Policy Research Working Paper 8922 Multi-Hazard Groundwater Risks to the Drinking Water Supply in Bangladesh Challenges to Achieving the Sustainable Development Goals Mohammad Shamsudduha George Joseph Sabrina S. Haque Mahfuzur R. Khan Anwar Zahid Kazi Matin U. Ahmed Water Global Practice June 2019 Policy Research Working Paper 8922 Abstract Groundwater currently provides 98 percent of all the drink- including access to drinking and irrigation water supplies ing water supply in Bangladesh. Groundwater is found and social vulnerability (that is, poverty), are overlaid on throughout Bangladesh but its quality (that is, arsenic these risk maps to estimate exposures. The multi-hazard and salinity contamination) and quantity (that is, water groundwater risk maps show that a considerable propor- storage depletion) vary across hydrological environments, tion of land area (5 to 24 percent under extremely high to posing unique challenges to certain geographical areas high risks) in Bangladesh is currently under combined risk and population groups. Yet, no national-scale, multi-haz- of arsenic and salinity contamination, and groundwater ard groundwater risk maps currently exist enabling water storage depletion. As few as 6.5 million (2.2 million poor) resource managers and policy makers to identify areas that to 24.4 million (8.6 million poor) people are exposed to are vulnerable to public health. This paper develops, for a combined risk of high arsenic, salinity, and groundwa- the first time, groundwater risk maps at the national scale ter storage depletion. The multi-hazard groundwater risk for Bangladesh that combine information on arsenic, salin- maps reveal areas and exposure of population groups to ity, and water storage, using geospatial techniques, linking water risks posed by arsenic and salinity contamination and hydrological indicators for water quality and quantity to depletion of water storage. construct risk maps. A range of socioeconomic variables, This paper is a product of the Water Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at gjoseph@worldbank.org, The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Multi-Hazard Groundwater Risks to the Drinking Water Supply in Bangladesh: Challenges to Achieving the Sustainable Development Goals Mohammad Shamsudduha1, George Joseph, Sabrina S. Haque, Mahfuzur R. Khan, Anwar Zahid, and Kazi Matin U. Ahmed JEL Classification: I39, Q53, Q56 Keywords: groundwater; water supply; risks to public health; global change; Bangladesh  1  Corresponding author; George Joseph, Global Water Practice, The World Bank, gjoseph@worldbank.org,  Mohammad Shamsudduha (m.shamsudduha@ucl.ac.uk,  1Institute for Risk and Disaster Reduction, University  College London, London WC1E 6BT, UK); Sabrina S. Haque (sabrina.sharmin.haque@emory.edu, Rollins School  of Public Health, Emory University, Atlanta, GA 30322, USA); Mahfuzur R. Khan  (m.khan@du.ac.bd,  Department of Geology, University of Dhaka, Dhaka 1000, Bangladesh); Anwar Zahid anwarzahidb@gmail.com,  Bangladesh Water Development Board, Green Road, Dhaka 1205, Bangladesh), and Kazi Matin U. Ahmed,  (kmahmed@du.ac.bd, Department of Geology, University of Dhaka, Dhaka 1000, Bangladesh)  This work was made possible by the Swedish International Development Cooperation Agency and benefited from  funding from the Government of Japan through the Japan Trust Fund for Scaling Up Nutrition. The findings,  interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the World Bank, its  Board of Executive Directors, or the governments they represent.    Multi-Hazard Groundwater Risks to the Drinking Water Supply in Bangladesh: Challenges to Achieving the Sustainable Development Goals Mohammad Shamsudduha2, George Joseph, Sabrina S. Haque Mahfuzur R. Khan, Anwar Zahid and Kazi Matin U. Ahmed 1. Introduction Groundwater is the largest store of freshwater that provides drinking, irrigation, and industrial water supplies globally (Taylor et al., 2013). In Bangladesh, approximately 32 km3 of groundwater is withdrawn annually of which 90% is used for irrigation, and 10% for domestic and industrial purposes combined (Figure 1) that is equivalent to ~4% of global groundwater withdrawal (Hanasaki et al., 2018). However, the sustainability of groundwater resources is threatened by hydrological and socioeconomic factors such as poor water quality, over- abstraction, inadequate governance, and impacts of changing climate that are not well