in Romania and to provide a comprehensive overview of current and future drought risk on different sectors and ecosystems in Romania, and, thus, to raise awareness among the relevant decision makers on the findings and to boost efforts towards increasing the resilience to climate risks. 142. Data have shown that various sectors and the natural ecosystems were significantly impacted by the 2022 drought and the risk is increasing under the current and future climate conditions. 143. In general, the scores of the precision and balanced accuracy metrics of the risk calculation models’ performance evaluation were excellent49: • Agriculture: for maize, wheat, sunflower seeds, potatoes, and the perennial fodder: precision ≥ 0.98 and BACC ≥ 0.9. In case of other crops, the precision score is higher than 0.95 and the BACC score is higher than 0.85. The ‘other edible roots’ category had the lowest performance (precision = 0.95, BACC = 0.5). • Energy production: the model performance is high as the precision nearly reaches 100% and the balanced accuracies is approximately at 95%. • Fluvial Transport: the model performance is high as both the precision and balanced accuracy almost reaches 100% (> 0.98). • Natural Ecosystems: the model performance is high for both ecosystems, as the precision and balanced accuracies are over 95%. • Water supply: the model performance is high as the precision nearly reaches 100% and the balanced accuracies is higher than 95%. 144. Although a formal analysis of the country’s existing drought risk management capacities did not form part of this study, valuable information on Romania’s approach to drought risk management was nevertheless gleaned through the process. This was done by relevant literature reviews, discussions with key stakeholders (including MEWF, ANAR, INHGA, RBAs, IGSU, ARA, ANRSC and the water service providers), discussions and feedback collected from participants at the conference “Drought Risk and Drought Risk Management in Romania and in Europe” 50. The event, held October 30th and 31st, 2023 in Bucharest, was organized by MEWF, the Danube Water Program and the World Bank, and counted 85 participants from European and national governmental institutions, donors, academic institutions, NGOs and other stakeholders. 145. Based on the drought risk assessment and this additional information, several opportunities to strengthen Romania’s capacity for drought risk management have been identified and presented in this chapter. Given the report’s emphasis on identifying risk rather than drought management, this list is not intended to be exhaustive, but to create momentum by helping policymakers and sector specialists begin the process towards greater water-, food-, and energy- security through enhanced disaster risk management. Summary of key drought hazards: • Romania has grown increasingly dryer since 1980, caused primarily by rising temperatures. • River flow in Romania has been consistently decreasing since 2011. • Under climate change, temperatures in Romania are expected to increase further, especially during the summer months. • Rising temperatures and more frequent and prolonged heat waves are predicted to intensify persistent drought conditions across the country. 49 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 50 https://www.iawd.at/eng/event/809/details/w/0/drought-risk-i-drought-risk-management-in-romania-and- in-europe/ 45 Summary of the drought impacts and risks: Agriculture Impact: in 2022 over 160,000 hectares were affected by a soil moisture deficit causing a 25-30 percent reduction in agricultural output, an increase in animal food prices and a reduction in livestock numbers. Risk: ▪ Present conditions: the AAL in crop yields ranges between 3.9% for fodder, 4.5% for fruit and vegetables, 4.7% for potatoes, to 7.4% for cereals and 7.7% for oil crops. ▪ Future conditions: the AAL in crop yields due to drought ranges between 8.6 – 9.4% for fodder, 11.0 – 11.4% for fruit and vegetables, 13.5 - 14% for potatoes, to 12.8 – 13.7% for cereals and 17.4 – 18.1% for oil crops. Energy production Impact: in 2022 only 85% of total yearly energy production was reached. Risk: ▪ Present conditions: Water-use for hydro-electricity production shows an AAL of about 6.3% at the national level. ▪ Future conditions: Changes in AAL in the future under different RCPs mostly worsen drought impacts on energy production, where all currently high impact regions expect significant increases, reaching AALs due to drought of 16.2% -19.9%. Fluvial transport Impact: in 2022 a 4.6% reduction in transport was observed compared to normal yearly values, with only 81% of the average transportation expected in Q3. Risk: ▪ Present conditions: the reduction in goods loadings/unloading in Romania associated with drought is estimated almost at 1.4%. ▪ Future conditions: Changes in AAL due to drought in the future for most RCPs are expected to worsen the drought impacts on inland water transport, in RCP 7.0 and 8.5 are expected to reach 3.05% and 3.25%, respectively by the end of the century. Industrial productivity Impact: Although the industrial disruption cannot be linked entirely with the drought, it should be noted that in 2022 the industrial productivity is systematically lower and correlate with drought conditions. Risk: The lack of detailed data didn’t allow the construction of a drought risk model for this sector, and the risk was not calculated. Natural ecosystems Impact: 2022 has been the worst year since recorded data on wildfire exists: not only their number increased, but as well the interval for extinguish them and the burned surface per fire doubled (more than 10,000 hectares were affected). Risk: ▪ Present conditions: The AAL due to drought for the NPP of forest ecosystem shows an AAL of about 2.9% at the national level, reaching the maximum value in Prut-Barlad RBA (6%). The AAL for the NPP of the aquatic ecosystems is about 2.8% at the national level and has increased values in Banat RBA, Jiu RBA and Crisuri RBA (> 3.5%). ▪ Future conditions: The AAL for NPP of the forest systems is expected to increase relative to the baseline only in 4 RBAs: Jiu RBA (4.1% -4.5%), Buzau-Ialomita RBA (5.1% - 6.7%), Dobrogea-Litoral RBA (6.7% - 8.1%) and Prut-Barlad RBA (13.8% - 15.1%). The AAL for NPP of the wetland systems is expected to further increase mainly in the southern and southeastern river basins, where local tributaries join the Danube, and in the delta area: Jiu RBA (11.3% - 12.7%), Arges-Vedea RBA (11% - 12.2%) and Dobrogea-Litoral RBA (6.2 - 7.4%). Water supply Impact: in 2022 reduced river flows disrupted public water supplies in 220 localities, restricting non-essential water use in many parts of the country. Risk: ▪ Present conditions: The AAL due to drought of the volumes for public water supply is estimated 3.4% loss at national level, reaching over 7% in the Arges-Vedea RBA. The impacts are highest in the most populated river basin administrations, and over 55% of Romania’s population live in river basins w ith AAL larger than 4%. ▪ Future conditions: The river basin administrations with the highest impact at current conditions, Arges- Vedea RBA and Buzau-Ialomita RBA also show the highest increases in the AAL in the future RCP scenarios, reaching average losses of 12.8% - 20.7%. 46 5.1 Recommendation 1: Conduct a comprehensive DRRA and introduce proactive drought risk management practices 146. Romania already has an advanced policy and institutional framework for drought risk management. Romania is one of only five EU member states to have a drought management strategy “in place” that covers the whole country51. Furthermore, Romania is one of about 10 EU members with drought management legislation, including one law that has already been adopted as primary regulation by Parliament, and subsidiary legislation and secondary regulations adopted by the Government, as well as additional emergency regulations in place. The country also counts with basin-level procedures for water distribution and restrictions in drought scenarios, linked to River Basin Management Plans. 147. However, existing coordination mechanisms are complex and focus primarily on drought emergencies, without providing, for example, mechanisms and measures to prevent hydrological and pedological drought (e.g., planification and implementation of water retention solutions or planning the irrigations to preserve the soil moisture). Roles and responsibilities have mainly been defined to manage drought emergencies, not the proactive management of drought risk. In many cases, the mandates of the bodies responsible for drought management at different levels, from the National Committee for Mitigating Droughts and Combating Land Degradation and Desertification, to ANAR, INHGA, to River Basin Committees, the IGSU, to actors at local levels are unclear and, hence, not fully completed. While risk mitigation is part of the existing agenda, in practice, agencies still focus on reactive management practices to respond during drought events. There is no mechanism in place to ensure that identified precautionary and early approaches are being implemented. The severe 2022 drought provided the impetus for stakeholders to universally acknowledge the need to improve drought preparedness and shift towards proactive drought management. So far, the existing framework has not introduced such practices and will eventually require adjustments and further additions. 