Present and Future Climate Risk Across Bangladesh: Integrated findings on hazards, exposure, and poverty-driven vulnerability September 28, 2022 Mattia Amadio Lander Bosch Sailesh Tiwari Ayago Wambile 1 Table of Contents List of Maps ........................................................................................................4 List of Figures ....................................................................................................5 List of Tables ......................................................................................................5 Acronyms and Abbreviations .............................................................................6 Acknowledgments ..............................................................................................7 Executive Summary ...........................................................................................8 1. Introduction ............................................................................................... 16 2. Spatial disparities in hazards and poverty rates, and their link to climate vulnerability ...................................................................................................... 16 3. Defining present-day disaster risk: definitions, data, and methodology. 19 4. Conceptualizing hazards and the current setting in Bangladesh ............ 21 a. Floods and storm surges ...............................................................................................................21 b. Heatwaves .......................................................................................................................................22 c. Droughts .........................................................................................................................................23 d. Landslides .......................................................................................................................................24 e. Tropical cyclones ...........................................................................................................................24 f. Air pollution ...................................................................................................................................25 5. Conceptualizing exposure and the current setting in Bangladesh .......... 26 a. Population .......................................................................................................................................26 b. Built-up assets ................................................................................................................................27 c. Agricultural land.............................................................................................................................28 6. Results: mapping of present-day hazards, exposure, and vulnerability .. 29 a. Riverine and coastal flood risk .....................................................................................................30 b. Heatwave risk .................................................................................................................................37 c. Drought risk ...................................................................................................................................40 d. Landslide risk..................................................................................................................................42 e. Tropical cyclone risk .....................................................................................................................45 f. Air pollution risk ............................................................................................................................47 7. The compounding effect of risk................................................................ 50 8. A look ahead: changing risks in the face of climate change .................... 53 2 a. Climate projections: the why and how .......................................................................................53 b. Modeled variables: projecting changes in precipitation and temperatures ............................54 c. Flooding and landslide projections .............................................................................................54 d. Drought projections ......................................................................................................................58 e. Heat projections .............................................................................................................................60 9. Compounding challenges: the multiplicity of mid-century climate risks 60 10. References .............................................................................................. 61 3 List of Maps Map 2.1: Poverty Rates at the Upazila Level, 2010 ..................................................................................................... 18 Map 2.2: Poverty Rates at the Upazila Level, 2016 ..................................................................................................... 18 Map 2.3: Change in Poverty Rates between 2010 and 2016 in Bangladesh’s Upazilas ......................................... 19 Map 5.1: WorldPop Bangladesh Population Model, 2020 ......................................................................................... 27 Map 5.2: WSF 2019 Distribution of Built-Up Areas across Bangladesh ................................................................. 28 Map 5.3: Bangladesh Land Use, 2020 ............................................................................................................................ 29 Map 6.1: Riverine (panel a) and Coastal (panel b) Flood Hazard at a 90m Resolution across Bangladesh for a 100-Year Return Period ................................................................................................................................................... 30 Map 6.2: Expected Annual Population Impact of Riverine (panel a) and Coastal (panel b) Floods – Mortality and Morbidity .................................................................................................................................................................... 31 Map 6.3: Expected Annual Damage to Built-Up Assets of Riverine (panel a, left) and Coastal (panel b, right) Floods ................................................................................................................................................................................. 32 Map 6.4: Expected Annual Exposure of Agricultural Land to Riverine (panel a, left) and Coastal (panel b, right) Floods ...................................................................................................................................................................... 32 Map 6.5: Expected Annual Impact of Riverine Floods on Population in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................................... 34 Map 6.6: Expected Annual Impact of Riverine Floods on Built-Up Assets in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................... 34 Map 6.7: Expected Annual Exposure of Agricultural Land to Riverine Floods in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................... 35 Map 6.8: Expected Annual Impact of Coastal Floods on Population in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................................... 36 Map 6.9: Expected Annual Impact of Coastal Floods on Built-Up Assets in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ............................................................................................................................... 36 Map 6.10: Expected Annual Exposure of Agricultural Land to Coastal Floods in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................... 37 Map 6.11: Heat Stress across Bangladesh for a Heat Event with a 20-Year Return Period ................................. 38 Map 6.12: Expected Annual Population Exposure to High Heat Stress for Bangladeshi Upazilas .................... 38 Map 6.13: Expected Annual Exposure of the Population to Heat Stress in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ............................................................................................................................... 39 Map 6.14: Drought Hazard for Agricultural Land across Bangladesh during the First (panel a) and Second (panel b) Cropping Seasons............................................................................................................................................. 40 Map 6.15: Kharif (panel a) and Rabi (panel b) Season Drought Exposure for Agricultural Land (by Zila and Upazila) ............................................................................................................................................................................... 41 Map 6.16: Expected Annual Exposure of Agricultural Land to Drought in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ............................................................................................................................... 42 Map 6.17: Landslide Hazard Index for Bangladesh at a 1km Resolution ................................................................ 43 Map 6.18: Population Exposure to High Landslide Risk at the Upazila Level....................................................... 43 Map 6.19: Built-Up Asset Exposure to High Landslide Risk .................................................................................... 44 Map 6.20: Expected Annual Exposure of the Population to Rainfall-Induced Landslides in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs .................................................................................................. 44 Map 6.21: Expected Annual Exposure of Built-Up Assets to Rainfall-Induced Landslides in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs .................................................................................................. 45 Map 6.22: Wind Speeds Reached across Bangladesh during Tropical Cyclones with a 100-Year Return Period .............................................................................................................................................................................................. 46 Map 6.23: Expected Annual Impact of Tropical Cyclones across Bangladesh’s Upazilas.................................... 46 4 Map 6.24: Expected Annual Exposure of Built-Up Assets to Tropical Cyclones in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................... 47 Map 6.25: Mean Air Pollution (PM2.5 concentration) across Bangladesh, 1998-2019 ........................................... 48 Map 6.26: Increased Mortality from PM2.5 Air Pollution, 1998-2019 ...................................................................... 49 Map 6.27: Expected Annual Impact of Air Pollution on the Population in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs ................................................................................................................................... 50 Map 7.1: Co-Occurring Hazard Exposure for Bangladesh’s Upazilas ..................................................................... 51 Map 8.1: Climate Change Projections – Rainfall over 10mm (days/year) for Bangladesh ................................... 55 Map 8.2: Climate Change Projections – Consecutive Wet Days (days/year) for Bangladesh .............................. 56 Map 8.3: Climate Change Projections – Maximum 5-Day Precipitation (mm) for Bangladesh .......................... 56 Map 8.4: Climate Change Projections – Extremely Wet Day Precipitation (mm) for Bangladesh ..................... 57 Map 8.5: Climate Change Projections – Average Sea Level Rise (cm) for Bangladesh ......................................... 57 Map 8.6: Climate Change Projections – Consecutive Dry Days (days/year) for Bangladesh .............................. 59 Map 8.7: Climate Change Projections – Standardized Precipitation-Evapotranspiration Index for Bangladesh .............................................................................................................................................................................................. 59 Map 8.8: Climate Change Projections – WBGT Heat Index (°C) for Bangladesh ................................................ 60 List of Figures Figure 3.1: Computation of Expected Annual Impact of Natural Hazards in Geospatial Analytics .................. 20 Figure 7.1: 2010 and 2016 Poverty Rates and Co-Occurring Hazards .................................................................... 52 Figure 7.2: Change in Poverty Rates and Co-Occurring Natural Hazards .............................................................. 52 List of Tables Table 3.1: Hazard, Exposure, and Impact – Upazila-Level Analysis ....................................................................... 21 Table 6.1: Hazard, Exposure, and Vulnerability Categories included for the Current Climate Analytics in Bangladesh ......................................................................................................................................................................... 29 Table 8.1: Climate Variables Underlying Climate Projections ................................................................................... 54 5 Acronyms and Abbreviations Acronym Definition AR Assessment Report (IPCC) BBS Bangladesh Bureau of Statistics CCDR Country Climate and Development Report CMIP Coupled Model Intercomparison Projects EAE Expected Annual Exposure EAI Expected Annual Impact FAO Food and Agriculture Organization of the United Nations GDP Gross Domestic Product HIES Household Income and Expenditure Survey IMF International Monetary Fund IPCC Intergovernmental Panel on Climate Change MISR Multi-angle Imaging Spectroradiometer MODIS Moderate Resolution Imaging Spectroradiometer NASA National Aeronautics and Space Administration NOAA National Oceanic and Atmospheric Administration PM Particulate Matter RCP Representative Concentration Pathway RP Return Period RPi Impact for an event with a specific RP SeaWiFS Sea-viewing Wide Field-of-view Sensor SPEI Standardized Precipitation-Evapotranspiration Index SSP Shared Socioeconomic Pathway UN United Nations WBG World Bank Group WBGT Wet Bulb Globe Temperature WHO World Health Organization WSF World Settlement Footprint 6 Acknowledgments This report has been prepared by Mattia Amadio, Lander Bosch, Sailesh Tiwari, and Ayago Wambile from the Poverty and Equity Global Practice in the Equitable Growth, Finance and Institutions Vice Presidency of the World Bank. It is part of the background work prepared for the Country Climate and Development Report (CCDR) in Bangladesh. We gratefully acknowledge the helpful input and support received from Ira Irina Dorband, and Arthur Hrast Essenfelder. Any remaining errors are our own. This work benefited from the general direction of CCDR Co-TTLs Gayatri Acharya, Swarna Kazi, Bernard Haven, Volker Treichel, and Persephone Economou. The team is also grateful for the overall guidance received from Andrew Dabalen and Yutaka Yoshino and acknowledges the advice from peer reviewers through the CCDR peer review process. The findings and interpretations in this report do not necessarily reflect the views of the World Bank, its affiliated institutions, or its Executive Directors. 