understood. Effective management of groundwater resources is critical in meeting national and international agendas for improved public health, economic development, and poverty alleviation (Conti et al., 2016). About 98% of drinking and 80% of dry-season irrigation water supplies come from groundwater at shallow depths (<150 m below ground level, bgl) (Shamsudduha, 2018). However, widespread arsenic (As) contamination of shallow aquifers affects some 26% of all tubewells across Bangladesh (World Bank 2018), making it the largest mass poisoning of a population in history (Smith et al., 2000). Furthermore, a recent World Bank led study estimates that nearly 40% of all tubewells have dangerous detections of fecal bacteria (World                                                              2  Corresponding author; George Joseph, Global Water Practice, The World Bank, gjoseph@worldbank.org,  Mohammad Shamsudduha, m.shamsudduha@ucl.ac.uk,  1Institute for Risk and Disaster Reduction, University  College London, London WC1E 6BT, UK, , Sabrina S. Haque, sabrina.sharmin.haque@emory.edu, Rollins School  of Public Health, Emory University, Atlanta, GA 30322, USA, Mahfuzur R. Khan, m.khan@du.ac.bd, Department  of Geology, University of Dhaka, Dhaka 1000, Bangladesh,  Anwar Zahid, anwarzahidb@gmail.com, Bangladesh  Water Development Board, Green Road, Dhaka 1205, Bangladesh, and Kazi Matin U. Ahmed,  kmahmed@du.ac.bd, Department of Geology, University of Dhaka, Dhaka 1000, Bangladesh      This work was made possible by the Swedish International Development Cooperation Agency and benefited from  funding from the Government of Japan through the Japan Trust Fund for Scaling Up Nutrition. The findings,  interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the World Bank, its  Board of Executive Directors, or the governments they represent.    2    Bank 2018). In some areas of the coast, aquifers are also subject to increasing salinity intrusion (Ayers et al., 2016; Khan et al., 2014). Additionally, shallow groundwater levels in the northwestern, north central, and southwestern areas of the country are declining over time, particularly affecting public water supplies during the dry season when there is notable groundwater abstraction for irrigation (Shamsudduha et al., 2009; Shamsudduha et al., 2012). Nevertheless, Bangladesh’s progress in providing nearly universal access to “improved” drinking water supply can mostly be credited to its abundance of groundwater sources. According to the 2011 Bangladesh Census, there are 31 million households that have access to primarily groundwater supplies via tubewells, tapwater and other water wells such as dugwells (BBS, 2015). The current number of privately-owned tubewells is approximately 17 million, compared to a previous estimate of 6 million to 11 million in the early 1990s (BGS and DPHE, 2001). In addition to these private wells, an additional 1.6 million public tubewells have been installed by the Department of Public Health Engineering of Bangladesh in order to provide As-safe drinking water supply in rural areas of which about 20% are considered deep (>150 mbgl) wells (DPHE, 2016). Consequently, sourcing groundwater from deep aquifers, has become an effective and popular mitigation strategy for drinking-water supply over the last couple of decades for both As and even salinity water quality issues (Burgess et al., 2010; van Geen et al., 2016). However, there is concern of potential contamination of deep groundwater due to ingress of As and salinity from shallow depths (Shamsudduha et al., 2018). Contrary to these hydrological concerns, from ethical consideration (i.e., positive impacts outweigh the negative consequences of long-term exposure to arsenic), Ravenscroft and others (Ravenscroft et al., 2013) advocate that the development of deep groundwater in Bangladesh is needed. Although it may lead to unsustainable exploitation over the long-term, it will lessen the burden of crippling disease and death from As and salinity contamination of shallow groundwater and surface water while also benefiting future generations by improving public health, education and economic conditions. Assessment of groundwater risks in Bangladesh has been conducted in isolation. To date, integrated, multi-hazard groundwater risk mapping has not been conducted at the national scale to guide water resources officials and policy makers in Bangladesh. As the new era of the United Nation’s