148. One instrument that could be adopted to successfully complete the transition from reactive to proactive drought management practices is the World Bank’s Drought Risk and Resilience Assessment (DRRA) approach. The DRRA builds on internationally recognized concepts, such as the “three pillars approach for drought resilience” promoted by the Integrated Drought Management Programme52 from the World Meteorological Organization, the Global Water Partnership and experiences of the World Bank. The approach brings together already-established tools, frameworks, and methodologies organized around the structure of four building blocks: (i) Characterization and Monitoring of droughts, (ii) Assessing impacts, vulnerabilities and risks, (iii) Characterization of current response and preparedness, and (iv) Prioritizing potential investments. Based on this report on drought risk assessment, the application of block (iii) and (iv) of the Bank’s DRRA would provide a full picture current drought resilience and help to develop a clear strategy to further strengthen the current framework and support its implementation for introducing a truly proactive drought risk management for all of Romania. 5.2 Recommendation 2: Develop drought risk management plans at river basin scale 149. Integrated Drought Risk Management Plans (IDRMPs) at the basin level, combined with existing River Basin Management Plans (RBMPs), are effective instruments to provide a structured framework for anticipating, mitigating, and responding to drought conditions. These IDRMPs encompass a comprehensive range of measures, from precautionary and operational to organizational and recovery measures. Water allocation and water use restrictions during drought events are an integral part of these plans. It is essential to ensure that, once crafted, these do not 51 https://edo.jrc.ec.europa.eu/edora 52 https://www.droughtmanagement.info/ 47 exacerbate structural water scarcity, as they should complement ordinary hydrological planning instead of being a vehicle for enacting contentious measures. Monitoring implementation, evaluating impacts of measures, and regularly updating these plans is essential. Linking IDRMPs to RBMPs and Flood Risk Management Plans (FRMPs) that are required to be updated every six years, would be beneficial and result in many additional synergies, for example in the collection and assessment of data or in stakeholder engagement. 150. Developing a pilot IDRMP in one or several river basins would be a good first step. This would consist of selecting river basins, based on drought impacts experienced, data availability, competencies, or any other factor considered. At the outset, it will be necessary to compile and validate an extensive set of water-related data, for example, surface and groundwater, or alternative resources such as wastewater reuse, and water demand by sector, including urban, agriculture, livestock, wetlands and environmental flows. Some of these are available in the RBMPs and in the already existing restriction plans. In addition to a thorough assessment of water supply and demand, a compilation of information on water system management rules (operational rules of reservoirs and infrastructure), and on restrictions like ecological flows to evaluate the possible constraints to resource management will be needed. As a next step, this information should be integrated into hydrological or “hydrosocial” models, such as Aquatool or other available decision support tools and models. These instruments allow different scenarios to be modelled, to assess the impact of changes in water demand or water availability and test different measures of proactive, precautionary, and reactive measures and their impact during drought years. 151. Water restrictions applied during drought events may lead to substantial damages for affected sectors and without adequate public awareness cause conflicts. Users must have a clear understanding of potential restrictions before their implementation. There is a good basis for this in many RBAs of Romania, as some binding restrictions on water use in drought events are currently defined according to the methodology established by the MEWF. Designing and implementing restrictions is a sensitive topic subject to controversy and conflicting interests. Developing them before an event when sensibilities are milder would benefit the process. Similarly, enforcement of restrictions should be supported by monitoring systems buoyed by comprehensive decision support tools and models. Making precautionary, organizational, monitoring and recovery measures at river basin level explicit will also contribute to validating the existing identification of competent authorities, stakeholders, and the clarification of their responsibilities at the national level. 152. Transparency and active stakeholder participation during the planning process will help raise awareness and build consensus. Engagement from the early stages of planning and preparation may contribute to better buy-in, better implementation of drought measures and may reduce conflicts in the event of droughts. Nevertheless, to be successful, public participation processes should be transparent about the scope of participation and include a feedback and/or grievance mechanism that addresses how the engagement was adequately included in the plan’s development. This process should be closely coordinated with the development of the RBMPs and FRMPs. 153. Insights gained from pilot IDRMPs can be used to develop a clear methodology for their wider application across Romania and benefit other countries within the Danube River Basin. The importance of international cooperation is noteworthy, given Romania’s river basins largely fall within the larger Danube River Basin covered by the International Danube River Protection Convention. This Convention is actively striving to enhance its approach to quantitative challenges and drought management. Consequently, the outcomes of pilot IDRMPs could be shared with the convention, fostering improved cooperation and communication among member countries. 48 5.3 Recommendation 3: Introduce drought risk mitigation and management in the water supply and sanitation sector 154. Droughts pose an increasing risk for domestic water supply in Romania, but local water utilities (ROCs or LOCs) lack the capacity to assess and manage drought. Few operating companies have written plans for water emergencies, and none of them are dedicated to drought. Drought management measures are mostly reactive, focusing on impact mitigation, and with little connection to the drought preparedness approaches devised at RBA level. Existing water monitoring activities by ROCs and LOCs do not effectively consider forecasting or updated assessments of water availability and impacts, potentially failing to capture and address the increasing frequency and severity of droughts. 155. Operators require support to assess and manage drought risks and develop and implement drought management plans. There is plenty of technical literature providing guidelines aiding utilities and operators in addressing drought risks, which could be adapted for application in Romanian cases (for example, there are several manuals developed in Spain53, and the USA54). These resources offer frameworks for targeted monitoring, planning, resource allocation, and in some cases advice on technological and infrastructure improvements, enabling utilities to better withstand and manage drought impacts on water supply. In 2023, ARA developed some guidelines for drought preparedness, reflecting a clear institutional response to the impact of the 2022 drought on ROCs. These guidelines must also be disseminated to LOCs. 156. The Danube Water Program's Utility Drought Risk Assessment initiative could support evaluations of the level of drought preparedness of Romania’s water utilities. This initiative aims to raise awareness about drought challenges and explore adaptation measures. It assesses annual average drought risk for participating utilities by analyzing customer bases, leakage losses, and water sources. This evaluation spans present conditions and three potential future climate change scenarios, ensuring a comprehensive understanding of current and evolving drought risks for effective adaptation planning. 5.4 Recommendation 4: Strengthen data sharing, collection and dissemination activities related to drought 157. Improve access to water data to unlock the full potential of drought monitoring and forecasting. The foundation of sound drought risk management lies in comprehensive and accessible data and drought risk information systems. Evaluating and refining the processes of data collection and information management is fundamental for enabling timely decision-making and strategic planning. Romania produces high-quality hydrometeorological information and has a well- advanced agrometeorological service, monitoring and forecasting (agricultural) droughts55. The INHGA is monitoring drought conditions, studies hydrological extremes and provides hydrological forecasts and warnings56. Based on these products, Romania is in a good position to develop further drought monitoring and forecasting products, adjusted for the general public and for (sector-) specific use groups and publish them through a drought observatory. EDO, the European Drought Observatory57 is a good example for these systems. 158. Comprehensive drought monitoring also includes the collection and analysis of information on drought impacts and damages. This information is not being systematically collected. It is important to continue and further advance data-driven methodologies for drought risk 53 https://www.miteco.gob.es/es/agua/temas/observatorio-nacional-de-la- sequia/guia_elaboraci%C3%B3n%20planes%20emergencia_tcm30-215447.pdf 54 https://www.epa.gov/waterutilityresponse/drought-response-and-recovery-water-utilities 55 https://www.meteoromania.ro/despre-noi/cercetare/agrometeorologie/ 56 https://www.hidro.ro/prognoze/ 57 https://edo.jrc.ec.europa.eu/edov2/php/index.php?id=1111 49 assessment in Romania. This involves further development of impact-chains for systems-at-risk within socio-economic and environmental domains. 