7 Present and Future Climate Risk Across Bangladesh Integrated findings on hazards, exposure, and poverty-driven vulnerability Executive Summary Overview Bangladesh is among the countries most affected by extreme weather events globally. The long-term Climate Risk Index (Eckstein, Künzel, and Schäfer 2021) ranks Bangladesh as the seventh most at- risk country over the 2000-2019 period. No other country is more vulnerable to flooding, with heatwaves, droughts, tropical cyclones, and more locally, landslides, posing compounding climate- related threats (ADB 2021). Average annual losses from disasters are estimated to be about US$3 billion, close to 2 percent of Bangladesh’s gross domestic product. While these mainly affect the agricultural sector, with wide-ranging effects on the rural economy, natural hazards increasingly impact the country’s rapidly expanding urban areas, where the service sector is predominant (WBG 2021a). Even under optimistic global climate scenarios, Bangladesh faces a strong increase in the frequency and severity of extreme weather events and long-term climatic changes throughout the twenty-first century, compounded by intensified environmental degradation and urbanization. A detailed study of Bangladesh’s current and future climate risks is therefore of paramount importance to devise tailored policies and targeted investments. Natural hazards do not occur uniformly across space, nor is their impact equally distributed. The frequency and intensity of hazards, and the vulnerability of exposed communities and assets to these hazards will determine how severely a specific location is struck. Poorer households might be disproportionally affected by extreme events because they tend to reside in more disadvantaged and hazard-prone areas, have less mobility and poorer access to critical services and early warning systems, and their assets are less well-insured against extreme climate events (Dasgupta 2015; WBG 2021a). Geography, environmental processes, and socioeconomic conditions thus play out hand in hand and need to be studied in great spatial detail when drawing up the risk profile of the country. At-risk areas, communities, and assets need to be identified, investments focused, and policymaking tailored to benefit these priority groups, with the aim of enhancing equity and protecting the most vulnerable. This report assesses and quantifies the impact of climate hazards and exposure on poverty. Considering the increasing impacts of climate change in Bangladesh, especially on poorer communities, we explored whether poverty trends over time show a consistent, diverging pattern for hazard-affected and exposed spatial units. We found that while vast swaths of Bangladesh have seen a steep decrease in poverty rates over the 2010-2016 period, there has been a reverse trend for a significant number of clustered upazilas (at times referred to as thanas). Indeed, in eastern and southwestern Chattogram, southern Barisal, northeastern Dhaka, southern Rangpur, and northern Rajshahi Divisions, poverty rates have increased by up to two-fifths in the six-year period between poverty assessments. We also found that areas whose populations and built-up assets are heavily exposed to three or four co-occurring natural hazards showed no substantial decline in poverty rates between 2010 and 2016, whereas those with lower co-occurring exposure showed a mild or even stark decline in poverty rates over the same period. While causation cannot be confirmed, this pattern 8 suggests that climate change-related natural hazards could pose a barrier to lowering poverty rates in hazard-affected and exposed spatial units. This could hinder general poverty reduction efforts in the country, which thus far have shown great success compared to most countries in the developing world, hampering the achievement of development outcomes, and lowering the resilience of communities. Increasing resilience to natural hazards should therefore be a top priority in eastern and southwestern Chattogram, southern Barisal, northeastern Dhaka, southern Rangpur, and Rajshahi Divisions, where both poverty and natural hazards pose a double burden and show a worsening trend over time. The following subsections summarize the report’s key findings in three important areas: (1) small- scale spatial disparities in present-day natural hazard exposure and impact, (2) climate risks and changes in poverty over time, and (3) changing risks in the face of climate change. 1. Small-scale spatial disparities in present-day natural hazard exposure and impact Current and future climate risk is conceptualized in this work as a function of (a) the spatial occurrence, frequency, and intensity of natural hazards; (b) the exposure of population groups, built- up assets, and agricultural land to these hazards; and (c) their vulnerability—their propensity or predisposition to be adversely affected when exposed. Using deterministic and probabilistic modeling as appropriate, first, the present-day disparities in exposure to and the impact of natural hazards for Bangladesh are calculated down to the upazila level (third-level administrative units). Two key aspects of impact are captured in this work. Structural impact entails the expected risk of mortality and injury for exposed populations and the damage to infrastructure and cropland where impact functions are available, or their expected exposure to these hazards where specific functions are unavailable. Socioeconomic impact captures the deep spatial disparities in poverty, welfare, and development outcomes through overlaying exposure maps with the latest (2016) HIES-based poverty headcount rates. All major natural hazards have been included, resulting in disaster risk profiles for riverine and coastal floods, heat stress, droughts, landslides, and tropical cyclones. Moreover, while not a climate hazard in and of itself, we also looked at air pollution. This is a critical environmental concern for Bangladesh, and indeed across large parts of South Asia. The spatial combination of these exposure and hazard data is performed at the grid level, at a resolution matching that of the selected exposure layers. In our analysis, this was set to a common 90-meter grid. The resulting exposure and impact picture was first obtained for individual hazards: 1. Combining the impact of 10-, 100-, and 1,000-year floods on population mortality shows that the population at risk of death from riverine flooding is largest along the banks of the Ganges and Brahmaputra rivers in the southeastern upazilas of Rajshahi Division. Bera, Santhia, Sujanagar, and Shahjadpur Upazilas stand out, with over 2,700 citizens exposed to mortality from riverine floods on an annual basis. Saghata Upazila in Rangpur Division and Keraniganj Upazila just south of Dhaka show a similarly high expected annual impact (EAI) from riverine floods on population mortality. Population mortality from coastal flooding is particularly concerning along Bangladesh’s western coast in Khulna Division. In Shyamnagar and Assasuni Upazilas alone, over 4,000 citizens face the risk of mortality from storm surges. Both the EAI of riverine and coastal floods in terms of damage to built-up assets and exposure of agricultural 9 land to a combination of water depths of 10-, 100-, and 1,000-year floods show highly similar patterns. Incorporating socioeconomic vulnerability through the overlay of the 2016 poverty headcount rates captures the vast disparities in deprivation between flood-affected upazilas. Poverty rates of flood-exposed communities in Rangpur, Mymensingh, and Chattogram Divisions are considerably higher than for those in Rajshahi Division, and these populations are therefore considered to be particularly vulnerable to the impacts of riverine flooding. The higher poverty level in these divisions might push citizens with less mobility and poorer insurance and adaptative or preventative measures against high water to reside in hazard-prone areas. This is particularly important since agricultural land is also most extensively exposed to severe flooding in these areas, and rural livelihoods dependent on agriculture are therefore set to be adversely affected. For coastal flooding, a concerning, compounding effect of high expected annual coastal flooding impact and intermediate and high rates of poverty are seen across upazilas in southern Khulna and Barisal, with the coastal west of Chattogram Division also affected. 2. Heatwaves and heat stress are rampant across the entirety of Bangladesh’s territory, with wet bulb globe temperatures of approximately 35°C commonplace during heat stress episodes with a return period of 20 years. The Chittagong Hill Tracts fare comparatively better, yet with temperatures of over 30°C, indicating “sweltering heat,” heat stress still poses a severe threat to health. Given the widespread nature of these extreme events, the highest annual exposure to heat stress is found in and around urban centers. Dhaka and Sylhet stand out clearly, with millions of Bangladeshis facing extreme heat each year. Overlaid with socioeconomic vulnerability, the more densely populated upazilas of northern Rangpur, Rajshahi, and Mymensingh Divisions, as well as of the very northeast of Dhaka Division and southwestern Chattogram Division stand out, with high rates of poverty coinciding with a large population exposure to heatwaves annually. The communities in these areas are therefore the most at risk of negative health consequences brought about by extreme heat stress. 3. Especially in the northwest of Bangladesh, vast expanses of cropland are exposed to regular drought, occurring about once every four to five years. Zilas (second-level administrative units) in the northwestern regions of the country stand out. During the first cropping season, the most frequent and intense droughts are found in the northwestern zilas of Rangpur (Dinajpur) and Rajshahi (Naogaon, Chapai Nababganj, Natore) Divisions, as well as Narail Zila in Khulna Division, and Bhola Zila in Barisal Division. Over 30 percent of cropland in these areas experience drought stress at least once every four years. Crop failures as a result of drought could have severe impacts on food security. During the second cropping season, drought is much less of a concern. 10 Studying the relationship between drought patterns and 2016 poverty figures, concerns arise once again around upazilas in Bangladesh’s northwest. Upazilas in Rangpur, Rajshahi, and Mymensingh Divisions are some of the worst-affected areas, with northern Khulna and southwestern Chattogram Divisions also standing out negatively. In these geographic areas, poor communities that rely on agriculture for their livelihoods are particularly at risk of reduced crop harvests because of drought, possibly triggering price spikes and food insecurity. 4. In the very southeast of the Chittagong Hill Tracts, rainfall-triggered landslide intensity is high. While less densely populated than the floodplains of Bangladesh’s main rivers, there are still a considerable number of people and built-up areas exposed to landslides. In Sadar and Ramu Upazilas of Cox’s Bazar, over 100,000 residents live in landslide-prone areas. Relating this to the 2016 poverty headcount rates, the eastern upazilas of Chattogram Division are among the poorest in Bangladesh. The extremely high exposure of people and assets to intense landslide hazard thus makes these areas particularly at risk of the negative consequences of rainfall-triggered landslides. 5. Because of its flat topography, high wind speeds that accompany tropical cyclones are able to penetrate deep into the inland areas of Bangladesh, although the largest EAI of damage to built-up assets is found in and around major urban centers closer to the coastline. Dhaka and western Chattogram Division stand out. The overlay of the expected annual damage of tropical cyclones to built-up assets with poverty rates suggests that the highest relative co-occurrence of exposure and poverty is found spread along Bangladesh’s coastline, but also as far inland as Rajshahi Division and upazilas in northern Dhaka Division. In absolute terms, however, the impact is rather limited. 6. Fine particulate matter (PM2.5) concentrations exceed the World Health Organization norm of 37.5 µg/m3 maximum for a 24-hour mean on average every single year for which data are available across the whole of Bangladesh. This all-encompassing hazard predicts severe public health consequences, especially in densely populated urban areas. In cities, Dhaka in particular, millions of citizens are at increased mortality risk because of air pollution. A similar pattern to what was seen for heat stress risk emerges for air pollution risk when 2016 poverty rates are overlaid with PM2.5 concentrations. Again, the more densely populated upazilas of northern Rangpur, Rajshahi, and Mymensingh Divisions, as well as of the very northeast of Dhaka Division and southwestern Chattogram Division show a possibly devastating combination of high poverty rates and the highest expected annual population impact on mortality because of pollution. In sum, two aggregations of upazilas in Bangladesh face particularly challenging environmental conditions. Firstly, the western upazilas of Mymensingh and the eastern upazilas of Rangpur face a 11 combined high annual impact of riverine flooding with a high exposure to heat stress, agricultural drought, and air pollution. The high rates of poverty in these areas call for the explicit attention of policymakers and development partners in ensuring that the lives and livelihoods of these population groups are protected in the face of climate-related disasters through adaptative and preventative measures. A similar level of attention should be given to Chattogram Division, which should be disaggregated into southeastern and southwestern sections, as different dynamics are at play. Whereas the highly deprived southeast faces considerable landslide risk, the southwest is confronted with co- occurring issues of coastal floods, heat, drought, tropical cyclones, and air pollution, which require specific and tailored preventative and adaptative interventions and capacities. 