Sustainable Development Goals (SDGs) has begun, Bangladesh needs to look beyond improved access to ‘improved and safe drinking water supply’ in order to address water quality (SDG 6.3: improve water quality by reducing pollution) and water sustainability 3    (SDG 6.4: ensure sustainable withdrawals and supply of freshwater to address water scarcity). Strategies in meeting these integrated targets can benefit from geospatial analysis of layers of digital data on groundwater quality and quantity. Here, we develop, for the first time, multi-hazard groundwater risk maps for Bangladesh incorporating three hydrological indicators: groundwater arsenic, salinity and water storage. Our national-scale maps identify critical areas across the entire Bangladesh that may have both groundwater quality (arsenic and salinity contamination) and quality (water storage decline) issues – affecting access to drinking and irrigation water supplies. We have estimated exposure of population groups, particularly poor and socially vulnerable population, number of households with access to tubewells and piped water supplies, and the access to irrigation water supply to various levels of groundwater risks. These groundwater risk maps, and associated hydrological information and data will be made publicly available so that additional water- related indicators (e.g., E. coli, access to sanitation services) can be incorporated to the models in future to develop further these national-scale groundwater risk maps to estimate population exposure. 2. Materials and Methods Water-related multi-hazard groundwater risk mapping at the national scale of Bangladesh is conducted in this study that applies a number of hydrological indicators or variables such as (i) shallow groundwater arsenic (As) concentrations, (ii) groundwater salinity (i.e., Electrical Conductivity (EC) of shallow groundwater), and (iii) the mean depth (2004–2013) to dry- season (December–April) groundwater levels (GWD). A brief description of the hydrological data is provided here but a broader description can also be found in the supplementary text. 2.1 Arsenic concentrations in groundwater We use the groundwater arsenic (As) data set from the national-scale survey of groundwater in Bangladesh (National Hydrochemical Survey) that was conducted jointly by the British Geological Survey (BGS) and the Department of Public Health Engineering (DPHE) (BGS and DPHE, 2001). According to the national hydrochemical survey, nearly 90% of these surveyed wells have an intake depth of <150 mbgl and are considered shallow wells (see Supplementary Texts for further details). Approximately 25% of these shallow tubewells were drawing groundwater with As concentrations >50 µg/L (Bangladesh drinking water standard); nearly 4    42% of these wells were recorded to contain As concentrations >10 µg/L (WHO drinking water standard). Our analysis applies interpolated As concentrations (Figure 2a) for developing groundwater risk maps at the national scale in Bangladesh. Arsenic concentrations at geographic points (n=3,207) were interpolated using the Inverse Distance Weighting (IDW) algorithm and resampled at 2.5-km × 2.5-km grid resolution using the ‘raster’ package in the ‘R’ programming language (R Core Team, 2017). 2.2 Groundwater salinity: Electrical conductivity Elevated groundwater salinity is common at shallow depth (<150 mbgl) in coastal aquifers of southern Bangladesh and is generally defined by total dissolved solids (TDS), or Electrical Conductivity (EC) or chemical constituent such as chloride (Cl) (Zahid et al., 2013). There is no national-scale, frequent (e.g. monthly) monitoring of groundwater salinity in Bangladesh. Over the years, a few studies (see Supplementary Texts for further details) have generated some contour maps of groundwater EC at the national scale but there is little detail on the data sets used for mapping. Here, we have generated a groundwater EC map at the national scale using data primarily from two sources. First, we collated groundwater EC data from 461 newly- installed monitoring boreholes that were installed recently under a regional-scale hydrogeological study conducted in 19 coastal districts by the Bangladesh Water Development Board (BWDB) (BWDB, 2013). Second, we have digitized and georeferenced the contoured map of groundwater EC (Rahman and Ravenscroft, 2003) and extracted point data of EC at 102 locations throughout the country, predominantly in the northern part where there is limited groundwater EC measurement as groundwater is generally fresh. Finally, we interpolated the point data (n=563) at the national-scale using the IDW algorithm in ArcGIS environment. The interpolation error was 50 µS/cm compared to the data range of 27 to 43,950 µS/cm with a mean of 5,251 µS/cm. We then rasterized, using the ‘R’ programming language, the interpolated EC data at a grid resolution of 2.5-km × 2.5-km (Figure 2b). 