159. Enhanced data transparency could significantly bolster integrated drought management in Romania. Access to data and willingness to share information, for data needed to assess drought hazards and data on drought damages is still very limited in Romania and was a major challenge for this assessment. Openly sharing this information through a broad system of public dissemination, perhaps through advertised online platforms, however, could significantly bolster integrated drought risk management in Romania. Better access to comprehensive and dynamic data regarding drought conditions, water availability, soil moisture, crop health, water usage and socio-economic indicators can revolutionize the approach to handling drought risk while increasing public awareness holds the potential to spark action across different societal levels, from individual households, to sectors, to international collaborations. At the grassroots, it prompts behavioral changes and community initiatives, while nationally it influences integrated and coordinated policies and drives investments in technology and research. Internationally, it could foster collaboration as well. 5.5 Recommendation 5: Earmark additional investments for drought risk management 160. Additional investments in water infrastructure will be needed to combat drought. With more frequent and more severe drought episodes due to climate change, additional investments in water infrastructure, such as storage expansion, advanced water conservation measures (including the ones needed to preserve the soil moisture), reducing non-revenue water and water losses, alternative water sources, and water reuse technologies will be needed. The level of investment needed to achieve water security has not been assessed and a more in-depth investigation, using data not currently available, such as groundwater data, would be necessary to fully understand the impacts of drought on water availability more broadly in Romania. 161. Financial tools are instrumental for managing drought risks effectively but are generally lacking in Romania. For instance, adjusting water pricing and tariffs during droughts encourages more responsible water usage, and both ANAR and ROCs/LOCs would benefit from a careful review and design of drought-specific tariffs, penalties and water conservation incentives. Higher prices during times of scarcity incentivize conservation, promoting efficient water use among consumers and industries. 162. Emergency funds are a vital resource during droughts. Risk transfer instruments at different scales, from the country to water users, and that rely on drought monitoring, would allow entities to transfer drought-related risks to financial markets, reducing their exposure to financial losses during dry spells. Governments or organizations could allocate specific funds or grants designed for rapid deployment during drought emergencies, including for example, social protections for farmers. These funds offer immediate relief and support to affected regions or communities, assisting in coping with the crisis. 163. It is important to explore options for donors or the government to support the funding of drought preparedness-related projects in different areas. In general, economic limitations hinder effective drought risk management, especially in rural areas. This is particularly evident in the water sector, but also the agricultural sector where low water billing prevails. ROCs and LOCs, and farms, will require external support for implementing drought risk management measures, such as investments in diversifying and connecting water sources, reducing non-revenue water and water losses, and enhancing integration within the distribution system. 50 References Berbel, J., Esteban, E. (2019). Droughts as a catalyst for water policy change. Analysis of Spain, Australia (MDB), and California. Global Environmental Change, 58, 101969. Blahušiaková, A., Matoušková, M., Jenicek, M., Ledvinka, O., Kliment, Z., Podolinská, J., Snopková, Z. (2020). Snow and climate trends and their impact on seasonal runoff and hydrological drought types in selected mountain catchments in Central Europe. Hydrological Sciences Journal, vol. 65, no. 12, pp. 2083–2096. Blauhut, V., Gudmundsson L., Stahl K. (2015). Towards pan-European drought risk maps: quantifying the link between drought indices and reported drought impacts. Environ. Res. Lett. 10(1): 014008. Broom, D. (2019). 5 droughts that changed human history. World Economic Forum. Available at: https://www.weforum.org/agenda/2019/05/5-droughts-that-changed-human-history/ Browder, G., Nunez Sanchez, A., Jongman, B., Engle, N., van Beek, E., Castera Errea, M., Hodgson, S. (2021). An EPIC Response: Innovative Governance for Flood and Drought Risk Management. World Bank, Washington, DC. Available at: https://openknowledge.worldbank.org/handle/10986/35754 Brus, D. J., Hengeveld, G. M., Walvoort, D. J. J., Goedhart, P. W., Heidema, A. H., Nabuurs, G. J., Gunia, K. (2012). Statistical mapping of tree species over Europe. European Journal of Forest Research, 131, 145-157. Burek, P., Satoh, Y., Kahil, T., Tang, T., Greve, P., Smilovic, M., Guillaumot, L., Zhao, F., Wada, Y. (2020). Development of the Community Water Model (CWatM v1. 04): A high-resolution hydrological model for global and regional assessment of integrated water resources management. Geoscientific Model Development, Volume 13, issue 7 GMD, 13, 3267–3298. Available at: https://doi.org/10.5194/gmd-13-3267-2020 Cammalleri, C., Micale F., Vogt, J. (2016). A novel soil moisture-based drought severity index (DSI) combining water deficit magnitude and frequency. HYDROLOGICAL PROCESSES 30 (2), p. 289- 301. JRC96439 Cammalleri, C., Arias-Muñoz, C., Barbosa, P., de Jager, A., Magni, D., Masante, D., Mazzeschi, M., McCormick, N., Naumann, G., Spinoni, J., Vogt, J. (2021). A revision of the Combined Drought Indicator (CDI) used in the European Drought Observatory (EDO). NHESS, 21, 481–495, 2021. Available at: https://doi.org/10.5194/nhess-21-481-2021 Carrão, H., Naumann, G., Barbosa, P. (2018). Global projections of drought hazard in a warming climate: a prime for disaster risk management. Clim Dyn 50, 2137–2155. Available at: https://doi.org/10.1007/s00382-017-3740-8 Center for International Earth Science Information Network - CIESIN - Columbia University. (2018). Gridded Population of the World, Version 4 (GPWv4): Population Count Adjusted to Match 2015. Revision of UN WPP Country Totals, Revision 11. Palisades, New York: NASA Socioeconomic Data and Applications Center (SEDAC). DOI: 10.7927/H4PN93PB Comité National Routier (CNR), Romanian National Union of Road Hauliers (UNTRR). (2020). Road freight transport in Romania in 2020. Available at: https://www.cnr.fr/download/file/publications/CNR%20- %20The%20Romanian%20road%20freight%20transport%20in%202020.pdf Copernicus Climate Change Service (C3S). (2022). ESOTC 2022 Overview: Drought. Available at: 51 https://climate.copernicus.eu/esotc/2022/drought Crângan, C. (2022). Marea foamete din 1917. Cum au ajuns românii să ucidă pentru hrană şi cât de adevărată este ipoteza canibalismului. Available at: https://adevarul.ro/stiri-locale/galati/marea-foamete-din-1917-cum-au-ajuns-romanii-sa- 1691682.html Delpla, I., Jung, A. V., Baures, E., Clement, M., Thomas, O. (2009). Impacts of climate change on surface water quality in relation to drinking water production. Environment International, 35(8), 1225- 1233. Ding, Y., Hayes, M. J., Widhalm, M. (2011). Measuring economic impacts of drought: a review and discussion. Disaster Prevention and Management: An International Journal, 20(4), 434-446. Directorate-General for Agriculture and Rural Development, European Commission (EC). (2024). Romania – CAP Strategic Plan. Available at: https://agriculture.ec.europa.eu/cap-my-country/cap-strategic- plans/romania_en#:~:text=Romania%20is%20one%20of%20the,sunflower%20seeds%2C%20 honey%20and%20plums.