2. Climate risks and changes in poverty over time While the hotspots of concern identified through present-day natural hazard, exposure, and impact modeling show the areas and communities at highest risk today, it is important to explore whether poverty trends over time show a consistent, diverging pattern for more, or less, hazard-affected, and exposed spatial units. To that end, the change in poverty headcount rates between 2010 and 2016 was compared with the number of included hazards—riverine floods, coastal floods, heat stress, drought, tropical cyclones, landslides, and air pollution—for which each upazila falls into the highest decile of relative population or built-up asset exposure. A worrisome observation emerges: in the areas that have seen the starkest increase in poverty headcount rates between 2010 and 2016, natural hazards tend to have the highest, co-occurring impacts. Upazilas in eastern and southwestern Chattogram, southern Barisal, and across Rangpur and Rajshahi Divisions fall into the highest exposure decile for three or even four natural hazards, with a similar intensity and co-occurrence of exposure also seen in southern Khulna and central Dhaka Divisions. This raises particular concern about the reduced ability of these impoverished communities to prevent or adapt to the impacts of climate change. Upazilas whose population and built-up assets are heavily exposed to three or four co-occurring natural hazards showed no substantial decline in poverty rates between 2010 and 2016, whereas those with lower co-occurring exposure (none, one, or two hazards) showed a mild or even stark decline in poverty rates over this period. While causation cannot be confirmed, this pattern suggests that climate change-related natural hazards could pose a barrier to lowering poverty rates in hazard-affected and exposed spatial units. This could hinder general poverty reduction efforts in the country, which thus far have shown great success compared to most countries in the developing world, hampering the achievement of development outcomes, and lowering the resilience of communities. In addition, the findings might also point to an out-migration of wealthier households, with poorer, more vulnerable households remaining behind in hazard-affected areas. Increasing resilience to natural hazards should therefore be a top priority in eastern and southwestern Chattogram, southern Barisal, Rangpur, and Rajshahi Divisions, where both poverty and natural hazards pose a double burden and show a worsening trend over time. 3. Changing risks in the face of climate change Having established the present-day risks posed by the exposure of people, built-up assets, and agricultural land to a wide variety of natural hazards, a forward-looking perspective needs to be adopted to explore how climate risks are set to develop across space in the decades ahead. The 12 screening and assessment of future impacts of natural hazards due to climate change typically involves the comparison of baseline conditions (observed or simulated) against future scenarios of climate variability. These baseline conditions are determined by computing the long-term average of a climate variable, setting a baseline historical value for a climate variable defined as the norm. Projections of this climate variable in the future will show anomalies, or deviations, of this variable relative to the historical baseline. To set this baseline and model projected anomalies for the future, we obtained data from climate models released under the Intergovernmental Panel on Climate Change Sixth Assessment Report (AR) framework (IPCC 2021a), supported by coordinated climate modeling efforts referred to as Coupled Model Intercomparison Projects (CMIP). The 2021 sixth AR features new CMIP6 models, which substantially expand the number of participating groups (49), climate models (100), and future scenarios examined (8) compared to previous ARs. We drew on CMIP6 data for our future modeling, and took into account three climate change scenarios, which are called Shared Socioeconomic Pathways (SSPs). These pathways cover the range of possible future scenarios of anthropogenic drivers of climate change by accounting for various future greenhouse gas emission trajectories, as well as a specific focus on carbon dioxide (CO2) concentration trajectories (IPCC 2021b). First, we included SSP1, a scenario with low greenhouse gas emissions and CO2 emissions declining to net zero after 2050, followed by net negative CO2 emissions. Next, we looked at SSP2, with intermediate greenhouse gas emissions and CO2 emissions remaining near current levels until the middle of the century before dropping off. As a final scenario, we incorporated SSP5, a very high greenhouse gas emission scenario with CO2 emissions that roughly double from current levels by 2050. As a time horizon, the period 2041-2060 was selected. With a gridded resolution of about 100 kilometers, we were able to aggregate the modeled information at the level of Bangladesh’s divisions and pinpoint which divisions are likely to face an increase or reduction in the frequency and intensity of a specific natural hazard under specific SSP scenarios. Underlying these projections are several key climate variables connected to changing patterns of precipitation and temperature. Looking at the number of consecutive wet days, days with rainfall over 10 millimeters, precipitation volume on extremely wet days, maximum five-day precipitation, and sea level rise allowed for the forecasting of likely changes in the risk of floods and rainfall-induced landslides across Bangladesh. We used the Wet Bulb Globe Temperature heat index to predict spatial changes in heat stress for the three climate scenarios. Finally, the number of consecutive dry days and the Standardized Precipitation-Evapotranspiration Index (SPEI) allowed for the assessment of projected changes in drought patterns for the country. No projections for air pollution or tropical cyclones can be made. The hazard-specific results can be summarized as follows: 1. Studying the climate projections for the five precipitation-related variables, a complex spatial pattern emerges. While patterns of cumulative rainfall over five days or consecutive wet days are likely to remain static or see a slight increase in standardized anomalies in the 2041-2060 period compared to the 1995-2014 baseline period, the number of days with rainfall over 10 millimeters and the amount of precipitation during extremely wet days show a strong upward trend. While the forecasts for days with rainfall over 10 millimeters are consistent across the three SSPs, projected anomalies in precipitation during extremely wet days are starker for the 13 higher-emission future scenarios, in particular SSP5. While no structural changes in overall precipitation patterns and conditions of wetness are thus expected for Bangladesh, the rainfall volumes during extremely wet days and the frequency with which such days occur are set to substantially increase. This is particularly the case in the northern divisions, which already face the most severe precipitation events, though increases are projected across the country throughout the 2041-2060 period. In the case of higher greenhouse gas emission scenarios, these increases will be exacerbated. Along the coastline, the starkly growing anomalies of sea level rise raise tremendous concerns, in particular for Khulna Division, which will see the greatest increase in standardized sea level rise anomalies by mid-century—of over 40 centimeters—in any climate change scenario. The projected increase in extreme rainfall events, particularly in the northern divisions, in combination with strong sea level rise, is alarming, particularly as this can go hand in hand with more frequent and severe coastal and riverine flooding and rainfall-induced landslides. Thus, adaptative measures to prevent an increase in mortality and damage to built-up and agricultural assets will need to be taken, with particular attention to the north of the country and along the coastal floodplains. 2. For drought, the SPEI shows little change in standardized anomalies under any of the SSPs compared to the baseline. This suggests that little change is expected in agricultural drought patterns during the 2041-2060 period compared to the 1995-2014 baseline period. The climate projections therefore do not foresee a significant worsening—or improvement—of this spatial pattern of agricultural drought in the decades ahead. However, under all three SSPs, and in particular under the high-emission SSP, the number of annual consecutive dry days is forecasted to increase mildly, as the standardized anomalies would increase above the historical baseline. However, this increase is limited, and uniform across Bangladesh. Areas with a historical pattern of a large number of consecutive dry days each year, mainly the western divisions, are thus set for a further increase in short-term dry episodes mid-century, particularly in the event of sharply rising greenhouse gas emissions. Elsewhere, standardized anomalies rise as well, yet this increase is comparatively smaller, suggesting the historical patterns are unlikely to shift significantly. This limited shift in projected drought patterns suggests that current agricultural drought concerns—closely related to food pricing and security—will remain valid in the decades ahead. 3. Under all three SSPs, and relatively uniformly across Bangladesh, the standardized anomalies of heat are forecasted to increase relative to the 1990-2010 baseline period used for this indicator. This increase is considerably stronger under the higher-emission future scenarios, and particularly significant in historically cooler divisions such as Chattogram. The projected rise in temperatures is concerning, as it indicates increased heat stress across Bangladesh by the 2041-2060 period, potentially putting even more lives at risk. The fact that this increase is fairly homogenous across the entirety of the country’s territory raises particular questions on adaptation and prevention for urban environments, where heat has historically been a cause of significant climate risk. Combining this diverse set of climate change variables, and cutting across SSPs, a worrisome picture emerges for Bangladesh, in particular in relation to heat stress, riverine and coastal flooding, and 14 landslides for the period 2041-2060. Current heat stress patterns are set to increase in severity, and while overall precipitation patterns are unlikely to considerably alter, extreme rainfall events are likely to increase in magnitude. The fact that we are only two decades away from these scenarios materializing calls for prompt and decisive adaptation, while mitigation of greenhouse gas emissions stands out as equally important, since the adverse effects of climate change will deepen under worse emission conditions. 15 1. Introduction Bangladesh is among the countries most affected by extreme weather events globally. The long-term Climate Risk Index ranks Bangladesh as the seventh most at-risk country over the 2000-2019 period, with 185 extreme events recorded and 0.38 fatalities per 100,000 inhabitants (Eckstein, Künzel, and Schäfer 2021). Calculating precise numbers of affected people and economic and welfare impacts remains challenging, yet average annual losses to disasters are estimated to be about US$3 billion, close to 2 percent of Bangladesh’s gross domestic product (GDP). While these mainly affect the agricultural sector, with wide-ranging effects on the rural economy, natural hazards increasingly impact the country’s rapidly expanding urban areas, where the service sector is predominant (WBG 2021a). Projections of global temperature and precipitation indicate that, even under optimistic global climate scenarios, Bangladesh faces a strong increase in the frequency and severity of extreme weather events and long-term climatic changes throughout the twenty-first century, compounded by intensified environmental degradation and urbanization. Future climate models, based on greenhouse gas concentration trajectories called Shared Socioeconomic Pathways or SSPs, forecast that without adaptative measures and under a business-as-usual scenario, over 7 million Bangladeshis will experience annual flooding in coastal areas (WBG 2021a). No other country is more vulnerable to flooding, with heatwaves, droughts, tropical cyclones, and more locally, landslides, posing compounding climate-related threats (ADB 2021). A detailed study of Bangladesh’s current and future climate risks is therefore of paramount importance to devise tailored policies and targeted investments. 2. Spatial disparities in hazards and poverty rates, and their link to climate vulnerability Natural hazards do not occur uniformly across space, nor is their impact equally distributed. The frequency and intensity of hazards, and the vulnerability of exposed communities and assets to these hazards will determine how severely a specific location is struck. Poorer households might be disproportionally affected by extreme events because they tend to reside in more disadvantaged and hazard-prone areas, have less mobility and poorer access to critical services and early warning systems, and their assets are less well-insured against extreme climate events (Dasgupta 2015; WBG 2021a). Geography, environmental processes, and socioeconomic conditions thus play out hand in hand and need to be studied in great spatial detail when drawing up the risk profile of the country. At-risk areas, communities, and assets need to be identified, investments focused, and policymaking tailored to benefit these priority groups, with the aim of enhancing equity and protecting the most vulnerable. While poverty rates have dropped dramatically in recent decades—from 48.9 percent at the start of the century to 24.3 percent in 2016 according to national poverty lines (WBG 2022a)—deep spatial disparities in poverty, welfare, and development outcomes remain. Large gaps can be seen across the urban-rural divide in all divisions (first-level administrative units), and across zilas and upazilas within divisions (second- and third-level administrative units). Map 2.1 shows the 2010 poverty rates at the upazila level, with division boundaries highlighted. Across upazilas in the north (Rangpur and 16 Mymensingh Divisions), southwest (Khulna, northern Barisal, and southern Dhaka Divisions) and farthest southeast (Chattogram Division), up to two-thirds of the population lived in poverty. Northern Chattogram, central Dhaka, Sylhet, and Rajshahi Divisions performed better, comparatively. Map 2.2 shows the 2016 poverty rates (the most recent available) at the upazila level. Also shown on the map again are the eight divisions. The data are drawn from the 2016-2017 Household Income and Expenditure Survey (HIES) carried out by the Bangladesh Bureau of Statistics (BBS) and matched with population data from the 2011 census (BBS 2011) for the 544 upazilas in the country at the time of the census. The indicator shown on the map is the poverty headcount rate in percent, using upper poverty lines. The poverty map for 2016 shows that monetary deprivation continues to differ greatly across space and is clustered at a highly granular level in specific divisions. Looking at the highest quintile of poverty, the most deprived communities are still found to reside in the northwest of the country, in particular across Rangpur and Mymensingh Divisions, as well as western Rajshahi Division and northeastern Dhaka Division. Similarly high poverty rates are observed in Bangladesh’ eastern Chattogram Division, especially in the Chittagong Hill Tracts along its eastern border. In these areas, more than two out of five Bangladeshis are poor compared to the country’s upper poverty line. Dhaka and Sylhet Divisions, and northwestern Chattogram Division perform comparatively better, with the majority of upazilas having a poverty rate below 20 percent. These strong disparities, and the observation that poverty rates differ greatly within divisions, emphasize the importance of zooming in to granular administrative units, where investment decisions are made and local policy should be targeted, and where the heterogeneity of socioeconomic conditions might make communities more, or less, vulnerable in the face of climate change. Especially in those upazilas with poverty rates over 40 percent, communities might be considerably less resilient because of a lower capacity to take preventative or adaptative measures, reduced chances to migrate away from at-risk areas, and a lower likelihood of being insured for natural disasters affecting their lives, assets, and livelihoods. In light of the increasing impacts of climate change, especially on poorer communities, it is important to explore whether poverty trends over time show a consistent, diverging pattern for more or less hazard-affected and exposed spatial units. Map 2.3 lays the foundation for this analysis by plotting the change in poverty headcount rates between 2010 and 2016. While vast swaths of Bangladesh have seen a steep decrease in poverty rates over the 2010-2016 period, there has been a reverse trend for a significant number of clustered upazilas. Indeed, in eastern and southwestern Chattogram, southern Barisal, northeastern Dhaka, southern Rangpur, and northern Rajshahi Divisions, poverty rates have increased by up to two-fifths in the six-year period between poverty assessments. 17 Map 2.1: Poverty Rates at the Upazila Level, 2010 Source: Map developed by the World Bank for this report based on data from BBS 2011. Map 2.2: Poverty Rates at the Upazila Level, 2016 Source: Map developed by the World Bank for this report based on data from the 2016 HIES (BBS 2016-2017). 18 Map 2.3: Change in Poverty Rates between 2010 and 2016 in Bangladesh’s Upazilas Source: Map developed by the World Bank for this report based on data from BBS 2011 and the 2016 HIES (BBS 2016-2017). 