2.3 Depth to dry-season groundwater levels We use mean depth to dry-season (December–April) groundwater levels as a hydrological indicator for representing groundwater storage depletion in Bangladesh. This is an important measure of sustainability of shallow groundwater abstraction in Bangladesh using low-lift pumps, for example, the No. 6 hand-operated pump, which is widely used for drinking and domestic purposes (see Supplementary Texts for further details). Mean depths to dry-season groundwater levels were estimated from weekly monitoring records from 236 boreholes from 5    2004 to 2013 (Figure 2c). These monitoring boreholes belong to a network of some 1,250 monitoring wells across the entire country that have been managed by the Bangladesh Water Development Board (BWDB) since the early 1960s. We estimated depth to mean dry-season groundwater levels (i.e., maximum depth below ground level) using the ground surface as a reference level. 2.4 Demography, access to water supply, and social vulnerabilities Demographic data sets on population, poverty, tubewells, and access to pipe water supplies in Bangladesh at the upazila level are collated from a GIS database (The Bangladesh Interactive Poverty Maps) published by the World Bank (2016). The country-level demographic database allows one to explore and visualize socioeconomic data at both Zila (district) and Upazila (sub- district) level. The online GIS-based mapping tool enables an easy access to different types of indicators including poverty, demographics of the population, children’s health and nutrition, education, employment, and household access to energy, water, and sanitation services (World Bank, 2016). These maps (see maps in supplementary Figure S1) were constructed by combining three different data sources all of which are publicly available: (i) 2010 Bangladesh Poverty Maps, (ii) 2011 Bangladesh Census of Population and Housing, and (iii) 2012 Undernutrition Maps of Bangladesh (BBS/WFP/IFAD, 2012). Children’s health and nutrition data sets were produced by the World Food Programme (WFP) and are constructed based on data from the Child and Mother Nutrition Survey of Bangladesh 2012 (MICS) and the Health and Morbidity Status Survey 2011 (HMSS). Upazila-level total population and percentage of poor population (i.e. percentage of the population that lives below the official national upper poverty line, which is based on household's poverty status assessed using per capita consumption) are shown in Figure S1. According to the 2011 National Population Census, conducted by the Bangladesh Bureau of Statistics, the total population of the country was 144 million with 76% of the total population living in rural areas in Bangladesh (see Supplementary Texts for further details). Upazila-level, percentage of households with access to tubewell and tapwater supplies in Bangladesh are shown in Figure S2. The data on access to water supply come from the 2011 Census of Population and Housing. The national average of households with an access to tubewell and tapwater supply (town or municipal water supply via piped network) is 82% and 6    10% respectively. Tapwater supply is limited to provincial towns and large metropolitan cities such as Dhaka, Chittagong, Rajshahi, Sylhet, Barisal and Rangpur (Figure 1). Access to groundwater-fed water supplies for irrigation is an important indicator for measuring food security in Bangladesh. Upazila-level, groundwater use for irrigation for the year of 2006– 07 Boro rice season is shown in Figure S2c. Groundwater irrigation has been estimated using reported information on irrigated area and the number of irrigation pumps surveyed under the minor irrigation campaign by the Bangladesh Agricultural Development Corporation (BADC). Additional information on irrigation requirement for dry-season rice cultivation under various soil types and their infiltration capacity (Ravenscroft, 2003) has been used to estimate groundwater irrigation. The social vulnerability is defined