&text=The%20area%20in%20Romania%20used,13.5%20million%20 hectares%20(ha). Accessed: 05/07/2024. Directorate-General for Environment, European Commission (EC). (2024). Water scarcity and droughts: Preventing and mitigating water scarcity and droughts in the EU. Available at: https://environment.ec.europa.eu/topics/water/water-scarcity-and- droughts_en#:~:text=Droughts%20are%20a%20temporary%20decrease,and%20are%20consi dered%20natural%20phenomena. Accessed: 3/07/2024. Dumitrescu, R. (2022). “Romania sees sevenfold increase in vegetation fires in 2022.” Available at: https://www.romania-insider.com/romania-sevenfold-increase-vegetation-fires-2022. Accessed: 04/07/24. European Commission (EC). (2007). Communication from the Commission to the European Parliament and the Council - Addressing the challenge of water scarcity and droughts in the European Union. Brussels, 18.7.2007 COM (2007) 414 final. Available at: https://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2007:0414:FIN:EN:PDF European Commission (EC). (2012). Communication from the Commission to the European Parliament and the Council- Report on the Review of the European Water Scarcity and Droughts Policy. Brussels, 14.11.2012 COM (2012) 672 final. Available at: https://climate-adapt.eea.europa.eu/metadata/publications/report-on-the-review-of-the-european- water-scarcity-and-drought-policy/11309505 European Commission (EC). (2020). Communication from the Commission to the European Parliament and the Council - European Union Strategy for Danube Region. Brussels, 6.4.2020. SWD (2020) 59. Available at: https://danube-region.eu/wp-content/uploads/2020/04/EUSDR-ACTION-PLAN-SWD202059-final.pdf European Commission (EC). (2023). Water scarcity and droughts [Webpage]. Energy, climate change, environment. Retrieved from: https://environment.ec.europa.eu/topics/water/water-scarcity-and- droughts_en#:~:text=In%20some%20regions%2C%20the%20severity,a%20result%20of%20cl imate%20change. European Environment Agency (EEA). (2017). Climate change adaptation and disaster risk reduction in Europe. Enhancing coherence of the knowledge base, policies and practices. EEA Report no. 15/2017. Available at: 52 https://www.eea.europa.eu/publications/climate-change-adaptation-and-disaster Food and Agriculture Organization of the United Nations (FAO). (2012). Coping with water scarcity: An action framework for agriculture and food security. FAO Waters Reports 38. Available at: https://www.fao.org/4/i3015e/i3015e.pdf Food and Agriculture Organization of the United Nations (FAO), International Institute for Applied Systems Analysis (IIASA). (2021). Global Agro Ecological Zones version 4 (GAEZ v4). Retrieved from: http://www.fao.org/gaez/ Food and Agriculture Organization Statistics (FAOSTAT). (2023). Crops and livestock products trade database. Retrieved from: https://www.fao.org/faostat/en/#data/TCL Forzieri, G., Feyen, L., Rojas, R., Flörke, M., Wimmer, F., Bianchi, A., (2014). Ensemble projections of future streamflow droughts in Europe. Hydrology and Earth System Sciences 18(1), 85–108. DOI: 10.5194/hess-18-85-2014. Friedl, M., Sulla-Menashe, D. - Boston University and MODAPS SIPS - NASA. (2015). MCD12Q1 MODIS/Terra+Aqua Land Cover Type Yearly L3 Global 500m SIN Grid. NASA LP DAAC. DOI: 10.5067/MODIS/MCD12Q1.006 General Inspectorate for Emergency Situations (IGSU). (2016). Country report: 5.1 Conditionality. Romania. Available at: https://igsu.ro/Resources/COJ/RapoarteStudii/Raport_Final_de_tara%20pt%20Condit%20ex- ante%202016.pdf Global Water Partnership Central and Eastern Europe (GWP CEE), EU Strategy for the Danube Region, Priority Area 5 (EUSDR PA5 Environmental Risks). (2020). Revision of the Policy Instruments and their potential to contribute to EU droughts and water scarcity policies. Available at: https://www.gwp.org/globalassets/global/gwp-cee_files/idmp- cee/revision_of_the_policy_instruments_of_drought_2020.pdf Government Decision (GD) no. 923/2007 for the approval of the Program of measures for the development of the National Strategy for reducing the effects of drought in the short, medium and long term. Published in the Official Gazzette no. 565/2007. Available at: https://legislatie.just.ro/Public/DetaliiDocument/84536 Hayes, M.J., Svoboda, M. D., Wilhite, D.A., Vanyarkho, O. V. (1999) Monitoring the 1996 drought using the standardized precipitation index. Bull Am Meteorol Soc 80:429–438 Hengl, T., Gupta, S. (2019). Soil available water capacity in mm derived for 5 standard layers (0-10, 10- 30, 30-60, 60-100 and 100-200 cm) at 250 m resolution (v0.1) [Data set]. Zenodo. DOI: 10.5281/zenodo.2629149 Hidroelectrica Romania. (2016). Romania Investor Day Conference. New York. Hisdal, H., Tallaksen, L. M., Clausen, B., Peters, E., Gustard, A. (2004). Hydrological Drought Characteristics. Pages 139–198 of: Tallaksen, L. M., Lanen, H. A. J. (eds), Hydrological Drought: Processes and estimation methods for streamflow and groundwater. Development in Water Science, no. 48. Elsevier. Huning, L. S., Agha Kouchak, A. (2020). Global snow drought hot spots and characteristics. Proceedings of the National Academy of Sciences of the United States of America, 117(33), 19753-19759. Available at: https://doi.org/10.1073/pnas.1915921117 53 International Commission for the Protection of the Danube River (ICPDR). (2015). Danube River Basin District Management Plan 2015. Available at: https://www.icpdr.org/main/publications/maps-danube-river-basin-district-management-plan-2015 International Commission for the Protection of the Danube River (ICPDR). (2016). The 2015 Droughts in the Danube River Basin. Available at: https://www.icpdr.org/main/resources/2015-droughts-danube-river-basin International Commission for the Protection of the Danube River (ICPDR). (2018). Danube Climate Change Adaptation Strategy. Available at: https://www.icpdr.org/flowpaper/viewer/default/files/nodes/documents/icpdr_climate_change_ada ptation_strategy_web.pdf International Commission for the Protection of the Danube River (ICPDR). (2021). Danube River Basin Management Plan (DRBMP). Updates 2021. Available at: https://www.icpdr.org/flowpaper/viewer/default/files/nodes/documents/ic231_drbmp_update_202 1_draft_v10.pdf International Food Policy Research Institute. (2019). Global Spatially - Disaggregated Crop Production Statistics Data for 2010 Version 2.0 [Data set]. Harvard Dataverse, V4. DOI: 10.7910/DVN/PRFF8V Inundatii Romania. (No Date). River Basins [Website]. Retrieved from: https://inundatii.ro/en/river-basins/ Intergovernmental Panel on Climate Change (IPCC). (2021). Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., Zhai, P., Pirani, A., Connors, S. L., Péan, C., Berger, S., Caud, N., Chen, Y., Goldfarb, L., Gomis, M. I., Huang, M., Leitzell, K., Lonnoy, E., Matthews, J. B. R., Maycock, T. K., Waterfield, T., Yelekçi, O., Yu, R., Zhou, B. (eds.)]. Cambridge University Press. In Press. Joint Research Centre (JRC). (2015). European drought climatologies and trends based on a multi- indicator approach. Global and Planetary Change, Volume 127, 2015, Pages 50-57, ISSN 0921- 8181. Joint Research Centre (JRC). (2019). World Atlas of Desertification. Available at: https://wad.jrc.ec.europa.eu/primaryproduction Joint Research Centre (JRC). (2023). Forest Fires in Europe, Middle East and North Africa 2022. Available at: https://publications.jrc.ec.europa.eu/repository/handle/JRC135226 Kchouk, S., Melsen, L. A., Walker, D. W., van Oel, P. R. (2021). A review of drought indices: Predominance of drivers over impacts and the importance of local context. Nat. Hazards Earth Syst. Sci. Discuss, 1-28. Kummu, M., Taka, M., Guillaume, J. H. (2018). Gridded global datasets for gross domestic product and Human Development Index over 1990–2015. Scientific Data, 5(1), 1-15. Kurnik, B., Louwagie, G., Erhard, M., Ceglar, A., Kajfež, L. B. (2014). Analysing Seasonal Differences between a Soil Water Balance Model and in Situ Soil Moisture Measurements at Nine Locations Across Europe, Environmental Modeling & Assessment 19, 19–34. DOI: 10.1007/s10666-013- 9377-z. Kurnik, B., Kajfež-Bogataj, L., Horion, S. (2015). An assessment of actual evapotranspiration and soil water deficit in agricultural regions in Europe. International Journal of Climatology 35(9), 2451–2471. DOI: 10.1002/joc.4154. 54 Lange, S., Menz, C., Gleixner, S., Cucchi, M., Weedon, G. P., Amici, A., Bellouin, N., Schmied, H. M., Hersbach, H., Buontempo, C., Cagnazzo, C. (2021). WFDE5 over land merged with ERA5 over the ocean (W5E5 v2.0). ISIMIP Repository. McKee, T. B., Doesken, N. J., Kleist, J. (1993). The relationship of drought frequency and duration to time scales. In Proceedings of the 8th Conference on Applied Climatology (Vol. 17, No. 22, pp. 179-183). Ministry of Agriculture and Rural Development, National Agency for Fisheries and Aquaculture. (2021). National Strategic Multi-Annual Plan for Aquaculture (2021 – 2030). https://www.anpa.ro/wp- content/uploads/file/Planul%20Strategic%20National%20Multianual%20pentru%20Acvacult ura%202021%20-%202030.pdf Ministry of Agriculture and Rural Development, Order no. 97/2020, and Ministry of Internal Affairs, Order no. 63/2020, regarding the approval of the Regulation on the management of emergencies caused by dangerous meteorological phenomena resulting in the occurrence of pedological drought. Published on Official Gazzette no. 320/2020. Available at: https://legislatie.just.ro/Public/DetaliiDocumentAfis/224915 Ministry of Environment and Climate Change. (2013). National Strategy of Romania for Climate Change 2013 – 2020. Available at: http://mmediu.ro/app/webroot/uploads/files/Strategia-Nationala-pe-Schimbari-Climatice-2013- 2020.pdf Ministry of Waters and Forests, Order no. 459/2019, and Ministry of Internal Affairs, Order no. 78/2019, regarding the approval of the Regulation on emergency situations management generated by dangerous hydrometeorological phenomena having as effect the generation of floods, hydrological drought, as well as incidents / accidents at hydrotechnical constructions, accidental pollution of watercourses and marine pollution in the coastal area. Published on Official Gazzette no. 558/2019. Available at: https://legislatie.just.ro/Public/DetaliiDocument/215946 Modarres, R. (2007). Streamflow drought time series forecasting. Stochastic Environmental Research and Risk Assessment, 21:223–233. Myneni, R., Knyazikhin, Y., Park, T. - Boston University and MODAPS SIPS - NASA. (2015). MOD15A3H MODIS/Combined Terra+Aqua Leaf Area Index/FPAR Daily L4 Global 500m SIN Grid. NASA LP DAAC. DOI: 10.5067/MODIS/MOD15A3H.006 National Administration “Romanian Waters” (ANAR). (2022). River Basin Management Plans. Available at: https://rowater.ro/consultarea-publicului/directiva-cadru-apa/materiale-utile/legaturi-web-cu- proiectele-planurilor-de-management-actualizate-ale-bazinelor-spatiilor-hidrografice-2021/ National Integrated Drought Information System (NIDIS). (2024). What is Drought: Drought Basics. Available at: https://www.drought.gov/what-is-drought/drought-basics. Accessed: 3/07/2024. National Meteorological Administration (ANM): https://www.meteoromania.ro/clim/caracterizare- anuala/cc_2022.html