3. Defining present-day disaster risk: definitions, data, and methodology The Intergovernmental Panel on Climate Change (IPCC) defines a hazard as the potential occurrence of a natural or human-induced physical event or trend that may cause loss of life, injury, or other health impacts, as well as damage and loss to property, infrastructure, livelihoods, service provision, ecosystems, and environmental resources (IPCC 2019). Exposure describes the location of people and assets in an environment where they may be threatened by these natural hazards. People and assets may be exposed, but not adversely impacted if they are not vulnerable. Vulnerability summarizes the propensity or predisposition to be adversely affected when exposed, measured by characteristics that favor the negative impact of a hazard if exposed to it. Disaster risk is then the probability of a negative impact in the future caused by a hazard. Together, hazard (H), exposure (E), and vulnerability (V) drive disaster risk (R) (IPCC 2012): R=f(H, E, V) Having defined these core concepts, a modeling approach for each of the included hazards needs to be selected. For baseline hazard modeling, there are two options. Modeling is either: 19 a. Deterministic, in the form of an individual spatial layer measuring the mean or maximum intensity of a hazard. b. Probabilistic, in the form of several spatial layers representing the maximum hazard level for a specific occurrence frequency, measured as a return period (RP). The spatial combination of exposure and hazard data is performed at the grid level, at a resolution matching that of the selected exposure layers. In our analysis, this is set to a common 90-meter grid. The impact model translates the physical intensity unit of the hazard into a damage factor (0 to 1), which is then multiplied by the exposure layer. Impact models can be quantitative or qualitative. When multiple probabilistic scenarios of a hazard are available, the impact for each scenario (RPi) is summed up for the smallest administrative unit level. In this case, the expected annual impact (EAI) is calculated by multiplying the impact with the frequency (1/RP) of each event, and then summing up to one value. This is illustrated in Figure 3.1 below. The exceedance frequency curve shown in this figure is built by the interpolation of these points, highlighting the relationship between the RP of each hazard and the estimated impact. The area below the curve represents the total annual damage. Figure 3.1: Computation of Expected Annual Impact of Natural Hazards in Geospatial Analytics Source: World Bank, adapted from Colorado Water Conservation Board (2020). In this geospatial section, we focus on disaster risk posed by riverine and coastal floods, heat stress, droughts, landslides, and tropical cyclones in Bangladesh. Moreover, while not a climate hazard, in and of itself, we also look at air pollution. This is a critical environmental concern for Bangladesh, and indeed across large parts of South Asia. It can also compound the impact of many climate hazards, including droughts and heatwaves, and exacerbate many aspects of vulnerability through its impact on health, educational attainment, and labor productivity. Geophysical hazards (earthquakes, tsunamis, volcanoes, and dry landslides) are not a standard consideration in this work but could be included with relatively little additional effort. To assess and quantify the impact of these hazards, whether extreme events or long-term climatic changes, we look at three types of exposure: (1) population, (2) built-up assets, and (3) agricultural 20 land. Where available, an impact function is added to the exposure variables to demonstrate the EAI on population health in terms of morbidity and mortality, or the potential damage to built-up assets and agricultural land. An overview of the key hazards for Bangladesh that are included in the analysis, the relevant exposure categories by hazard, and whether an impact function or exposure metric is used is shown in Table 3.1 below, along with the main data sources. Table 3.1: Hazard, Exposure, and Impact – Upazila-Level Analysis Exposure Categories Hazards Population Built-up assets Agricultural land (Health - mortality) (Physical damage) (Physical damage) Flood  Exposure Impact function  Impact function Fathom 90m Flood Model classification Landslide Exposure Exposure classification World Bank-ARUP Landslide Index at 1km classification Heat stress Exposure World Bank-GFDRR Wet Bulb Globe   classification Temperatures at 10km Drought Exposure     FAO Agricultural Stress Index at 1km classification Tropical cyclone Impact function NOAA Wind Speeds at 30km Air pollution Van Donkelaar et al. Surface PM2.5 Impact function Concentrations at 1.1km Source: Fathom 2022; World Bank-ARUP 2022; World Bank-GFDRR 2017; FAO 2022; NOAA 2018; and Van Donkelaar et al. 2021. 4. Conceptualizing hazards and the current setting in Bangladesh a. Floods and storm surges Located downstream of the Ganges, Brahmaputra, and Meghna river basins with particularly intense and voluminous rainfall during the June to September monsoon season, and facing storm surges of several meters high and tropical cyclones along the coastline, Bangladesh is particularly vulnerable to flooding (WMO 2018). The flat topography and extreme—and worsening—climate variability add to this burden. Each year, on average, more than a fifth of the country’s surface floods, extending over three million hectares. During the extreme flooding events of 1988 and 1998, this reached up to 70 percent of Bangladesh’s territory (Tufts University 2016). Average annual extreme weather event- related losses amounted to 1.8 percent of GDP over the 1990-2008 period, with peaks of up to 5 percent in the 1998 flood year (IMF 2019). Historical extreme events have shown how flooding induced by monsoon rains can cause widespread devastation. During the 2020 monsoon season starting in June, over a quarter of the country was flooded, ruining the livelihoods of millions and killing at least 100 over the space of two months (NASA 2020). The most vulnerable groups are particularly at risk. In the period between the end of 21 July and early September 2021, Cox’s Bazar experienced over 1,300 millimeters of rain, with over 200 millimeters falling in just the 48 hours of 27-28 August. Five hundred and forty-two monsoon-related incidents were subsequently reported in the refugee settlements, affecting nearly 12,000 shelters (UNOCHA 2021). Along the coast, storm surges become concerning during tropical cyclone events. For instance, in 1970, Cyclone Bhola caused a storm surge of nine meters, which swept away houses, crops, and hundreds of thousands of livestock (ADB 2021). With 60 percent of the population living less than five meters above sea level, the threat is tremendous. Beyond people and their built-up assets, agricultural land is also considerably at risk. The 2017 floods affected eight million people and caused severe damage to crops, resulting in spiraling food prices and severe trade disruptions, costing some 0.3 percent of GDP (IMF 2019). Climate change is set to increase the frequency and intensity of these events. Extreme river flows, now occurring once every century, are likely to be repeated at least every 25 years (WBG 2021a). Up to 900,000 people could be forced to migrate out of coastal areas because of permanent inundation by mid-century in the Ganges Delta, where the sea level could increase by up to 10 millimeters per year under a negative climate change scenario (WBG 2021a). To study the spatial patterning of this flood hazard across Bangladesh, riverine and coastal flood hazard probabilities are depicted by the Fathom Global 2.0 dataset, at a 90-meter spatial resolution. The dataset covers 10 RPs, depicting how often flooding of a certain inundation depth occurs. The risk and the associated damage to infrastructure, crops, and livestock, as well as mortality and morbidity, vary substantially with variations in land use and settlement patterns. The three exposure categories studied for flood hazard—population, built-up assets, and agricultural land—address the three key sources of impact: the people, the economy, and the environment. The impact of floods will also differ by water depth. While little damage and mortality are expected at low water depths, damage can increase substantially as flood depth rises. This impact has been approximated by impact functions for population and built-up area, respectively. This yields an EAI, which is aggregated at the upazila level for floods with RPs of 10, 100, and 1,000 years. Simple impact functions for agriculture are not easily suited to assess agricultural impact, so a measure of exposure of cropland is included in these analytics instead. b. Heatwaves Historically, the average temperature in Bangladesh has been about 26°C, with temperatures ranging between 15°C and 34°C throughout the year—though dangerous (over 40°C) and extremely dangerous (over 54°C) days have been commonplace over the last half century (WBG 2021b; Mahmud, Raza, and Hossain 2021). Nonetheless, Bangladesh has been growing and continues to grow hotter—a 0.5°C increase in mean temperature was recorded between 1976 and 2019 —with temperatures also becoming more erratic. Summers have become longer and warmer, with temperatures for March to October rising by up to 1.3°C compared to the start of the last century. The median number of days per year during which Bangladesh experiences a heat index of greater than 35°C is about 70 days, reflecting a highly frequent, heat-stressed environment (WBG 2021a). Warmer winters also mean Bangladesh is losing some of its distinct seasonality (Mahmud, Raza, and 22 Hossain 2021). As climatic changes continue to spiral, mean temperatures across Bangladesh are set to increase by between 1.4°C and 2.4°C by 2100 (Mahmud, Raza, and Hossain 2021). Heat discomfort increases when hot temperatures are associated with high humidity (Coffel, Horton, and de Sherbinin 2017). Periods of “abnormally and uncomfortably hot and humid weather” (AMS 2012) threaten people’s health and infrastructure. Being a very wet country, with over 2,200 millimeters of rainfall a year, this is a particular concern for Bangladesh (WBG 2021b). These conditions take a toll on human health and overwhelm health systems in Bangladesh, with greater consequences for places where extreme heat occurs in the context of aging populations, urbanization, urban heat island effects, and health inequalities (Mahmud, Raza, and Hossain 2021). Heatwaves pose manifold risks to human development in a variety of ways, ranging from direct impacts on health, to indirect and downstream risks such as power cuts (Allen-Dumas, Binita, and Cunliff 2019), to long-term impairment of learning capacity and educational attainment as well as labor productivity and thus, income (Goodman et al. 2018). Heat-related mortality among the elderly has increased by 53.7 percent in the past 20 years and people’s productivity at work has declined due to heat stress. An estimated 148 work hours per person were lost in 2019 due to heat stress in Bangladesh, which translates to 18.2 billion work hours lost in total for 2019, up from 13.3 billion work hours in 2013 (Mahmud, Raza, and Hossain 2021). Understanding the spatial distribution of heatwave risk is important for designing adaptation policies. In the current analysis, the focus lies on direct risks from heatwaves on health and productivity. Therefore, heatwave hazard probability, the exposure of people, and their vulnerability are mapped. Heatwave magnitude and frequency are altered by climate change. Hence, future projections of extreme heat are analyzed based on simulations of climate models. Heatwave probability is depicted by the maximum daily wet bulb globe temperature (WBGT, in °C) obtained from the World Bank Global Facility for Disaster Risk and Recovery. The WBGT combines temperature and humidity, both critical components in determining heat stress. Only WBGT values greater than 30°C were considered. Three RPs of heat events are included: once in 5, 20, and 100 years. c. Droughts Bangladesh faces an annual probability of severe meteorological drought of approximately 4 percent, with less severe droughts occurring more frequently. Some 1 percent of the population was exposed to drought in the 2001-2013 period (WBG 2021a). Future models are unclear about the drought patterns across Bangladesh over the course of the twenty-first century. Moreover, large-scale climate phenomena, such as the El Niño Southern Oscillation and the Indian Ocean Dipole, make predictions of change highly complex. Nonetheless, the rising impacts of climate change on agriculture are particularly concerning for Bangladesh, where agriculture is a primary source of livelihood, contributing 14.2 percent to the GDP and employing about 40.6 percent of the labor force. Annual climate change impacts are forecasted to reduce agricultural GDP by 3.1 percent on an annual basis (Mondol et al. 2021). 23 Drought-prone areas are located in the northwestern and northern regions of Bangladesh, spread over some 5.46 million hectares (ADB 2021). These droughts are associated with irregular monsoon rains and intermittent dry spells, whose frequency has increased by at least 10 percent in parts of Rangpur and Mymensingh Divisions over the 1979–2018 period (Mondol et al. 2021). These episodes have severe repercussions on food security, food prices, and the economy of Bangladesh. The consecutive droughts of 1978 and 1979 directly affected 42 percent of cultivated land and cut rice production by an estimated two million tons. The 1997 drought reduced grain production by one million tons, entailing a loss of about US$500 million (ADB 2021). This can have a devastating effect on local economies and communities that rely on agriculture for their livelihoods, and the occurrence of droughts and agricultural exposure to dry episodes need to be studied. Drought probability is depicted by the FAO Agricultural Stress Index, which plots the annual drought exposure of agricultural land (cropland and pastureland combined) between 1984 and 2022 at a 1 kilometer resolution. This information is available for the two main cropping seasons in Bangladesh: Kharif, which runs from May to October, and Rabi, which commences mid-November and continues through April (Mohsenipour et al. 2018). Especially during Kharif season, crops are vulnerable to a variety of natural hazard events, including drought (Aziz et al. 2021). Exposure has been quantified by the cropland land cover class of ESA WorldCover 2020 (ESA 2020). d. Landslides Landslide hazards are particularly prevalent in the eastern Chittagong Hill Tracts of Bangladesh. While the root cause of landslides is primarily heavy rainfall during the monsoon season, a recent trend of spontaneous urbanization in the hills has posed a sharply increased risk of loss of life and damage to critical infrastructure and ecosystems (Ahmed 2021). Landslides thus rarely occur in isolation and pose a compounding threat for vulnerable communities. The excessive rainfall in Cox’s Bazar referred to earlier also went hand in hand with landslides that destroyed refugee shelters in July 2021 (UN 2021). While the geographic area in which landslides occur in Bangladesh is limited, the human and economic impacts can be devastating. In the last 30 years, landslides have caused more than 350 deaths. A series of landslides in June 2017 caused landslides in 145 places in the Chittagong Hill Tracts, killing 168, damaging 40,000 houses, and resulting in an economic loss of over US$200 million (Abedin et al. 2020). Landslide probability is depicted by the World Bank-ARUP landslide hazard index at a 1 kilometer resolution. Since there are no tailored impact functions available for landslides, the focus lies on the exposure of the population, built-up assets, and agricultural land to this hazard. e. Tropical cyclones Tropical cyclones are events that can trigger multiple hazard processes at once, such as strong winds, intense rainfall, extreme waves, and storm surges. However, when studying tropical cyclones as part of the Country Climate and Development Report analytics, we considered only the wind component of the cyclone hazard, while the flood and storm surge components were considered separately, as described above. 