here by the lack of nutrition in children under the age of five. Four indicators measuring the level of under-nutrition in children: the number of underweight and severely underweight, and stunted and severely stunted children at the upazila level of Bangladesh (Figure S3) are used here to characterize social vulnerabilities. According to the 2012 Undernutrition Maps of Bangladesh, underweight and severely underweight children are those whose standardized weight-for-age is less than two and three standard deviations, respectively, below the median for the international reference following the WHO standard. Similarly, stunted and severely stunted children are those, whose standardized height- for-age is less than two and three standard deviations, respectively, below the median for the international reference following the WHO standard. 2.5 Groundwater risk mapping at the national scale We construct two multi-hazard groundwater risk maps based on the following selection criteria applying on groundwater arsenic (As), salinity (i.e. electrical conductivity, EC) and dry-season depth to shallow groundwater levels (GWD) at the national scale in Bangladesh. Mapped areas of groundwater risks represent four zones with specific ranges of groundwater As and salinity concentrations that are considered risks to public health, and depth to dry-season groundwater levels below or greater than a certain threshold (4-8 m bgl) that is considered unsustainable to low-lift pumping technologies and economically expensive. Cut-off values of hazard parameters such as arsenic, salinity and groundwater depletion are based on various drinking- water standards and lifting capacity of traditional pumping technologies used in Bangladesh. For instance, drinking water standards for arsenic are 10 µg/L and 50 µg/L according to the 7    World Health Organization (WHO) and Bangladesh Government guidelines, respectively. There are no Bangladesh standards for salinity in drinking water; however, according to the WHO the “palatability of water with a total dissolved solids (TDS) level of less than about 600 mg/L is generally considered to be good” (WHO, 2011). According to the secondary drinking water guidelines by the US Environmental Protection Agency’s (EPA, 2018), the drinking water standard for TDS is 500 mg/L that is equivalent to ~750 µS/cm of electrical conductivity (EC) considering a ratio of 0.65 between TDS and EC (Rusydi, 2018). Finally, the maximum lifting capacity of hand-operated pumps in Bangladesh is 7-8 m below ground level (bgl) (UNICEF, 2010). We consider two threshold-based multi-hazard risk modeling of groundwater into four categories: extremely high, high, medium and low risks. Groundwater risk map is based on the following criteria (Model 1) for four risk categories:  Extremely High Risk: As > 100 µg/L & EC > 1500 µS/cm & GWD > 8 m bgl  High Risk: As > 50 µg/L & EC > 750 µS/cm & GWD > 6 m bgl  Medium Risk: As > 10 µg/L & EC > 750 µS/cm & GWD > 5 m bgl  Low Risk: As > 10 µg/L & EC > 500 µS/cm & GWD > 4 m bgl Groundwater risk map is based on the following criteria (Model 2) for four risk categories:  Extremely High Risk: As > 100 µg/L & EC > 1500 µS/cm  High Risk: As > 50 µg/L & EC > 750 µS/cm  Medium Risk: As > 10 µg/L & EC > 750 µS/cm  Low Risk: As > 10 µg/L & EC > 500 µS/cm Groundwater risk maps are generated using the above sets of criteria in ‘R’ programming language (R Core Team, 2017). First, raster maps of spatial grid resolution of 2.5-km × 2.5-km for the three hydrological risk variables are generated using the Inverse Distance Weighting (IDW) algorithm in ESRI’s ArcGIS environment (software version 10.3.1). Raster data sets are imported into ‘R’ statistical software environment and the above sets of criteria are applied to individual grid points (total number of grids: 19,613 covering an area of ~148,000 km2). 3. Results 3.1 Groundwater risk mapping at the national scale 8    A composite groundwater risk map (Figure 3) in red, green and blue (RGB) colors represents a combination of three hydrological indicators such groundwater arsenic concentration (Green), salinity or groundwater EC (blue) and depth to dry-season groundwater levels (Red) in Bangladesh. The composite map clearly shows that the northwestern and northcentral Bangladesh is featured by deep dry-season groundwater levels whereas arsenic problem in shallow groundwater is mainly concentrated in southcentral region, and the southern, coastal region of Bangladesh has the highest level of groundwater salinity. The composite map clearly highlights the areas where more than one hydrological indicator or risk variable (i.e. high arsenic and high salinity; or high arsenic and deep groundwater levels; or a combination of all three elements) exists to a varying degree. 