Palomo-Hierro, S., Loch, A., Perez-Blanco, C. D. (2022). Improving water markets in Spain: Lesson- drawing from the Murray-Darling Basin in Australia. Agricultural Water Management, 259, 107224. Available at: https://doi.org/10.1016/j.agwat.2021.107224 55 Rossi, L., Wens, M., de Moel, H., Cotti, D., Sabino Siemons, A., Toreti, A., Maetens, W., Masante, D., van Loon, A., Hagenlocher, M., Rudari, R., Naumann, G., Meroni, M., Avanzi, F., Isabellon, M., Barbosa, P. (2023). European Drought Risk Atlas. Publications Office of the European Union, Luxembourg. Available at: https://publications.jrc.ec.europa.eu/repository/handle/JRC135215 Running, S., Zhao, M. (2019). MOD17A3HGF MODIS/Terra Net Primary Production Gap-Filled Yearly L4 Global 500 m SIN Grid V006 [Data set]. NASA EOSDIS Land Processes DAAC. DOI: 10.5067/MODIS/MOD17A3HGF.006 Slovenian Environment Agency. (2019). Better prepared for drought: Danube drought strategy. Ljubljana SPEEH Hidroelectrica S.A.: https://www.hidroelectrica.ro/article/20 Spinoni, J., Barbosa, P., Bucchignani, E., Cassano, J., Cavazos, T., Christensen, J. H., Christensen, O. B., Coppola, E., Evans, J., Geyer, B., Giorgi, F., Hadjinicolaou, P., Jacob, D., Katzfey, J., Koenigk, T., Laprise, R., Lennard, C. J., Kurnaz, M. L., Li, D., Llopart, M., McCormick, N., Naumann, G., Nikulin, G., Ozturk, T., Panitz, H., Porfirio da Rocha, R., Rockel, B., Solman, S. A., Syktus, J., Tangang, F., Teichmann, C., Vautard, R., Vogt, J. V., Winger, K., Zittis, G., Dosio, A. (2020). Future Global Meteorological Drought Hot Spots: A Study Based on CORDEX Data, Journal of Climate, 33(9), 3635-3661. Statista. (2024). Distribution of electricity generation in Romania in 2023, by source. Available at: https://www.statista.com/statistics/1236358/romania-distribution-of-electricity-production-by- source/ Statistical Office of the European Union (Eurostat). (2021). EUROPOP2019 data browser. Updates from March 2021. Available at: https://ec.europa.eu/eurostat/databrowser/view/tps00002/default/table?lang=en Stănilă, I. (2022). Marea foamete care a lovit România. Lipsa alimentelor i-a făcut pe oameni să mănânce leşuri de animale şi opinci. Available at: https://adevarul.ro/stiri-locale/calarasi/marea-foamete-care-a-lovit-romania-lipsa-2095938.html Sugg, M., Runkle, J., Leeper, R., Bagli, H., Golden, A., Handwerger, L. H., Magee, T., Moreno, C., Reed- Kelly, R., Taylor, M., Woolard, S. (2020). A scoping review of drought impacts on health and society in North America. Climatic Change, 162, 1177-1195. Technical and scientific support to the European Drought Observatory (EDO) for Resilience and Adaptation - Lot 2: In-depth assessment of drought management plans and a report on climate adaptation actions against drought in different sectors. ENV/2021/OP/0009. Member States Survey – Romania. Telesca, L., Lovallo, M., Lopez-Moreno, I., Vicente-Serrano, S. (2012). Investigation of scaling properties in monthly streamflow and Standardized Streamflow Index time series in the Ebro basin (Spain). Physica A: Statistical Mechanics and its Applications, 391(4):1662–1678. Thomas, D. S., Wilhelmi, O. V., Finnessey, T. N., Deheza, V. (2013). A comprehensive framework for tourism and recreation drought vulnerability reduction. Environmental Research Letters, vol. 8, no. 4, 044004. United Nations Framework Convention on Climate Change (UNFCCC). (2018). Glossary of Climate Change Acronyms and Terms. Available at: https://unfccc.int/process-and-meetings/the-convention/glossary-of-climate-change-acronyms-and- terms. United Nations Office for Disaster Risk Reduction. (2021). GAR Special Report on Drought 2021. Geneva. 56 United Nations. (2024). Global Issues – Water. Available at: https://www.un.org/en/global-issues/water United States Agency for International Development (USAID). (2014). Hydro-technical infrastructure in Romania. The SEPIC project task 1.4. Bucharest, Romania. Van Loon, A. (2013). On the propagation of drought. How climate and catchment characteristics influence hydrological drought development and recovery. PhD Thesis, Wageningen University, Wageningen, the Netherlands. Vicente-Serrano, S. M., Beguería, S., López-Moreno, J. I. (2010). A multiscalar drought index sensitive to global warming: the standardized precipitation evapotranspiration index. Journal of Climate, 23(7), 1696-1718. Vogt, J. V., Naumann, G., Masante, D., Spinoni, J., Cammalleri, C., Erian, W., Pischke, F., Pulwarty, R., Barbosa, P. (2018). Drought Risk Assessment and Management. A conceptual Framework. Publications Office of the European Union, Luxembourg. World Meteorological Organization (WMO), United Nations Educational, Scientific and Cultural Organization (UNESCO). (2012). International Glossary of Hydrology (WMO – No. 385). Geneva. World Bank (WB). (2023a). Systematic Country Diagnostic (SCD). Available at: https://documents1.worldbank.org/curated/en/099134003102323181/pdf/BOSIB0480d508207e080 5908b215a1d78b8.pdf World Bank (WB). (2023b). Romania Country Climate and Development Report (CCDR). Available at: https://openknowledge.worldbank.org/server/api/core/bitstreams/9693d7f2-e4b9-45c9-ac1c- ca28146e0248/content World Bank (WB). (2024). World Bank Data Bank. Agriculture, forestry, and fishing, value added (% of GDP), Agricultural land (% of land area), and Employment in agriculture (% of total employment) (modeled ILO estimate). Available at: https://data.worldbank.org/ World Bank (WB). (2024a). Drought Risk and Resilience Assessment Methodology. Enhancing the World Bank’s approach to managing droughts. World Bank (WB). (2023b). World Bank Data Bank. Population, Total Indicator. Available at: https://data.worldbank.org/indicator/SP.POP.TOTL World Meteorological Organization (WMO), Global Water Partnership (GWP). (2014). National Drought Management Policy Guidelines: A Template for Action (D.A. Wilhite). Integrated Drought Management Programme (IDMP) Tools and Guidelines Series 1. WMO, Geneva, Switzerland and GWP, Stockholm, Sweden. Available at: https://www.droughtmanagement.info/literature/IDMP_NDMPG_en.pdf World Meteorological Organization (WMO), Global Water Partnership (GWP). (2016). Handbook of Drought Indicators and Indices. Available at: https://www.droughtmanagement.info/literature/GWP_Handbook_of_Drought_Indicators_and_Indi ces_2016.pdf World Meteorological Organization (WMO), Global Water Partnership Central and Eastern Europe (GWP CEE). (2017). Integrated Drought Management in Central and Eastern Europe. Compendium of Good Practices. Available at: 57 https://reliefweb.int/sites/reliefweb.int/files/resources/idmp-cee_compendium_en.pdf Yildirim, G., Rahman, A., Singh, V. P. (2022). Meteorological and hydrological drought hazard, frequency and propagation analysis: A case study in southeast Australia. Journal of Hydrology: Regional Studies, Volume 44, 101229 https://doi.org/10.1016/j.ejrh.2022.101229 Zamfir, A. (2003). Centrala de la Cernavodă și-a întrerupt activitatea/Cernavodă Power Plant interrupted its activity. Ziua de Constanța, August 25, 2003. Available at: https://www.ziuaconstanta.ro/stiri/social/centrala-de-la-cernavoda-si-a-intrerupt-activitatea- 12072.html Zargar, A., Sadiq, R., Naser, B., Khan, F. I. (2011). A review of drought indices. Environmental Reviews, 19, 333-349. Zaveri, E. D., Damania, R., Engle, N. L. (2023). Droughts and Deficits - Summary Evidence of the Global Impact on Economic Growth. World Bank Group. Available at: http://documents.worldbank.org/curated/en/099640306142317412/IDU03b9849a60d86404b600bc 480bef6082a760a Zomer, R. J., Xu, J., Trabucco, A. (2022). Version 3 of the global aridity index and potential evapotranspiration database. Scientific Data, 9(1), 409. 58 Annex 1: List of local and regional operating companies (LOCs/ROCs) interviewed Regional Operating Companies (ROCs) Botosani - Nova Apaserv S.A. Botoșani Buzau - Compania de Apă S.A. Buzău Covasna - Gospodărie Comunală S.A. Sf. Gheorghe Dambovita - Compania de Apă Târgoviște-Dâmbovița S.A. Gorj - Aparegio Gorj S.A. Târgu-Jiu Hunedoara - Apa Prod Deva S.A. Iasi - ApaVital S.A. Iași Local Operating Companies (LOCs) SGCL Amara Servicii Edilitare pentru Comunitate Mioveni S.R.L. ECO-SCUP S.R.L. Serviciul de Alimentare cu Apa Paunesti UAT Puscasi Serviciul Public de Alimentare cu Apa si Canalizare (SPAAC) Rasova 59 Annex 2: Questionnaire for the qualitative interview with ROCs and LOCs Survey of the ROCs WB study on the Benefits of actions and the costs of inactions for drought risk management in Romania's water sector. Background: Over the past few years, Romania has experienced an increased frequency and severity in drought episodes. The purpose of this survey is to record qualitative evidence on how the ROCs were impacted by the drought events of recent years and their associated restrictions implemented by ANAR, to investigate which measures and changes to their operations the ROCs adopted as a result, and how to further increase their resilience to drought events. This information will help us to get a deeper understanding of the situation and will complement the analysis of the benchmarking data. Only with such information can we do a comprehensive impact analysis of the drought on the ROCs and help advance drought risk management for and by the ROCs. De facto, we garantee the anonymity of of all our respondents. This means that in the data analysis, your company will only appear under an identification number. In case we would like to use extracts from this present survey to illustrate our analysis with examples / case studies, would you be willing to have the name of your company used instead of the identification number? Should any of the questions below