24 Bangladesh is particularly vulnerable to tropical cyclones, as illustrated by Cyclone Bhola, which caused 300,000 to 500,000 fatalities in November 1970—the deadliest tropical cyclone in history (WMO 2020). At least 12 major tropical cyclones have hit the country since 1965 (ADB 2021). Over the last half century, tropical cyclones have caused a startling 718,000 deaths. However, cyclone-related mortality in Bangladesh has declined considerably since the 1970s, with numbers dropping more than one hundred-fold, largely because of better preventative measures. Currently, there are 12,000 functionally active cyclone shelters and a warning system in use along the coastline, which are able to provide shelter to and alert nearly 5 million people (UCL 2020). Despite these preventative measures, millions of people—and their assets—are still exposed to frequent cyclones. In 2007, 9 million people were exposed to Cyclone Sidr, with Category 1 tropical Cyclone Mora exposing some 10 million people to wind speeds of 120 kilometers per hour in 2017 (WBG 2021a). Despite these tremendous potential impacts, the ways in which climate change will interact with cyclones is currently poorly understood. Sea level rise, increased wind speed, and precipitation intensity all come into play, and while the total number of cyclones is set to decrease according to certain models, the intensity and severity of extreme events is likely to increase (WBG 2021a). To model wind speeds associated with tropical cyclones, we have used the International Best Track Archive for Climate Stewardship wind speed dataset, with five RPs and a resolution of 30 kilometers. f. Air pollution Bangladesh has the worst air pollution of any country in the world. Air pollution shortens the life expectancy of the average Bangladeshi by 6.7 years, and up to 8.1 years in the worst-affected areas (AQLI 2022). Every single inhabitant lives in an area where the World Health Organization (WHO) air quality norms are regularly exceeded. At 65.5 µg/m3, fine particulate matter (PM2.5) concentrations are at least four times the WHO recommended limits for daily exposure (AQLI 2022). These small particles, of less than 2.5 micrometers in diameter, pose the greatest health risks—because of their small size, they lodge deeply into the lungs (WHO 2019). There they cause respiratory infections, cardiovascular disease, and lung cancer. Pollution caused some 28 percent of deaths in 2015, more than 10 times the number of deaths from traffic accidents, and cost the country US$6.5 billion, or 3.4 percent of its GDP (WBG 2018). The capital, Dhaka, is the most polluted city in the world (Associated Press 2021). While not a climate hazard in and of itself, air pollution is adversely affected by climate change (UCAR 2022). Moreover, it is associated with numerous hydrometeorological hazards that will increase in frequency and intensity because of changing climatic conditions, such as droughts and heatwaves. Using surface-level PM2.5 concentrations with a 1.1 kilometer resolution, and drawing from the research of Van Donkelaar et al. (2021), who combined Aerosol Optical Depth retrievals from the NASA MODIS, MISR, and SeaWiFS instruments with the GEOS-Chem chemical transport model, the high-resolution spatial patterning of air pollution can be mapped for Bangladesh. It is important to note that this combines both human-induced PM2.5 emissions caused by car engines, coal- or natural 25 gas-fired power plants, fireplaces, and biomass burning (NRDC 2022), and natural sources of PM2.5, which include forest fires and desert dust (McDuffie et al. 2021). PM2.5 data are available from 1998 to 2020. 5. Conceptualizing exposure and the current setting in Bangladesh a. Population Whether a household will be affected by a natural hazard depends on their place of residence. Especially for smaller-scale, well-defined extreme events such as floods or landslides, administrative unit-level population data, for instance census data, do not offer the necessary granularity to explore the precise settlement location of citizens and to study their exposure. Therefore, the high-resolution location of the population was plotted using the WorldPop 2020 population model, where each cell has a 90-meter resolution. Map 5.1, panel a shows the distribution of the population across the country. High population densities are visible in and around the Dhaka metropolis in central Bangladesh. Urban sprawl is noticeable to the northwest and northeast of the country. Bangladesh’s second city, Chattogram, also stands out in the southeast. This accurately depicts the dense urban areas along Bangladesh’s major riverine systems. By contrast, the low population densities in the very southwest and east stand out. Two notes of caution need to be added with regard to the WorldPop data. First, as lower-density population cells are predominant, they visually crowd out cells with higher populations on Map 5.1, panel a, and only become visible when zooming in on urban areas. This is shown through the zoomed in view of the city of Dhaka in Map 5.1, panel b, where large urban areas with high population densities become visible. Second, WorldPop is a constrained population model—hence, it allocates population numbers proportionally to remotely sensed built-up density. This can generate model errors, particularly in mountainous environments, where populations are allocated to a limited number of built-up cells in valleys. This aggregation of population in valleys and allocation to specific cells might entail an overestimation of natural hazard risk in those areas, if affected. The aggregation of exposure and impacts at the level of administrative units, like upazilas, reduces this error through the scaling of population estimates. However, a residual error cannot be ruled out. As a predominantly flat country, with limited mountainous areas, this model error is hypothesized to be limited for Bangladesh. 26 Map 5.1: WorldPop Bangladesh Population Model, 2020 Panel a: All of Bangladesh Panel b: Zoomed in on the city of Dhaka Source: World Bank, original maps developed for this report based on data from WorldPop 2020. b. Built-up assets Built-up assets include homes, industrial complexes, road infrastructure, and facilities, among other infrastructure. For the analysis, 2019 World Settlement Footprint (WSF) data are employed. This is a high-resolution (10 meters) remotely sensed dataset which indicates whether each cell is primarily built up. To match the resolution of the WorldPop population data, WSF data were resampled at a 90- meter resolution. For each cell, the share of built-up land was then calculated, as shown on Map 5.2. A pattern similar to the population distribution emerges, whereby built-up assets are most common around Dhaka, with secondary urban centers scattered across the expanse of Bangladesh, in particular around Chattogram. 27 Map 5.2: WSF 2019 Distribution of Built-Up Areas across Bangladesh Source: World Bank, original map developed for this report based on data from WSF 2019. c. Agricultural land The 2020 WorldCover dataset from the European Space Agency (ESA 2020) was used to identify agricultural land, aggregated from its original 10-meter resolution to a 90-meter resolution to match the population model, as per Map 5.3. Unsurprisingly, as a primarily flat, fertile country, cropland is found across most of Bangladesh’s floodplains around the Ganges, Brahmaputra, and Meghna rivers, with wetlands directly bordering these river systems. In the southwestern delta, brackish waters are predominant, with forests covering most of the eastern Chittagong Hill Tracts. 28 Map 5.3: Bangladesh Land Use, 2020 Source: World Bank, original map developed for this report based on data from ESA 2020. 6. Results: mapping of present-day hazards, exposure, and vulnerability Bringing together hazard, exposure and vulnerability, Table 6.1 provides an overview of the categories of exposure for which results are available (population health, built-up assets, and/or agricultural land), as well as the structural and socioeconomic vulnerability included in this work. Table 6.1: Hazard, Exposure, and Vulnerability Categories included for the Current Climate Analytics in Bangladesh Hazard Exposure Vulnerability Structural Socioeconomic River/Coastal Population (Health/Mortality) Impact function floods Built-up assets (Physical damage) Impact function Poverty Agricultural land (Physical damage) Exposure classification Heatwaves Population (Health/Mortality) Exposure classification Poverty Droughts Agricultural land Exposure classification Poverty Landslides Population (Health/Mortality) Exposure classification Poverty Built-up assets (Physical damage) Exposure classification Tropical Built-up assets (Physical damage) Impact function Poverty cyclones 29 Hazard Exposure Vulnerability Structural Socioeconomic Air pollution Population (Health/Mortality) Impact function Poverty Source: World Bank, original table developed for this report. a. Riverine and coastal flood risk The floodplains around the Ganges, Brahmaputra and Meghna rivers are the lifeline and agricultural heartland of Bangladesh. As seen in the exposure maps above, most people, assets, and agricultural land are located here. The large intra-annual variation in precipitation across the country translates into different riverine flooding hazards, as seen in Map 6.1, panel a. Flood depths of up to five meters are observed along the riverine systems in the case of a flood event with a 100-year RP, with less extensive pockets of high flood hazard also found along the northeastern border in Mymensingh and Sylhet Divisions. Similarly high flood depths through coastal flooding are observed in the southwestern delta where the river systems converge, in particular in the Sundarbans mangrove area. Map 6.1, panel b depicts the water depths associated with a coastal flood happening once every 100 years, which in this region is often triggered by tropical cyclones. Map 6.1: Riverine (panel a) and Coastal (panel b) Flood Hazard at a 90m Resolution across Bangladesh for a 100-Year Return Period Panel a Panel b Source: World Bank, original maps developed for this report based on data from Fathom 2022. Floods will only have an impact if they occur in areas with population and asset exposure. The riverine and coastal flood disaster mapping therefore provides a spatial profile of where the highest annual impact on population and built-up assets is expected under the current climate, and where agriculture is most exposed. Combining the impact of 10-, 100-, and 1,000-year floods on population mortality shows that the population at risk of death from riverine flooding is greatest along the banks of the Ganges and 30 Brahmaputra rivers in the southeastern upazilas of Rajshahi Division (Map 6.2, panel a). Bera, Santhia, Sujanagar, and Shahjadpur Upazilas stand out, with over 2,700 citizens exposed to mortality by riverine floods on an annual basis. Saghata Upazila in Rangpur Division, and Keraniganj Upazila just south of Dhaka show similarly high EAIs of riverine floods on population mortality. Population mortality through coastal flooding is particularly concerning along Bangladesh’s western coast, in Khulna Division (Map 6.2, panel b). In Shyamnagar and Assasuni Upazilas alone, over 4,000 citizens face a risk of mortality from storm surges on an annual basis. Both the EAI of riverine and coastal floods in terms of damage to built-up assets (Map 6.3, panels a and b) and of exposure of agricultural land (Map 6.4, panels a and b) to a combination of water depths of 10-, 100-, and 1,000-year floods show highly similar patterns. In eastern Rajshahi Division, multiple upazilas have over 30 hectares of infrastructure threatened by water damage annually by riverine floods, while the same is seen in the very southwest of the country around Shyamnagar Upazila. In those areas, over 1,000 hectares of agricultural land are also annually exposed to inundations of over 0.5 meters, with potentially dramatic impacts on crop harvests, food prices, and food security. However, flood depth cannot drive the assessment of flood risk on agriculture on its own. Additional variables such as the types of crops and livestock, the timing of the flood in relation to the agricultural production cycle, and the duration of the flood should be considered. However, global datasets do not provide this information. Map 6.2: Expected Annual Population Impact of Riverine (panel a) and Coastal (panel b) Floods – Mortality and Morbidity Panel a Panel b Source: World Bank, original maps developed for this report based on data from Fathom 2022 and WorldPop 2020. 31 Map 6.3: Expected Annual Damage to Built-Up Assets of Riverine (panel a, left) and Coastal (panel b, right) Floods Panel a Panel b Source: World Bank, original maps developed for this report based on data from Fathom 2022 and WSF 2019. Map 6.4: Expected Annual Exposure of Agricultural Land to Riverine (panel a, left) and Coastal (panel b, right) Floods Panel a Panel b Source: World Bank, original maps developed for this report based on data from Fathom 2022 and ESA 2020. Given that the vulnerability of citizens and their capacity to mitigate or prevent adverse effects from this exposure to floods is primarily driven by their socioeconomic status, the socioeconomic component needs to be included in the equation when determining climate risk. This integrated picture 32 is produced by overlaying the population, built-up asset, and agricultural land expected annual exposure (EAE) and/or EAI with the 2016 poverty headcount rate for upazilas. For each overlay, a dual approach was followed. Firstly, both the EAE/EAI and poverty headcount were broken into three equal quantile classes. This resulted in a 3x3 matrix, with both the exposure to and impact of natural hazards as well as the socioeconomic vulnerability divided into low, intermediate, and high categories of equal size. Such quantile approach allows us to depict— irrespective of the absolute number or size of the impacted populations or assets—which upazila communities, built-up areas, and agricultural land run the highest risk. Secondly, maintaining the poverty quantiles, we broke down the EAI and EAE into three self-defined classes, with boundaries depicting a low, intermediate, and high exposure and/or impact based on absolute values. Particularly for hazards that have a limited geographic extent or cause limited loss of lives and livelihoods, a classification based on absolute rather than relative values can reveal a more grounded understanding of the scope of climate risks in Bangladesh. Maps 6.5, 6.6, and 6.7 below show the EAI on population and built-up assets of, and agricultural exposure to, riverine floods overlaid by the upazila-level headcount rates from the 2016 poverty map, with the natural hazard component included both as quantiles (panel a of each of the maps), and through predefined class boundaries (panel b of each of the maps). The quantile maps show a consistent spatial pattern for all three exposure categories —population, built-up assets, and agricultural land. The EAI of riverine floods on mortality and morbidity (Map 6.5, panel a), damage to built-up assets (Map 6.6, panel a), and annually exposed cropland and pasture (Map 6.7, panel a) is greatest upstream and near the confluence of the Ganges and Brahmaputra rivers, covering the southeastern upazilas of Rajshahi and Rangpur Divisions, and western Mymensingh Division. The quantile maps show that the eastern part of Chattogram Division also falls in the highest quantile of population exposure to riverine floods. In absolute terms however, the highest impact on population health (>1,000 potential deaths annually, Map 6.5, panel b) and infrastructure (>30 hectares damaged, Map 6.6, panel b), and the largest swaths of agricultural land flooded (>1,000 hectares, Map 6.7, panel b) is limited to southeastern Rajshahi and Rangpur Divisions, and western Mymensingh Division. While exposure is thus considerable across northwestern Bangladesh, extreme poverty rates differ greatly between affected upazilas. Poverty rates of flood-exposed communities in Rangpur, Mymensingh, and Chattogram Divisions are considerably higher than for those in Rajshahi Division, and these populations are therefore considered to be particularly vulnerable to the impacts of riverine flooding. The higher poverty level in these divisions might push citizens with less mobility and poorer insurance and adaptative or preventative measures against the impacts of flooding to reside in hazard- prone areas. This is particularly important since these are the same areas where agricultural land is the most extensively exposed to severe flooding, and rural livelihoods dependent on agriculture are therefore set to be adversely affected. 