3.2 Groundwater risks and population exposure Two groundwater risk maps for Bangladesh using two sets of combination criteria (model 1: groundwater arsenic, salinity and storage; model 2: groundwater arsenic and salinity) are shown in Figure 4. Each groundwater risk map has four classes: extremely high risk, high risk, medium risk and low risk that are based on a set of criteria (models) described in the method section. In addition, groundwater risks posed by arsenic and salinity alone are mapped considering various concentrations of arsenic and electrical conductivity in water (see risk maps in supplementary Figures S7 and S8). In the groundwater risk map under model 1 (Figure 4a), the mapped area (208 km2; Table 1) within the zone of extremely high risk is small and located mainly in central Comilla district of the southeastern part of Bangladesh (Figure 4a). There are some 219,193 (0.15% of total population) people exposed to the extremely high risk of groundwater (Table 2). The high risk zone covers a large area of 7,146 km2 (5.3% of total land area) from southwestern to southeastern Bangladesh encompassing a number of districts such as Satkhira, Jessore, Narail, Fardipur, Dhaka, Munshiganj, Comilla, Feni and Chittagong where some 6.4 million (4.4%) people are exposed to groundwater risks. The medium risk zone covers a larger area of 10,350 km2 (7%) over most part of the Ganges-Brahmaputra-Meghna delta except for the exposed coastal region and about 9.5 million (6.6%) people are exposed to medium groundwater risks. Combined population of approximately 16 million (11%) are currently at variable risk (extremely high to medium risk) of elevated arsenic, salinity and depth to dry-season groundwater levels. Lastly, the mapped zone under low risk in model 1 covers an area of 28,123 9    km2 (20%) where 21 million (15%) people are currently at combined low risk of arsenic, salinity and groundwater storage depletion. In the groundwater risk map under the model 2 (Figure 4b), the mapped area within the zone of extremely high risk is substantial (18,297 km2; 13% of total land area) and it covers most parts of southeastern Bangladesh where nearly 13 million (9% of total population) people are exposed (Tables 1 and 2). The high-risk zone covers an area of 15,147 km2 (11%) and an estimated 12 million (8%) people are currently exposed. The medium-risk zone covers an area 13,160 km2 (9%) and an estimated 9 million (6%) people are currently exposed. Finally, an estimated land area of 18,797 km2 (13%) and 13 million (9%) people are located in an area where there is low groundwater risk due to relatively less contamination of arsenic and salinity. Considering arsenic and salinity individually, we find that approximately 33 million (23% of total population) people are currently at variable risk of elevated (>50 µg/L) arsenic; and nearly 26 million (18%) are exposed to very high (>1500 µS/cm) salinity in shallow groundwater. 3.3 Groundwater risks and access to drinking and irrigation water supplies In addition to demographics of the population, we estimate exposure of poor people and household access to water supplies within the mapped groundwater risk zones under both models. Results are summarized in Tables 1 and 2. Groundwater risk mapping reveals that 3 million to 6 million households with an access to tubewells are currently at risk of various degrees under model 1 and 2, respectively. Furthermore, an estimated 237,000 to 418,000 households with access to tapwater supply are at risk of arsenic and salinity contamination as well as depletion of shallow groundwater. However, it is noteworthy that tapwater supplies in towns and metropolitan cities often derive from deep groundwater that may not be at immediate risk to arsenic or salinity contamination. 