need further checks/clarifications from your side, additional information can also be sent to us after the online survey. Finally, do you agree for this online survey to be recorded - so taht we can go back to the video later on in case we did not capture all your feedback in the text? Name of operating company / ARA 1 member Address of the operating company / ARA 2 member Name and position of the contact 3 person(s) for this operator 4 Phone number of the contact respondent 5 Email address of the contact respondent 6 Date of the survey being completed 7 Place 60 Measures in place Answer of the operating company m.1 Has your company got a set of actions to be implemented in times of drought, i.e. drought coping measures? m.2 Beyond the water safety plans that exist for each water source: has your company got a risk management plan specifically targetting drought? m.2.1 If yes, since when? m.2.2 Are you aware of the drought risk management plan of ANAR and are you relating your own plan to the national one? m.2.3 Are you aware of the recommendations for drought mitigation developped by ARA? Have you already applied any of these recommendations? m.3 Has your company implemented drought risk mitigation actions? m.3.1 If yes, since when? m.4 Do you increase the monitoring of water flows / levels when experiencing a drought? m.4.1 If yes, how? m.4.2 If yes, when has this happened since 2010? (you can provide several periods) m.4.3 If yes, please describe what triggers (e.g. hydromet indicator values) the increase in monitoring. m.4.4 Is the monitoring specific to water availability, water demand, or both? m.4.5 How do you monitor availability? m.4.6 How do you forecast demand? m.5 In times of drought, do you use alternative water sources within your RBD? m.5.1 If yes, please list them. m.5.2 If yes, please describe shortly why you have used alternative water sources during the droughts. m.5.3 Do you believe ANAR should consult you on which alternative water sources in your RBD they should use during a drought and that are not used yet? m.6 In times of drought, do you use alternative water sources outside your RBD? m.6.1 If yes, please list them. m.7 Do you apply water conservation measures? If yes, do these include… m.7.1 Maintenance of the distribution network/fixing leaks m.7.2 Water re-use m.7.3 Low-flow technologies m.7.4 Awareness campaigns m.7.5 Others m.7.6 Are any of the conservation measures defined in a plan, document or agreement? m.7.7 If yes please list the measures and associated documents. m.8 Have you adjusted water prices in recent years? If yes to m.8: m.8.1 Do the adjustments aim to cover the increase in operational costs caused by the drought? m.8.2 Do the adjustments aim to change water demand and behavior? m.8.3 Other aims? Please list them. m.9 Are other financial instruments in place to counter the economic effects of the drought? m.9.1 If yes, please describe. m.10 In times of drought, have you intentionally restricted water supply to domestic users? m.10.1 If yes, when? m.11 Do you have plans in place to deliver emergency water supplies to the population? If yes to m.11, … m.11.1 Who is responsible to organise the emergency water deliveries? m.11.2 Who determines when emergency water delivery is needed? m.11.3 Who covers the costs of emergency water supply? m.11.4 Can you please describe the plans for emergency water supply? 61 Annex 3: EDORA methodology EDORA methodology One of European Union initiatives towards addressing the drought risk was the European Drought Observatory for Resilience and Adaptation (EDORA) project, which aim was to widen the scope of the European Drought Observatory (EDO) portal (which initially was focused on addressing the drought risk for agriculture) by assessing the drought risk in connection to multiple sectors and various spatial scales. A methodology to assess drought risks in connection to multiple sectors (also considering future scenarios) and various scales, and an Atlas of Drought Risk at the EU27 level were developed under this project. The EDORA methodology presents a comprehensive approach to drought risk assessment, focusing on the complex relationship between the drivers of drought risk, such as hazard, exposure, and vulnerability, and their connection to drought impacts on various natural and human systems (Rossi et al., 2023). The EDORA approach aims to close the gap of the limited understanding of this relationship by implementing a holistic method that enables both broader and more precise analysis of drought risks. Drought Hazard Assessment Drought hazard indices rely on hydro-meteorological data to quantify the severity, intensity, duration, and timing of different drought events. The EDORA data-driven approach applies multiple sectoral-relevant drought indices to avoid underestimating the compounding effect associated with assessing drought risk based on a single hazard indicator. These indicators include: • the SPI (SPI, McKee et al., 1993) and the SPEI (SPEI, Vicente-Serrano et al., 2010), for meteorological drought analysis. • the SSFI (SSFI, Modarres, 2007, Telesca et al., 2012) for hydrological drought conditions. • etc. For all these indicators, the extremes were expressed in terms of relative deviation from the median (drought hazard indices representing magnitude of water deficit / evaporative excess) and in terms of their standardized anomalies (drought hazard indices representing frequency of water deficit / evaporative excess). Finally, the severity, intensity, duration, and timing of different drought events are represented by the monthly, annual mean, and range of the indicator values. Data from multiple national and international sources were used to assess the drought hazard indices (Table 1). Hydro-meteorological data is derived from the global Community Water Model (CWatM; Burek et al., 2020). For the historical period 1990 –2019, meteorological forcing from ISIMIP3 repository (W5E5 dataset, Lange et al., 2021) was used. Further, a drought risk under climate change was calculated for the period 2021 – 2100 for four different Representative Concentration Pathways (RCP): 2.6, 4.5, 7.0, and 8.5, and using the following five Global Climate Models (GCM): GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI-RSM2-0, UKESM1-0LL. Hydro-meteorological inputs, extracting the total monthly precipitation and potential evapotranspiration for each river basin, average monthly soil moisture, and the minimum monthly streamflow associated with the largest river were processed. The Standardized Precipitation Index (SPI) The SPI is a widely used index to characterize meteorological drought. In 2009, WMO recommended SPI as the main meteorological drought index that countries should use to monitor and follow drought conditions58. Among the advantages of using this index are: simplicity in calculation and usage, 58 WMO, GWP 2016 62 applicability to all climate regimes, variable time scale, which is helpful for the analysis of drought dynamics, especially the determination of onset and cessation59. It shows the anomalies (deviations from the mean) of the observed total precipitation for a period from the long-term average, for that period for any given location. Negative SPI values indicate conditions of less precipitation (more drought), while positive values suggest wetter conditions. The magnitude of the value reflects the severity of the condition, with more negative values indicating more severe drought and vice versa. For the Romania drought risk assessment, SPI of different scales were calculated for the years 1951 to 2022. For example, SPI12 represents the SPI calculated over a 12-month timescale, so it is a standardized drought index that measures how anomalous the accumulated precipitation is over a 12- month period compared with the long-term average for a given location. SPI48 does the same thing over a 48-month interval, measuring anomalies at each timestep over a four-year period. Each SPI timescale captures different aspects of the hydrological cycle. For example, SPI48 captures longer-term drought conditions, which can potentially be more related to depleted groundwater reserves, reduced reservoir storage and diminished snowpack than SPI12, which often connects better to more immediate impacts like soil moisture deficits, crop stress, and reduced streamflow. Comparing the scales can provide a more complete picture of drought conditions over different timescales, in this case medium-term and long-term drought respectively. The Standardized Precipitation-Evapotranspiration Index (SPEI) The SPEI is an extension of the SPI and includes a temperature component, allowing the index to account for the effect of temperature on drought development. SPEI is usually used to identify and monitor conditions associated with a variety of drought impacts60. High temperatures increase evapotranspiration rates, which may exacerbate drought conditions even if precipitation levels remain the same, and this effect is captured by the SPEI. By comparing monthly precipitation and monthly potential evapotranspiration, a simple monthly climatic water balance is computed. The calculated values are then aggregated at different time scales following the same procedure as the SPI. For this study, SPEIs up to scale 48 were calculated for the years 1951 to 2022. Similar to the SPI, the timescales capture impacts on different aspects of the water cycle. The Standardized Streamflow Index (SSFI) The SSFI was used to assess the hydrological drought condition at multiple timescales. The SSFI is calculated using monthly streamflow values and the methods of normalization associated with SPI61. It was calculated for the years 1990 to 2022. Discharge data was taken from Eurostat for the period from 1990 to 2020 and complemented with more recent data on yearly water volumes from ANAR. Drought risk assessment EDORA project proposed a complex methodology for drought risk assessment, by integrating the factor approach (focused on risk drivers) and the outcome approach (focused on the socio-economic and environmental factors influence on the negative impacts registered) to increase the accuracy of the drought risk calculation. 