33 Map 6.5: Expected Annual Impact of Riverine Floods on Population in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Fathom 2022, and WorldPop 2020. Map 6.6: Expected Annual Impact of Riverine Floods on Built-Up Assets in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b 34 Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Fathom 2022, and WSF 2019. Map 6.7: Expected Annual Exposure of Agricultural Land to Riverine Floods in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Fathom 2022, and ESA 2020. While riverine flooding primarily affects Bangladesh’s northwest, the south is exposed to considerable coastal flooding. Highly similar trends emerge from the maps showing the quantiles and predefined impact and exposure classes overlaid with 2016 poverty headcount rates (Maps 6.8, 6.9, and 6.10 for population impact, built-up asset impact, and agricultural exposure, respectively). A concerning, compounding effect of high expected annual coastal flooding impact and intermediate and high rates of poverty are seen across upazilas in southern Khulna and Barisal Divisions, with the coastal west of Chattogram Division also affected. Further north, in Dhaka, Mymensingh, and Sylhet Divisions, the impact of and exposure to coastal flooding is considerably lower. The combination of poor socioeconomic conditions and high impact of storm surges, particularly in Khulna and southeastern Barisal Divisions, draws critical attention to the risk coastal communities in both divisions face. This is especially critical given rising sea levels combined with the impact of tropical cyclones. Urgent adaptative measures need to be taken to protect the residents of these at- risk upazilas from loss of life and livelihoods, and damage to their assets. 35 Map 6.8: Expected Annual Impact of Coastal Floods on Population in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Fathom 2022, and WorldPop 2020. Map 6.9: Expected Annual Impact of Coastal Floods on Built-Up Assets in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Fathom 2022, and WSF 2019. 36 Map 6.10: Expected Annual Exposure of Agricultural Land to Coastal Floods in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Fathom 2022, and ESA 2020. b. Heatwave risk Heatwaves and heat stress are rampant across the entirety of Bangladesh’ s territory, with WBGTs of around 35°C commonplace during heat stress episodes with an RP of 20 years, as captured by Map 6.11. The Chittagong Hill Tracts fare comparatively better, yet with temperatures of over 30°C, indicating “sweltering heat”, heat stress still poses a severe threat to health. Given the widespread nature of these extreme events, we expect to see the highest annual exposure in and around urban centers. This is affirmed by Map 6.12, which shows the annual expected population exposure by combining heat stress episodes with RPs of 5, 20, and 100 years. Dhaka and Sylhet Divisions stand out clearly, while the expected annual population exposure to heat stress in less densely populated and higher elevated Chattogram Division is considerably lower. In total, millions of Bangladeshis face extreme heat each year. 37 Map 6.11: Heat Stress across Bangladesh for a Heat Event with a 20-Year Return Period Source: World Bank, original map developed for this report based on data from World Bank-GFDRR 2017. Note: WBGT - Wet Bulb Globe Temperature. Map 6.12: Expected Annual Population Exposure to High Heat Stress for Bangladeshi Upazilas Source: World Bank, original map developed for this report based on data from World Bank-GFDRR 2017 and WorldPop 2020. Note: WBGT - Wet Bulb Globe Temperature. 38 Since heat stress is felt throughout Bangladesh, and affects nearly all Bangladeshis to the same, highly frequent and highly intense extent, when the expected annual population exposure to this heat is overlaid with the socioeconomic vulnerability proxied by the 2016 poverty headcount rates in upazilas, a pattern emerges that reflects a combination of population density and poverty distribution across the country, both for the quantile classification (Map 6.13, panel a) and the predefined absolute exposure classes (Map 6.13, panel b). The resulting picture reveals that in the more densely populated upazilas of the northern Rangpur, Rajshahi, and Mymensingh Divisions, as well as in the very northeast of Dhaka Division and southwestern Chattogram Division, high rates of poverty coincide with a large population exposure to heatwaves annually. The communities in these areas are therefore most at risk of the negative health consequences brought about by extreme heat stress. In and around less deprived urban centers, including Dhaka and Chattogram, a high exposure to heat goes hand in hand with lower poverty rates. This suggests that these communities have a lower vulnerability since their socioeconomic condition might allow for improved adaptative and preventative measures. The exception to this dynamic of high poverty and high population density entailing high heat stress risk is eastern Chattogram Division, where the elevated terrain results in lower population exposure to heat. By consequence, heat risk in this area is lower, irrespective of the high poverty rates found in upazilas in this area. Map 6.13: Expected Annual Exposure of the Population to Heat Stress in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report using data from the 2016 HIES (BBS 2016-2017), World Bank- GFDRR 2017, and WorldPop 2020. 39 c. Drought risk Map 6.14, panels a and b show the percentage of years throughout the 37-year baseline period (1984- 2022) of the FAO drought data during which at least 30 percent of agricultural land in each grid cell experienced drought stress for the first and second cropping seasons (Kharif and Rabi, respectively). The hypothesis emerging from the literature review that drought is of particular concern during the Kharif cropping season is confirmed. The maps indicate that, especially in the northwest of Bangladesh, vast expanses of cropland are exposed to regular drought, occurring about once every four to five years. Particular zilas stand out, as shown in Map 6.15, panels a and b. During the first cropping season, the most frequent and intense droughts are found in the northwestern zilas of Rangpur (Dinajpur) and Rajshahi (Naogaon, Chapai Nababganj, Natore) Divisions, as well as Narail Zila in Khulna Division, and Bhola Zila in Barisal Division. Over 30 percent of cropland in these areas experience drought stress at least once every four years. Crop failures as a result if this drought could have severe impacts on food security. During the second cropping season, drought is much less of a concern. No upazila experiences drought stress more frequently than once every five years. Map 6.14: Drought Hazard for Agricultural Land across Bangladesh during the Kharif (panel a) and Rabi (panel b) Cropping Seasons Panel a Panel b Source: World Bank, original maps developed for this report using data from FAO 2022. 40 Map 6.15: Kharif (panel a) and Rabi (panel b) Season Drought Exposure for Agricultural Land (by Zila and Upazila) Panel a Panel b Source: World Bank, original maps developed for this report using data from FAO 2022. Since drought is of particular concern during the first cropping season (Kharif) in Bangladesh, overlaying the frequency with which at least 30 percent of cropland and pasture in an upazila is affected by agricultural drought during this season with the 2016 poverty figures, concerns arise once again around upazilas in Bangladesh’s northwest. The quantile map (Map 6.16, panel a) highlights that again upazilas in Rangpur, Rajshahi, and Mymensingh Divisions are some of the worst-affected areas, with northern Khulna and southwestern Chattogram also standing out negatively. In these upazilas, drought and poverty coincide, where previously it was also shown that riverine and coastal flooding, as well as heat, pose severe risks to people and assets. In these geographic areas, poor communities relying on agriculture for their livelihoods are particularly at risk of reduced crop harvest because of droughts, possibly triggering price spikes and food insecurity. They are thus priority areas for tackling the potential adverse impacts of drought on rural livelihoods. Working with predefined boundaries (Map 6.16, panel b) still highlights the same areas of concern but showcases that the effects are felt less than once every four years across the country. 41 Map 6.16: Expected Annual Exposure of Agricultural Land to Drought in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017) and FAO 2022. d. Landslide risk Whereas the plains around the Ganges, Brahmaputra, and Meghna riverine system tend to be more affected by floods, droughts, and extreme heat, mountainous areas tend to be more susceptible to landslides, as is evident from the World Bank-ARUP landslide hazard index for Bangladesh shown in Map 6.17. In the Chittagong Hill Tracts in Bangladesh’s southeast, rainfall-triggered landslide intensity is high. While less densely populated than the floodplains of Bangladesh’s main rivers, there are still a considerable number of people and built-up areas exposed to landslides. Map 6.18 shows the upazila- level population exposure to landslides, while Map 6.19 shows the exposure of built-up assets for these administrative units. There is considerable overlap between upazilas whose population and built-up assets are exposed to landslide hazards, with some small geographic nuances. In Sadar and Ramu Upazilas of Cox’s Bazar, over 100,000 residents live in landslide-prone areas. The highest infrastructural exposure is found in Ruma, Rowangchhari, and Bandarban Sadar Upazilas, with over 50 hectares of built-up assets lying in areas with high landslide intensity, threatening loss of life and damage to infrastructure. 42 Map 6.17: Landslide Hazard Index for Bangladesh at a 1km Resolution Source: World Bank, original map developed for this report using data from World Bank-ARUP 2022. Map 6.18: Population Exposure to High Landslide Risk at the Upazila Level Source: World Bank, original map developed for this report using data from World Bank-ARUP 2022 and WorldPop 2020. 43 Map 6.19: Built-Up Asset Exposure to High Landslide Risk Source: World Bank, original map developed for this report using data from World Bank-ARUP 2022 and WSF 2019. The quantile and predefined class landslide maps show a similar picture for both the annually exposed population (Map 6.20, panels a and b) and built-up assets (Map 6.21, panels a and b) in relation to the 2016 poverty headcount rates. The eastern upazilas of Chattogram Division are among the poorest in Bangladesh. The extremely high exposure of people and assets to intense landslide hazard then makes these same areas particularly at risk of the negative consequences of rainfall-triggered landslides. Efforts to improve the resilience and capacity of communities to adapt to landslide risks should therefore be focused on the upazilas in Bangladesh’s southeasternmost region. Map 6.20: Expected Annual Exposure of the Population to Rainfall-Induced Landslides in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b 44 Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), World Bank-ARUP 2022, and WorldPop 2020. Map 6.21: Expected Annual Exposure of Built-Up Assets to Rainfall-Induced Landslides in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), World Bank-ARUP 2022, and WSF 2019. e. Tropical cyclone risk Because of its flat topography, high wind speeds that accompany tropical cyclones are able to penetrate deep into the inland areas of Bangladesh, with only the eastern hills of Chattogram Division altering a nearly linear south-to-north pattern (Map 6.22). The largest EAI of damage to built-up assets is therefore unsurprisingly found in and around major urban centers closer to the coastline. Dhaka and western Chattogram Division stand out, as can be seen in the risk map (Map 6.23). 45 Map 6.22: Wind Speeds Reached across Bangladesh during Tropical Cyclones with a 100-Year Return Period Source: World Bank, original map developed for this report based on data from NOAA 2018. Map 6.23: Expected Annual Impact of Tropical Cyclones on Built-Up Assets across Bangladesh’s Upazilas Source: World Bank, original map developed for this report based on data from NOAA 2018 and WSF 2019. The quantile map showing the overlay of the expected annual damage to built-up assets of tropical cyclones with poverty rates (Map 6.24, panel a) suggests that the highest relative co-occurrence of exposure and poverty is found spread along Bangladesh’s coastline, but also as far inland as Rajshahi Division and upazilas in northern Dhaka Division. However, the map with absolute damage 46 thresholds from gusts of wind with poverty rates overlaid (Map 6.24, panel b) shows that the absolute impact is rather limited, with intermediate overlapping poverty rates and asset damage visible most strongly in northern Khulna and western Chattogram Divisions. The EAI is also pronounced in and around Dhaka, yet the lower poverty rates there suggest adaptative measures could be put in place more easily, thereby lowering the vulnerability—and hence the risk—of communities in these better- off urban environments. Map 6.24: Expected Annual Exposure of Built-Up Assets to Tropical Cyclones in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), NOAA 2018 and WSF 2019. f. Air pollution risk PM2.5 concentrations have exceeded the WHO norm of 37.5 µg/m3 maximum for a 24-hour mean on average every single year for which data are available across the whole of Bangladesh. By this metric, most of the country even breaks the 80 µg/m3 boundary (dark brown on Map 6.25 below), indicating just how severe the air pollution situation is in Bangladesh. This all-encompassing hazard predicts severe public health consequences, especially in densely populated areas where concentrations exceed health norms. Linking hazard to population exposure and adding the mortality impact function associated with high PM2.5 concentrations, as done in Map 6.26 for upazilas confirms this worrying hypothesis. In highly urbanized areas, Dhaka in particular, millions of citizens are at increased mortality risk because of air pollution. 47 Map 6.25: Mean Air Pollution (PM2.5 concentration) across Bangladesh, 1998-2019 Source: World Bank, original map developed for this report based on data from Van Donkelaar et al. 2021. 