3.4 Groundwater risks and socially vulnerable population Our analysis shows that poor people are exposed to groundwater risks at various degrees. According to the groundwater risk map (model 1), approximately 5 million poor people are currently exposed at various degrees of risks with a potential of another 6 million poor people being affected in future. The groundwater risk map based on model 2 shows a greater number of poor people (11 million) are currently exposed to shallow groundwater with elevated arsenic and salinity contamination. Our analysis shows that 9,740 (extremely high risk) to 229,192 (high risk) children are underweight, and 10,206 (extremely high risk) to 274,545 (high risk) 10    children are stunted under risk model 1 (Table 2). According to risk model 2, an estimated 475,340 (extremely high risk) to 921,537 (high risk) children are underweight, and 540,627 (extremely high risk) to 1,062,574 (high risk) children are stunted (Table 2). 4. Discussion Bangladesh has made significant progress in food security and public health in the last few decades. The contribution of groundwater resources to economic development, social well- being and public health has been paramount. Despite the importance of groundwater and its critical role in all sectors including health, no multi-hazard risk maps for Bangladesh currently exist. The links between adverse human health and long-term exposure to elevated arsenic (As) and salinity concentrations in drinking water is well-established in the country. Access to drinking and irrigation water supplies is also affected due to widespread depletion of shallow groundwater storage in various parts of Bangladesh threating food security. Here, for the first time, we develop multi-hazard groundwater risk maps for Bangladesh at the national scale by applying geospatial techniques and formulating a set of criteria involving both groundwater quality and storage indicators. We generate two groundwater risk maps combining multiple hydrological hazards such as elevated groundwater As, salinity and depleted dry-season groundwater levels – all of which pose a threat to public health through unsafe and unsustainable drinking water supply from shallow groundwaters. Furthermore, we test the validity of our national-scale groundwater risk maps using an exposure data set of As-affected population in Bangladesh. The arsenic-affected population data set provided by the Directorate General of Health Services in Bangladesh (DPHE/JICA, 2010) shows that some 18,672 (51%) of the total of 37,039 patients (i.e. primarily skin lesions) are located in areas where >80% wells are found to be As contaminated. Overlaying on our groundwater risk maps, we find that some 1,000 (3% of total patients) to 22,450 (61%) As- affected patients are located within the extremely high to high-risk zones (Figure 5). One of the limitations of our analysis of multi-hazard based groundwater risk mapping is that the maps represent an aggregated risk of static levels of arsenic, salinity and declining groundwater storage. We employ groundwater arsenic data from the National Hydrochemical Survey that was conducted in 1998-99 with the assumption that As concentration has limited seasonal and temporal variations (Bhattacharya et al., 2011; Dhar et al., 2008) and, at the basin scale, the 11    spatial pattern of background As levels has remained largely unchanged (BBS and UNICEF, 2010). Similarly, our groundwater risk maps represent an average level of salinity though it varies seasonally (Zahid et al., 2013). Our groundwater risk maps are based on multi-hazard indicators of various levels. Therefore, any area outside the mapped risk categories may or may not be free from a single hydrological hazard such as elevated arsenic, salinity or depletion of water storage. For example, an area with high salinity (>1500 µS/cm) but low arsenic (<10 µg/L) is not mapped as groundwater risk as it does not fulfill the set criteria of multi-hazard indicators but considered to be affected by individual hazards. Our groundwater risk maps update the current, rather loosely constrained, estimates of population exposure to elevated As and salinity concentrations in Bangladesh. Our groundwater risk mapping reveals that an estimated 33 (23% of total population) and 64 (44%) million people in Bangladesh are currently exposed to As concentrations greater than 50 and 10 µg/L (WHO standard), respectively in shallow (<150 mbgl) groundwater. Using the groundwater electrical conductivity (EC) map, we estimate that approximately 26 million (18%) people, predominantly in the coastal region of Bangladesh are exposed to considerably high salinity (EC >1500 µS/cm) in groundwater. Our estimate of population exposure to high salinity in groundwater is substantially greater than a previous