59 Hayes et al., 1999 60 WMO, GWP 2016 61 WMO, GWP 2016 63 To assess drought risks for the different sectors and systems using EDORA methodology, the following steps were taken: 1. Identification of relevant proxies for drought impacts For each system at risk, a proxy that accurately represents the "drought impact" was determined. This serves as a reference point for understanding how drought affects specific systems or regions. 2. Development of system-specific conceptual models This phase involved the creation of conceptual models for drought risk that were tailored to various systems. For this step, qualitative impact chains, linking different aspects of drought and how they influence particular areas and systems were obtained from the results of the EDORA application at EU level (Rossi et al., 2023). Researchers used scientific and grey literature with further validation from interviews with experts. 3. Quantitative analysis of hazard, exposure and vulnerability This part consisted of three sub-steps: • Calculation of drought hazard indices (SPI, SPEI, SSFI, etc.) using hydro-meteorological data. • Combination of key system-specific vulnerability indicators to gauge susceptibility to drought. To tackle data gaps, vulnerability characterization combined general and sector- specific proxies, grouping regions with similar hazard-impact responses. This assumed similar likelihoods of negative direct impacts. Relevant indicators, derived from impact chains, were selected based on vulnerability drivers identified by Rossi et al., 2023. • Spatial representation of exposed assets and production. This involved mapping the physical areas and systems that are exposed to drought risks. Spatial datasets of assets and production (e.g., average wheat yield, average hydropower production) were used to estimate drought exposure. For the data-driven analysis, direct exposure data consisted of a detailed spatial representation of the different (sub-)systems and system types present in Romania. 4. Calibration of the models The collected and processed data from the previous steps were used to train the model via machine learning techniques, that were calibrated to understand the complex relationships between hazards, vulnerabilities, exposures, and impacts. 5. Calculation of the current and future risk metrics Finally, the calibrated models were used to calculate the risk metrics. These metrics approximate the sectoral drought risks, providing valuable insights into how drought might affect various parts of the economy and environment now and in the future. Data inputs Complex data sets were used for drought risk calculation, having various sources, spatial and time scales (see Table 4). Drought hazard: The three drought indices (SPI, SPEI, and SSFI) were used, along with additional indicators (e.g., soil-moisture indicators). Climate change models were used to calculate future drought indices. Exposure: Mapping the areas or systems exposed to drought risk required spatial dataset of assets and production. Often, exposure indicators are directly linked to the impact data used in the data-driven approach. For systems for which impact data was derived from continuous indicators, the average value of the last five years was used to establish the presence and value of the exposure. For other 64 systems, specific exposure maps (e.g., related to location of hydropower dams, thermal power plants, specific ecosystems, etc.) as identified in the impact chains, were considered. For Romania, exposure of wetland and forest systems to drought is represented by the fraction of wetland/forest in each river basin (Table 4). Vulnerability: Vulnerability indicators were grouped together to form vulnerability classes (or ‘clusters’). For each a common hazard-impact relationship was assumed. Following the EDORA system-specific impact chains, open-source datasets containing the vulnerability factors were used. Multiple indicator combinations were tested for different classes, each running an individual model. The best-performing class, determined by average balanced accuracy across regions, was chosen for the final vulnerability map. This tailored method offers a detailed representation of vulnerability in drought risk assessment. Impact: System-specific drought impact (proxy) indicators were identified and developed. A constraint on the choice of the "drought impact” was the availability of impact data, because consistent, continuous and quality assured data on drought-related impacts covering all the study area addressed is challenging. However, the collaboration of relevant local stakeholders allowed the completion of information contained in the global data sets and a more accurate drought risk calculation. Figure 54: Simplified representation of the water supply impact chain Source: WB water team’s analysis, adapted after Rossi et. al, 2023 The parameters used for Exposure, Vulnerability and Impact for this study are shown in Table 5. For the identified systems-at-risk, impact magnitude categories were determined using negative anomalies in continuous impact datasets. Some time series were detrended to account for changes like technological advancements, which can be done using linear regression or moving window algorithms such as the locally weighted scatterplot smoothing (LOWESS) algorithm. Anomalies were then calculated as shares below or above the expected trended values, and by using various thresholds to form discrete and relative impact categories. An important aspect is to separate the losses caused only by droughts from the ones caused by other natural or human-made shocks, like storms, floods, or economic recessions. Extreme losses may result from compound shocks like heat and drought, drought and pests, or poor water management. 65 This issue is addressed by teaching the machine learning algorithms to recognize and minimize the overestimation of drought's influence. Consequently, results that slightly underestimate the drought impact were looked for, recognizing that the ML model does not account for several other influential factors in its predictions. A decision tree algorithm, classifier type, was used per cluster and impact category to identify the hazard conditions (a specific set of drought indices and relevant thresholds) likely to create impacts of a selected severity, thus a prioritization in the the identification of the most influential predictors was done and therefore predictor thresholds to predict drought impact were employed. The optimization was done based on both precision and balanced accuracy, with the best result picked after manual screening. A forest of decision trees was used to overcome the limitation of identifying a single specific set of hydrometeorological conditions that leads to impact. The impact time-series data consisted of subnational data for at least five years resampled to river basin scale. One exception is the inland water transport sector, for which input data was only available on a national scale. Most input data were provided at NUTS3 level and were resampled to river basin scale based on the proportion of different NUTS3 in each river basin, as described in Equation 1. Equation 1: Rescaling NUTS3 impact data to river basin scale. ∩ = ∑ × ∈ ∩ Note: xr and xi are the rescaled impact data points, and is the share of the area of a given NUTS i in each basin r. Linking drought hazards and impacts Hazard and Impacts were linked using a statistical approach. This step was necessary to understand the system's sensitivity to drought extremes, using an event-based and sector-specific approach. By employing machine learning algorithms with time series data of drought impact and hazard, the hydro- meteorological conditions leading to an impact were identified. This included the analysis of local sectoral direct vulnerability, with more resilient systems reacting differently to drought hazards and severities compared to more sensitive ones. Decision trees and decision forests were used in this assessment to optimize the hazard-impact link for each sector and vulnerability class. This method assumes similar hazard-impact links for regions with similar vulnerabilities and can also estimate the likelihood of impact in data-scarce areas. Through decision tree algorithms, the stressor-response link is established for each system-at-risk, categorizing different impact severity classes, and identifying key hazard indicators. This machine learning approach enhances the modelling of historical drought impacts, learning from the hydro-meteorological drivers usually linked to observed impacts and objectively identifying the best predictors of risk. The standard metric to optimize these decision trees is the accuracy score (Equation 2). The accuracy score describes the capacity of the model to truthfully predict the occurrence or absence (miss) of an impact. However, given the presence of non-drought factors, the occurrence of misses was considered less significant. In addition to accuracy score, the decision trees were also optimized using precision score metrics (Equation 3), with the goal of minimizing false alarms, that is, simulated reductions in production