48 Map 6.26: Increased Mortality from PM2.5 Air Pollution, 1998-2019 Source: World Bank, original map developed for this report based on data from Van Donkelaar et al. 2021 and WorldPop 2020. A risk pattern similar to that seen for heat stress risk emerges for air pollution risk when the 2016 poverty rates are overlaid on the PM2.5 concentrations, both in the quantile (Map 6.27, panel a) and predefined class distribution (Map 6.27, panel b). Again, the more densely populated upazilas of northern Rangpur, Rajshahi, and Mymensingh Divisions, as well as of the very northeast of Dhaka Division and southwestern Chattogram Division show a possibly devastating combination of high poverty rates and the highest expected annual population impact on mortality from pollution. The communities in these areas are therefore most at risk of the negative health consequences brought about by high PM2.5 concentrations. Similar to what was seen for heat, the exception to the double burden of high pollution and high poverty/high population density is eastern Chattogram Division. The lower population density and slightly lower concentration of pollutants implies that risk in this area is lower, irrespective of the high poverty rates found in the upazilas there. 49 Map 6.27: Expected Annual Impact of Air Pollution on the Population in Bangladesh using Quantile (panel a) and Predefined (panel b) Cutoffs Panel a Panel b Source: World Bank, original maps developed for this report based on data from the 2016 HIES (BBS 2016-2017), Van Donkelaar et al. 2021, and WorldPop 2020. 7. The compounding effect of risk The overview of the population, built-up asset, and agricultural exposure to and impact of natural hazards in combination with the 2016 poverty rates at the upazila level shows the grave risks to communities and their assets arising from a co-occurring set of climate-related risks, causing disproportional damage and disrupting lives and livelihoods. Two aggregations of upazilas in Bangladesh face particularly challenging environmental conditions. Firstly, the western upazilas of Mymensingh and the eastern upazilas of Rangpur combine a high annual impact of riverine flooding with a high exposure to heat stress, agricultural drought, and air pollution. The high rates of poverty in these areas call for the explicit attention of policymakers and development partners in ensuring that the lives and livelihoods of these population groups are protected in the face of climate-related disasters through adaptative and preventative measures. A similar level of attention should be given to Chattogram Division, which should be disaggregated into southeastern and southwestern sections, since different dynamics are at play. Whereas the highly deprived southeast faces considerable landslide risk, the southwest is confronted with co-occurring issues of coastal floods, heat, drought, tropical cyclones, and air pollution, which require specific and tailored preventative and adaptative interventions and capacities. While these hotspots of concern show the areas and communities at highest risk today, it is important to explore whether poverty trends over time show a consistent, diverging pattern for more, or less, hazard-affected and exposed spatial units. To that end, we refer back to Map 2.3, which shows the 50 change in poverty headcount rates between 2010 and 2016. In addition, Map 7.1 shows for how many of the seven included hazards—riverine floods, coastal floods, heat stress, drought, tropical cyclones, landslides, and air pollution—each upazila falls into the highest decile of relative population or built- up asset exposure. Map 7.1: Co-Occurring Hazard Exposure for Bangladesh’s Upazilas Source: World Bank, original map developed for this report based on data from Fathom 2022, World Bank-ARUP 2022, World Bank-GFDRR 2017, FAO 2022, NOAA 2018, and Van Donkelaar et al. 2021. Comparing the 2010-2016 poverty trends with the co-occurring exposure to natural hazards captured in Map 7.1, a worrisome observation emerges: in the areas that have seen the starkest increase in poverty, natural hazards tend to have the highest, co-occurring impacts. Upazilas in eastern and southwestern Chattogram, southern Barisal, and across Rangpur and Rajshahi Divisions fall into the highest exposure decile for three or even four natural hazards, with a similar intensity and co- occurrence of exposure seen in southern Khulna and central Dhaka Divisions. This raises particular concern about the reduced ability of these impoverished communities to prevent or adapt to the impacts of climate change. This correlation between the compounding impact of natural hazards and poverty is emphasized by box plots showing the correlation between the 2010 and 2016 poverty rates and the number of hazards by which upazilas are severely affected (Figure 7.1), as well as the relationship between the change in poverty rates over time and these coinciding hazards (Figure 7.2). 51 Figure 7.1: 2010 and 2016 Poverty Rates and Co-Occurring Hazards for Hazard-Exposed Upazilas Source: World Bank, original figure developed for this report based on data from Fathom 2022, World Bank- ARUP 2022, World Bank-GFDRR 2017, FAO 2022, NOAA 2018, and Van Donkelaar et al. 2021. Figure 7.2: Change in Poverty Rates and Co-Occurring Natural Hazards for Hazard-Exposed Upazilas Source: World Bank, original figure developed for this report based on data from Fathom 2022, World Bank- ARUP 2022, World Bank-GFDRR 2017, FAO 2022, NOAA 2018, and Van Donkelaar et al. 2021. The figures highlight that upazilas with populations and built-up assets that are heavily exposed to three or four co-occurring natural hazards show no substantial decline in poverty rates between 2010 and 2016, whereas those with lower co-occurring exposure (0, 1, or 2 hazards for which they appear in the top 10 percent of the most exposed upazilas) show a mild or even stark decline in poverty rates over this period. While causation cannot be confirmed, this pattern suggests that climate change- related natural hazards could pose a barrier to lowering poverty rates, hampering the achievement of development outcomes, and lowering the resilience of communities. In addition, the findings might 52 also point to an out-migration of wealthier households, with poorer, more vulnerable households remaining behind in hazard-affected areas. Increasing resilience to natural hazards should therefore be a top priority in eastern and southwestern Chattogram, southern Barisal, Rangpur, and Rajshahi Divisions, where both poverty and natural hazards pose a double burden and show a worsening trend over time. 8. A look ahead: changing risks in the face of climate change a. Climate projections: the why and how Climate change is a major risk to the achievement of good development outcomes and the dual objectives of the World Bank Group (WBG 2013) to end extreme poverty and promote shared prosperity. Having established the present-day risks posed by the exposure of people, built-up assets, and agricultural land to a wide variety of natural hazards, a forward-looking perspective needs to be adopted to explore how climate risks are set to develop across space in the decades ahead. Forecasting the geographic risk patterns speaks to the objectives of the WBG’s 2019 Action Plan on Adaptation and Resilience, aiming to help countries shift from addressing adaptation as an incremental cost and isolated investment to systematically incorporating climate risks and opportunities at every phase of policy planning, investment design, implementation, and evaluation of development outcomes (WBG 2021a). The screening and assessment of future impacts of natural hazards from climate change typically involve the comparison of baseline conditions (observed or simulated) against future scenarios of climate variability. These baseline conditions are determined by computing the long-term average of a climate variable, setting a baseline historical value for a climate variable defined as the norm. Projections of this climate variable in the future will show anomalies, or deviations, of this variable relative to the historical baseline. The magnitude of these anomalies depends strongly on the future time horizon studied and the climate change scenario used. To set this baseline and model projected anomalies for the future, we obtained data from climate models released under the IPCC Sixth Assessment Report (AR) framework (IPCC 2021a). ARs are supported by coordinated climate modeling efforts referred to as Coupled Model Intercomparison Projects (CMIP). IPCC AR5, released in 2013, featured climate models from CMIP5. The 2021 AR6 features new CMIP6 models, which substantially expands the number of participating groups (49), climate models (100), and future scenarios examined (8). We drew on CMIP6 data for our future modeling, and took into account three climate change scenarios, called Shared Socioeconomic Pathways (SSPs)—previously called Representative Concentration Pathways or RCPs in CMIP5. These pathways cover the range of possible future scenarios of anthropogenic drivers of climate change by accounting for various future greenhouse gas emission trajectories, as well as a specific focus on carbon dioxide (CO2) concentration trajectories (IPCC 2021b). First, we included SSP1/RCP2.6, a scenario with low greenhouse gas emissions and CO2 emissions declining to net zero after 2050, followed by net negative CO2 emissions. Next, we looked at SSP2/RCP4.5, with intermediate greenhouse gas emissions and CO2 emissions remaining 53 near current levels until the middle of the century before dropping off. As a final scenario, we incorporated SSP5/RCP8.5, a very high greenhouse gas emission scenario with CO2 emissions that roughly double from current levels by 2050. As a time horizon, the period 2041-2060 was selected, allowing us to assess climate risks around the middle of the century. b. Modeled variables: projecting changes in precipitation and temperatures The three included climate scenarios each predict different spatial patterns, intensities, and frequencies of future natural hazards. This provides crucial information on which geographic areas are at the highest risk of climate-related disasters under a specific climate pathway. With a gridded resolution of about 100 kilometers, we were able to aggregate this modeled information at the level of Bangladesh’s divisions and pinpoint which divisions are likely to face an increase or reduction in the frequency and intensity of a specific natural hazard under specific SSP scenarios. Underlying these projections are several key climate variables connected to changing patterns of precipitation and temperature, summarized in Table 8.1. Looking at the number of consecutive wet days, days with rainfall over 10 millimeters, precipitation volume on extremely wet days, maximum five-day precipitation, and sea level rise allowed for the forecasting of likely changes in the risk of floods and rainfall-induced landslides across Bangladesh. We used the WBGT heat index to predict spatial changes in heat stress for the three climate scenarios. Finally, the number of consecutive dry days and the Standardized Precipitation-Evapotranspiration Index (SPEI) allowed for the assessment of projected changes in drought patterns for the country. No projections for air pollution or tropical cyclones can be made. Table 8.1: Climate Variables Underlying Climate Projections Hazard Associated Climate Indices Unit of Measurement Rainfall > 10 mm Days per year Consecutive wet days Days per year Floods and Maximum 5-day precipitation mm landslides Extremely wet days mm Sea level rise mm/year Annual SPEI Dimensionless Drought Consecutive dry days Days per year Heat stress WBGT heat index °C Source: World Bank, original table developed for this report. Note: SPEI – Standardized Precipitation-Evapotranspiration Index; WBGT – wet bulb globe temperature. c. Flooding and landslide projections Five variables underly the projection of changes in floods and rainfall-induced landslides: the modeled annual days of rainfall with over 10 millimeters of precipitation (Map 8.1), the annual number of consecutive wet days (Map 8.2), the maximum precipitation over five days (in millimeters, Map 8.3), the precipitation during extremely wet days (in millimeters, Map 8.4), and the projected annual average rise in sea levels (in centimeters/year, Map 8.5). 54 Each of these mapping ensembles has a similar structure. The gridded historical mean over the baseline period 1995-2014 is shown in the top left, with the historical division average over this period shown on the map at the bottom left. The second, third, and fourth columns represent the projected anomalies for the climate variables under SSP1 – 2.6 (second column), SSP2 – 4.5 (third column), and SSP5 – 8.5 (fourth column). The top row shows the gridded, standardized anomalies derived from CMIP6 for our time horizon of 2041-2060, while the bottom row shows the average standardized anomaly for each division in Bangladesh. Below the maps, for each climate variable, the historical variation during the baseline period is shown together with the projected future anomalies for the three SSPs. The historical baseline period shows a clear pattern of wet conditions across Bangladesh, particularly in the eastern and southern divisions, across the four precipitation-related variables. Rangpur, Mymensingh, and Sylhet Divisions face the most intense precipitation patterns, whereas Khulna Division has the overall lowest annual number of consecutive wet days, days with rainfall over 10 millimeters, extreme precipitation events, and volume of rainfall over a five-day period. In contrast, however, Khulna Division is particularly vulnerable to sea level rise. Map 8.1: Climate Change Projections – Rainfall over 10mm (days/year) for Bangladesh Source: World Bank, original maps developed for this report. Note: SD – standard deviation; ADM1 – First-level administrative unit, divisions. 55 Map 8.2: Climate Change Projections – Consecutive Wet Days (days/year) for Bangladesh Source: World Bank, original maps developed for this report. Note: SD – standard deviation; ADM1 – First-level administrative unit, divisions. Map 8.3: Climate Change Projections – Maximum 5-Day Precipitation (mm) for Bangladesh Source: World Bank, original maps developed for this report. Note: SD – standard deviation; ADM1 – First-level administrative unit, divisions. 56 Map 8.4: Climate Change Projections – Extremely Wet Day Precipitation (mm) for Bangladesh Source: World Bank, original maps developed for this report. Note: SD – standard deviation; ADM1 – First-level administrative unit, divisions. Map 8.5: Climate Change Projections – Average Sea Level Rise (cm) for Bangladesh Source: World Bank, original maps developed for this report. Studying the climate projections for the five precipitation-related variables, a rather more complex spatial pattern emerges. While patterns of cumulative rainfall over five days or consecutive wet days are likely to remain static or see a slight increase in standardized anomalies in the 2041-2060 period compared to the 1995-2014 baseline period, the number of days with rainfall over 10 millimeters and quantities of precipitation during extremely wet days show a strong upward trend. While the forecasts for days with rainfall over 10 millimeters are consistent across the three SSPs, projected anomalies in precipitation during extremely wet days are starker for the higher-emission future scenarios, in particular SSP5 – 8.5. 57 While no structural changes in overall precipitation patterns and conditions of wetness are thus expected for Bangladesh, the rainfall volumes during extremely wet days and the frequency with which such days occur are set to substantially increase. This is particularly the case in the northern divisions, which already face the most severe precipitation events, though increases are projected across the country throughout the 2041-2060 period. In the case of the higher greenhouse gas emission scenarios, these increases will be exacerbated. Along the coastline, the starkly growing anomalies of sea level rise raise tremendous concerns, in particular for Khulna Division, which will see the greatest increase in standardized sea level rise anomalies by mid-century—of over 40 centimeters—in any climate change scenario. The projected increase in extreme rainfall events, particularly in the northern divisions, in combination with strong sea level rise, is alarming, particularly as this can go hand in hand with more frequent and severe coastal and riverine flooding and rainfall-induced landslides. Thus, adaptative measures to prevent an increase in mortality and damage to built-up and agricultural assets will need to be taken, with particular attention to the north of the country and along the coastal floodplains. d. Drought projections Two variables underpin the projection of changes in drought patterns across Bangladesh: the number of consecutive dry days per year (Map 8.6), and the 12-month SPEI (Map 8.7). The SPEI has been found to be closely related to drought impacts on ecosystems, crop, and water resources, and has been designed to take into account both precipitation and potential evapotranspiration in determining drought (WBG 2022b). It is important to note that negative SPEI values indicate drier conditions. The mapping ensembles for these drought-related variables follow a similar design to the precipitation-related variables, again with the 1995-2014 period as a historical baseline. Both variables offer a complementary, yet slightly different, spatial picture. The SPEI shows little change in the standardized anomalies under any of the SSPs compared to the baseline. This suggests that little change is expected in agricultural drought patterns during the 2041-2060 period compared to the 1995-2014 baseline period. The climate projections therefore do not foresee a significant worsening—or improvement—of this spatial pattern of agricultural drought in the decades ahead. However, under all three SSPs, and in particular under the high-emission SSP (SSP5 – 8.5), the number of annual consecutive dry days is forecasted to increase mildly, as the standardized anomalies will increase above the historical baseline. However, this increase is limited, and uniform across Bangladesh. Areas with a historical pattern of a large number of consecutive dry days each year, mainly the western divisions, are thus set for a further increase in short-term dry episodes mid-century, particularly in the event of sharply rising greenhouse gas emissions. Elsewhere, standardized anomalies rise as well, yet this increase is comparatively smaller, suggesting the historical patterns are unlikely to shift significantly. This limited shift in projected drought patterns suggests that current agricultural drought concerns— closely related to food pricing and security—will remain valid in the decades ahead. 58 Map 8.6: Climate Change Projections – Consecutive Dry Days (days/year) for Bangladesh Source: World Bank, original maps developed for this report. Note: SD – standard deviation; ADM1 – First-level administrative unit, divisions. Map 8.7: Climate Change Projections – Standardized Precipitation-Evapotranspiration Index for Bangladesh Source: World Bank, original maps developed for this report. Note: ADM1 – First-level administrative unit, divisions. 59 e. Heat projections Projections for the change in heat patterns under the three climate change scenarios were determined by modeling the WBGT heat index (Map 8.8). Under all three SSPs, and relatively uniformly across Bangladesh, the standardized anomalies of heat are forecasted to increase relative to the 1990-2010 baseline period used for this indicator. Please note the slightly altered baseline period due to data availability. This increase is considerably greater under the higher-emission future scenarios, and particularly significant in historically cooler divisions such as Chattogram. The projected rise in temperatures is concerning, since it indicates increased heat stress across Bangladesh by the 2041-2060 period, potentially putting even more lives at risk. The fact that this increase is fairly homogenous across the entirety of the country’s territory raises particular questions on adaptation and prevention for urban environments, where heat has historically been a cause of significant climate risk. Map 8.8: Climate Change Projections – WBGT Heat Index (°C) for Bangladesh Source: World Bank, original maps developed for this report. Note: SD – standard deviation; ADM1 – First-level administrative unit, divisions. 9. Compounding challenges: the multiplicity of mid-century climate risks Combining this diverse set of climate change variables, and cutting across SSPs, a worrisome picture emerges for Bangladesh, in particular in relation to heat stress, riverine and coastal flooding, and landslides for the period 2041-2060. Current heat stress patterns are set to increase in severity, and while overall precipitation patterns are unlikely to considerably alter, extreme rainfall events are likely to increase in magnitude. The fact that we are only two decades away from these scenarios materializing calls for prompt and decisive adaptation and prevention, while mitigation of greenhouse 60 gas emissions stands out as equally important, as the adverse effects of climate change will deepen under worse emission conditions. 10. References Abedin, J., Y.W. Rabby, I. Hasan, and H. Akter. 2020. “An Investigation of the Characteristics, Causes, and Consequences of June 13, 2017, Landslides in Rangamati District Bangladesh.” Geoenvironmental Disasters 7(23). doi: 10.1186/s40677-020-00161-z. ADB (Asian Development Bank). 2021. Bangladesh Climate and Disaster Risk Atlas: Hazards—Volume I. Available at: https://www.adb.org/publications/bangladesh-climate-disaster-risk-atlas-volume-1. Last accessed: August 24, 2022. Ahmed, B. 2021. “The Root Causes of Landslide Vulnerability in Bangladesh.” Landslides volume 18: 1707–1720. Allen-Dumas, M.R., K.C. Binita, and C.I. Cunliff. 2019. Extreme Weather and Climate Vulnerabilities of the Electric Grid: A Summary of Environmental Sensitivity Quantification Methods. Oak Ridge National Library, ORNL/TM-2019/1252. AMS (American Meteorological Society). 2021. “Glossary of Meteorology - heat wave.” Available at: https://glossary.ametsoc.org/wiki/Heat_wave. Last accessed: August 24, 2022. AQLI. 2022. “Country Spotlight - Bangladesh.” Available: https://aqli.epic.uchicago.edu/country- spotlight/bangladesh/#:~:text=Some%20areas%20of%20Bangladesh%20fare,in%20the%20most% 20polluted%20district.&text=All%20of%20Bangladesh's%20161%20million,the%20country's%20o wn%20national%20standard. Last accessed: August 24, 2022. Associated Press. 2021. “Pakistan’s Lahore Becomes World's 3rd Most Polluted City.” Available at: https://www.washingtonpost.com/world/pakistans-lahore-becomes-worlds-3rd-most-polluted- city/2021/11/23/4bac278a-4c76-11ec-a7b8-9ed28bf23929_story.html. Last accessed: August 24, 2022 Aziz, M.A., A.B.M.Z. Hossain, M. Moniruzzaman, Rokib Ahmed, Taslima Zahan, Saiful Azim, Md. Abdul Qayum, Md. Abdullah Al Mamun, Md. Abdul Kader, and Niaz Md. Farhat Rahman. 2021. “Mapping of Agricultural Drought in Bangladesh using Geographic Information System (GIS).” Earth Systems and Environment. doi: 10.1007/s41748-021-00231-8. BBS (Bangladesh Bureau of Statistics). 2011. Population and Housing Census 2011. Dhaka. Available at: http://203.112.218.66/WebTestApplication/userfiles/Image/BBS/Socio_Economic.pdf. BBS. 2016-2017. Household Income and Expenditure Survey 2016-2017. Dhaka. Coffel, E.D., R.M. Horton, and A. de Sherbinin. 2017. “Temperature and Humidity Based Projections of a Rapid Rise in Global Heat Stress Exposure during the 21st Century.” Environmental Research Letters 13, number 1. Available at: https://iopscience.iop.org/article/10.1088/1748-9326/aaa00e. Last accessed: August 24, 2022. 61 Colorado Water Conservation Board. 2020. “What does expected annual damage mean?” Available at: https://storymaps.arcgis.com/stories/7878c89c592e4a78b45f03b4b696ccac. Last Accessed: August 23, 2022. Dasgupta, S. 2015. “Left unattended, 5.3 million of Bangladesh’s poor will be vulnerable to the effects of climate change in 2050.” World Bank Blogs. April 6, 2015. Available at: https://blogs.worldbank.org/developmenttalk/left-unattended-53-million-bangladesh-s-poor-will- be-vulnerable-effects-climate-change-2050. Last accessed: August 24, 2022. Eckstein, D., V. Künzel, and L. Schäfer. 2021. Global Climate Risk Index 2021. Available at: https://germanwatch.org/sites/default/files/Global%20Climate%20Risk%20Index%202021_1.pdf. Last accessed: August 24, 2022. ESA. 2020. “Worldwide Land Cover Mapping.” Available at: https://esa-worldcover.org/en. Last accessed: August 24, 2022. FAO (Food and Agricultural Association of the United Nations). 2022. “Earth Observation.” Available at: https://www.fao.org/giews/earthobservation/asis/index_1.jsp?lang=en. Last accessed: August 24, 2022. Fathom. 2022. “Fathom Global.” Available at: https://www.fathom.global/. Last accessed: August 24, 2022. Goodman, J., M. Hurwitz, J. Park, and J. Smith. 2018. “Heat and Learning.” NBER Working Series Paper 24639. IMF (International Monetary Fund). 2019. Climate Change Mitigation and Adaptation in Bangladesh: Policy Options. Available at: https://www.elibrary.imf.org/view/journals/002/2019/300/article-A001- en.xml. Last accessed: August 24, 2022. IPCC (Intergovernmental Panel on Climate Change). 2012. Determinants of Risk: Exposure and Vulnerability. Available at: https://www.ipcc.ch/site/assets/uploads/2018/03/SREX- Chap2_FINAL-1.pdf. Last accessed: August 24, 2022. IPCC. 2019. “Glossary Search.” Available at: https://apps.ipcc.ch/glossary/. Last accessed: August 24, 2022. IPCC. 2021a. Sixth Assessment Report. Available at: https://www.ipcc.ch/assessment-report/ar6/. Last accessed: August 24, 2022. IPCC. 2021b. Climate Change 2021 – The Physical Science Basis: Summary for Policymakers. Switzerland: Intergovernmental Panel on Climate Change. Mahmud, I., W. Raza, and R. Hossain. 2021. Bangladesh: Finding It Difficult to Keep Cool. Available at: https://openknowledge.worldbank.org/handle/10986/36534. Last accessed: August 24, 2022. McDuffie, E.E., R.V. Martin, J.V. Spadaro, R. Burnett, S.J. Smith, P. O’Rourke, M.S. Hammer, A. van Donkelaar, L. Bindle, V. Shah, L. Jaeglé, G. Luo, F. Yu, J.A. Adeniran, J. Lin, and M. Brauer. 2021. “Source Sector and Fuel Contributions to Ambient PM2.5 and Attributable Mortality across Multiple Spatial Scales.” Nature Communications 12(3594). 62 Mohsenipour, M., S. Shahid, E. Chung, and X. Wang. 2018. “Changing Pattern of Droughts during Cropping Seasons of Bangladesh.” Water Resources Management 32: 1555–1568. Mondol, M.A.H., X. Zhu, D. Dunkerley, and B.J. Henley. 2021. “Observed Meteorological Drought Trends in Bangladesh Identified with the Effective Drought Index (EDI).” Agricultural Water Management 255(107001). NASA ((US) National Aeronautics and Space Administration). 2020. “Intense Flooding in Bangladesh.” Available at: https://earthobservatory.nasa.gov/images/147057/intense-flooding-in- bangladesh. Last accessed: August 24, 2022. NOAA ((US) National Oceanic and Atmospheric Association). 2018. “NCDC International Best Track Archive for Climate Stewardship (IBTrACS) Project, Version 3.” Available at: https://www.ncei.noaa.gov/access/metadata/landing-page/bin/iso?id=gov.noaa.ncdc:C00834. Last accessed: August 24, 2022. NRDC. 2022. “The Particulars of PM 2.5.” Available at: https://www.nrdc.org/onearth/particulars- pm- 25#:~:text=PM%202.5%20comes%20primarily%20from,crazy%20central%20and%20southern%2 0California. Last accessed: August 24, 2022. Tufts University. 2016. “Swept Away? Mapping Flood-vulnerability in Bangladesh.” Available at: https://sites.tufts.edu/gis/files/2016/01/Sarkar_Aditya_DHP207_2016.pdf. Last accessed: August 24, 2022. UCAR. 2022. “Air Quality and Climate Change.” Available at: https://scied.ucar.edu/learning- zone/air-quality/air-quality-and-climate- change#:~:text=Because%20of%20climate%20warming%2C%20the,more%20often%20in%20hot %20temperatures. Last accessed: August 24, 2022. UCL. 2020. “Analysis: Bangladesh has Saved Thousands of Lives from a Devastating Cyclone – Here’s How.” Available at: https://www.ucl.ac.uk/news/2020/jun/analysis-bangladesh-has-saved- thousands-lives-devastating-cyclone-heres-how. Last accessed: August 24, 2022. UN. 2021. “Bangladesh: Deadly Flooding, Landslides Devastate Rohingya Refugees.” Available at: https://news.un.org/en/story/2021/07/1096622. Last accessed: August 24, 2022. UNOCHA. 2021. “Bangladesh: Floods and Landslides - Jul 2021.” Available at: https://reliefweb.int/disaster/fl-2021-000097-bgd. Last accessed: August 24, 2022. Van Donkelaar, A., M.S. Hammer, L. Bindle, M. Brauer, J.R. Brook, M.J. Garay, N.C. Hsu, O.V. Kalashnikova, R.A. Kahn, C. Lee, R.C. Levy, A. Lyapustin, A.M. Sayer, and R.V. Martin. 2021. “Surface PM2.5.” Available at : https://sites.wustl.edu/acag/datasets/surface-pm2-5/#V5.GL.02. Last accessed: August 24, 2022. WBG (World Bank Group). 2013. “Ending Extreme Poverty and Promoting Shared Prosperity.” Available at: https://www.worldbank.org/en/news/feature/2013/04/17/ending_extreme_poverty_and_promot ing_shared_prosperity. Last accessed: August 24, 2022. 63 WBG. 2018. “Clean and Resilient Growth in Bangladesh.” Available at: https://www.worldbank.org/en/news/feature/2018/09/16/clean-and-resilient-growth-in- bangladesh. Last accessed: August 24, 2022. WBG. 2019. Action Plan on Climate Change Adaptation and Resilience. Washington, DC: World Bank. WBG. 2021a. Climate Risk Country Profile – Bangladesh. Available at: https://climateknowledgeportal.worldbank.org/sites/default/files/country-profiles/15502- WB_Bangladesh%20Country%20Profile-WEB.pdf. Last accessed: August 24, 2022. WBG. 2021b. “Bangladesh - Current Climate > Climatology.” Available at: https://climateknowledgeportal.worldbank.org/country/bangladesh/climate-data- historical#:~:text=Bangladesh's%20historical%20climate%20has%20experienced,February)%20is% 20colder%20and%20drier. Last accessed: August 24, 2022. WBG. 2022a. “Poverty headcount ratio at national poverty lines (% of population) – Bangladesh.” Available at: https://data.worldbank.org/indicator/SI.POV.NAHC?end=2016&locations=BD&start=2000&vie w=chart. Last accessed: August 24, 2022. WBG. 2022b. “Metadata Glossary.” Available at: https://databank.worldbank.org/metadataglossary/environment-social-and-governance-(esg)- data/series/EN.CLC.SPEI.XD. Last accessed: August 24, 2022. WHO (World Health Organization). 2019. “Health Consequences of Air Pollution on Populations.” Available at: https://www.who.int/news/item/15-11-2019-what-are-health-consequences-of-air- pollution-on- populations#:~:text=Exposure%20to%20high%20levels%20of,people%20who%20are%20already %20ill. Last accessed: August 24, 2022. WMO (World Meteorological Organization). 2018. Flood Forecasting and Warning in Bangladesh. Available at: https://public.wmo.int/en/resources/bulletin/flood-forecasting-and-warning-bangladesh. Last accessed: August 24, 2022. WMO. 2020. “World’s Deadliest Tropical Cyclone was 50 Years Ago.” Available at: https://public.wmo.int/en/media/news/world%E2%80%99s-deadliest-tropical-cyclone-was-50- years-ago. Last accessed: August 24, 2022. World Bank-ARUP. 2022. “Assessing the Global Threat from Landslides.” Available at: https://www.arup.com/projects/global-landslide-hazard-map. Last accessed: August 24, 2022. World Bank-GFDRR. 2017. “EH-GLOBAL-VITO-20.” Available at: https://www.geonode- gfdrrlab.org/layers/hazard:intensity_returnperiod20y. Last accessed: August 24, 2022. WorldPop. 2020. “Population Counts.” Available at: https://hub.worldpop.org/project/categories?id=3. Last accessed: August 24, 2022. 64 WSF. 2019. “World Settlement Footprint - Where do Humans Live?.” Available at: https://www.dlr.de/blogs/en/all-blog-posts/world-settlement-footprint-where-do-humans- live.aspx. Last accessed: August 24, 2022. 65