level that were not actually observed. Equation 2: Calculation of the balanced accuracy metric   ( )  +    (ℎ)   =      Equation 3: Calculation of the precision metric  (ℎ)   =    (ℎ)  +    ( ) Decision trees, unlike regression analysis, do not use all predictors but instead seek the best predictors to forecast drought impact. They divide the dataset into categories like "will lead to impact" 66 and "will lead to no impact", allowing for a more nuanced analysis that considers non-linear relationships and various hydro-meteorological conditions. This method surpasses the common composite-indicator approach, providing a more flexible system or region-specific response. While decision trees identify one specific set of conditions leading to impact, the use of decision forests (multiple trees) allows for predicting various combinations of conditions that may lead to impact, encompassing everything from flash droughts to multiyear events. The technique was applied for each impact severity category, creating binary series of different magnitudes of relative impact (e.g., 5% below normal, 10% below normal). Separate trees were made for each category, and relevant hazard indicators may vary between them. In this data-driven approach, all hazard indicators were considered, contemplating spatial differences in exposure and vulnerability. Consequently, the likelihood of impactful drought can be determined for various systems, vulnerability clusters, and magnitudes of impact, ensuring that no potential links between hazard and impact are overlooked. Decision tree models across the different vulnerability classes established were used to select the most reliable model for estimating and predicting drought impacts. These models aim for balanced accuracy or precision within each class and impact level. They were rigorously validated against independent observed data and prevent overfitting through the Stratified Repeated KFold algorithm. Analyzing the ML models for the Romania’s drought risk calculation, better performance was found using the balanced accuracy metric, crucial in imbalanced scenarios. This metric, averaging Sensitivity (the "True positive rate") and Specificity (the "True negative rate"), gauges the model's accuracy in predicting presence or absence of drought impact. The focus was on balanced accuracy, aiming for a score close to 1 while guarding against overestimation of the drought impact, acknowledging the model's limitations in considering all influential factors. In general, the scores of the precision and balanced accuracy metrics were excellent62: • Agriculture: for maize, wheat, sunflower seeds, potatoes, and the perennial fodder: precision ≥ 0.98 and BACC ≥ 0.9. In case of other crops, the precision score is higher than 0.95 and the BACC score is higher than 0.85. The ‘other edible roots’ category had the lowest performance (precision = 0.95, BACC = 0.5). At the group level, cereals, oil crops, and fodder crops have relatively high performance, and the fruit and vegetables group have a mixed higher and lower performance. Some models slightly underestimate the observed impacts, which may be explained by impacts of non-drought drivers (e.g., mismanagement, heat waves etc.). • Energy production: the model performance is high as the precision nearly reaches 100% and the balanced accuracies is approximately at 95%. The model predicts energy production well, with slight underestimates for most river basins, which may be explained by impacts of non-drought drivers. • Fluvial Transport: the model performance is high as both the precision and balanced accuracy almost reaches 100% (> 0.98). The model runs nationally and slightly underestimates the impact on navigation, which may be explained by the involvement of non-drought drivers. • Natural Ecosystems: the model performance is high for both ecosystems, as the precision and balanced accuracies are over 95%. The models predict impacts on ecosystems well, with slight underestimates for most river basins, which may be explained by impacts of non-drought drivers. • Water supply: the model performance is high as the precision nearly reaches 100% and the balanced accuracies is higher than 95%. The model predicts water abstraction quite 62 In general, a score for precision or BACC > 0.85 is an excellent score, > 0.7 is a good one, and any other score can be considered as poor. 67 well, with slight under/over-estimates for specific river basins, which may be explained by impacts of non-drought drivers. Drought risk estimation The likelihood of impactful drought events in specific regions can be calculated using decision trees for specific systems, impact categories, and vulnerability classes. By using continuous impact data (e.g., time series of crop yields) the model determined the likelihood of impactful drought for each reduction category (e.g., reduction in crop yields over time). As a result, exceedance probability curves were created, which relate specific impacts to return periods, and allow the relative average annual loss (AAL) (or impact) due to drought conditions to be calculated. For this study, the relative average annual loss (AAL) in current and future conditions for different systems-at-risk is provided. Being a data-driven process, the accuracy of the estimated risk depends strongly on the quality of the input information available. More information and details on the modelling exercise can be found in the “Romania Drought Impact and Risk Assessment – Deep dive report” prepared by the International Institute for Applied Systems Analysis and Marthe Wens for this drought risk assessment. 68 Table 4: Summary table of the datasets used in Romania’s drought risk assessment Component Spatial Data item Source Temporal resolution Temporal coverage Comments resolution 1990 - 2019 Precipitation Global CWatM, ISIMIP 3a & 3b 0.5 deg Daily/monthly 2021 - 2100 Based on the FAO 1990 - 2019 Potential Evapotranspiration Global CWatM, ISIMIP 3a & 3b 0.5 deg Daily/monthly Penmann- 2021 - 2100 Monteith equation Hazard 1990 - 2019 Soil moisture Global CWatM, ISIMIP 3a & 3b 0.5 deg Daily/monthly 2021 - 2100 Monthly minimum 1990 - 2019 Streamflow Global CWatM, ISIMIP 3a & 3b 0.5 deg Daily/monthly streamflow for the 2021 - 2100 largest river National Institute of Statistics 18 crops and crop Crop yields NUTS3 Annual 1990 - 2021 TEMPO Online statistics Romania groups Water abstraction for public EUROSTAT River basin Annual 2008 - 2017 water supply Water use for hydropower Impact ANAR Romania River basin Annual 2009 - 2020 generation Transport in vessels for National Institute of Statistics National Annual 2005 - 2021 Romania TEMPO Online statistics Romania Net primary productivity MODIS/Terra NPP Gap-Filled 500 meters Annual 2001 - 2022 (NPP) Yearly Global Used landcover classification by MODIS/Terra+Aqua Land cover the IGBP, five Exposure Forest/Wetland map 500 meters Annual 2015 type yearly L3 Global forest landcovers, and one wetland covers Vulnerability Dominant tree species Brus et al., 2012 0.0174 deg - Multi annual 69 Component Spatial Data item Source Temporal resolution Temporal coverage Comments resolution EEA Digital map of ecological Ecological zones - - Multi annual regions GDP Kummu et al., 2020 0.008333 deg Annual 2015 Annual average Aridity index Zomer and Trabucco, 2022 0.008333 deg 1970 - 2000 Average of Government effectiveness EQI European Commission NUTS3 Annual 2021 indicators Agro-ecological zones FAO-IIASA GAEZ v.4 0.008333 deg Annual average 1980 - 2010 Data on these aspects is also available in Share of irrigation relative to IIASA/IFPRI SPAM crop 0.08333 deg Annual 2010 National Institute arable land distribution model. of Statistics TEMPO Online statistics Romania Available water capacity of Hengl and Gupta, 2019 250 meters Annual average 1950 - 2017 the soil Soil compaction as Soil compaction ISRIC GLASOD - - ~1990 major land degradation level 70 Table 5: Summary table of parameters for Exposure, Vulnerability and Impact Sector Exposure Vulnerability Impact Indicators: • Agro-Ecological Zones (AEZ), • Soil compactness, • Soil water retention capacity, • Crop yield levels, • Reduction of annual yield of 18 Agriculture • 18 crops • Presence of irrigation systems different crops between 1990 –2019 Classes: • Intersection of all indicators except AEZ; • Intersection of all indicators. Indicators: • Aridity index, • GDP per capita • Reduction in water supply for hydroelectric power generation Energy production • Hydropower plants Classes: during or after drought events • Intersection of Aridity index between 2009 – 2020 and GDP per capita; • GDP per capita Indicator: • SPEI 12 (discharge data was • Reduction of goods transported in Fluvial transport • Freight transport not available) vessels for Romania during or after Classes: drought events between 2005 - 2019 • SPEI 12 Forests: • Reduction of annual global Net Terrestrial and aquatic ecosystems • Forests and wetlands Indicators: Primary Production (NPP) between 2001 - 2022 • Dominant tree species, 71 Sector Exposure Vulnerability Impact • Ecological region Classes: • Intersection of dominant tree species and ecological region; • Dominant tree species Wetlands: Indicators: • Agro-Ecological Zones (AEZ), • WEI+ water stress indicators, • Ecological regions Classes: • Intersection of the ecological regions and the WEI+ indicator; • Intersection between all three indicators Indicators: • Aridity index, • GDP per capita, • Reduction of water abstraction Water Supply • Water supply • Quality of Governance index volumes for water supply between Classes: 2008 -2020 • Intersection of the aridity index and GDP per capita; • Intersection of all indicators 72