estimate of 20 million (Khan et al., 2011) people in coastal Bangladesh. Furthermore, the analysis of population exposure to various levels of groundwater risks reveals that approximately 37 million (26%) to 47 million (33%) people are currently at risk of elevated arsenic, salinity and declined groundwater storage combined that covers 31% to 44% of the land area of Bangladesh. Given the proven public-health risk to elevated arsenic and salinity in drinking water and considerable threat to food security associated with depleted water storage, our national-scale multi-hazard groundwater risk maps have the potential to guide current practices and policies towards achieving the UN’s sustainable development goals in Bangladesh. 5. Conclusions We draw the following main conclusions from this first-ever, national-scale multi-hazard groundwater risk mapping for Bangladesh: 12    (1) Approximately 5% (risk model 1: arsenic, salinity water-storage depletion combined) to 24% (risk model 2: arsenic and salinity) land area in Bangladesh is exposed to extremely high to high risks of elevated arsenic, salinity and groundwater depletion hazards. (2) An estimated 4.5% (risk model 1) to 17% (risk model 2) of total population (144 million) of Bangladesh are exposed to extremely high to high risks of elevated arsenic, salinity and groundwater depletion hazards; 2.2 million to 8.6 million of these vulnerable populations are poor. (3) Nearly a million (risk model 1) to over 4 million (risk model 2) children, who are exposed to extremely high to high risks of multi-hazard groundwater risks are either underweight or stunted. (4) An estimated 32 km3 of groundwater is withdrawn annually in Bangladesh for irrigation (90%), and domestic and industrial (10%) use; this volume is equivalent to ~4% of global groundwater withdrawals. Furthermore, an estimated 2.3 million (risk model 1) to over 5.7 million (risk model 2) people, who depend on groundwater-fed irrigation are exposed to extremely high to high risks of multi-hazard groundwater risks. Finally, we have updated the existing, rather speculative estimates of population exposure to arsenic or salinity contamination individually. Our new estimation using the geospatial risk mapping suggests that nearly 33 million people are exposed to >50 µg/L (Bangladesh drinking- water standard) of arsenic concentrations, and nearly 26 million are exposed to considerably high (electrical conductivity value >1,500 µS/cm) salinity in groundwater. Contributors MS conceived the idea of groundwater risk mapping at the national scape in Bangladesh. MS also designed overall study and prepared groundwater arsenic, salinity and storage datasets with technical support from AZ. GJ and SSH provided extensive inputs to the preparation of the manuscript and commented on several versions of the text. KMA and MRK have provided arsenic-affected population exposure data. All authors read the manuscript and provided their comments on the manuscript. 13    Declaration of interests We declare no competing interests. Acknowledgments We acknowledge the kind support provided by the World Bank under a project “Bangladesh WASH (Water, Sanitation and Hygiene) and Poverty Diagnostic” for this work. Support from the Swedish International Development Cooperation Agency and the Japan Trust Fund is acknowledged. We acknowledge the Bangladesh Climate Change Trust (BCCT) for funding a regional-scale hydrological campaign in coastal region that was carried out by the Bangladesh Water Development Board (BWDB). Finally, we acknowledge the Director General of Health Services (DGHS) of Bangladesh for arsenic-affected population exposure dataset and the Bangladesh Agricultural Development Corporation (BADC) for irrigation dataset. 14    References Ayers, J.C., Goodbred, S., George, G., Fry, D., Benneyworth, L., Hornberger, G., Roy, K., Karim, M.R., Akter, F. (2016) Sources of salinity and arsenic in groundwater in southwest Bangladesh. Geochemical Transactions 17, 1-22. BBS, (2015) Population and housing census 2011, Socio-economic and demographic report, National Report, vol. 4. Bangladesh Bureau of Statistics (BBS) Dhaka. BBS and UNICEF, (2010) Multiple Indicator Cluster Survey (MICS) 2009, Volume I: Technical Report, Progotir Pathey. Bangladesh Bureau of Statistics (BBS) and UNICEF Bangladesh. BBS/WFP/IFAD, (2012) Undernutrition Maps of Bangladesh 2012. Bangladesh Bureau of Statistics (BBS), World Food Programme (WFP), International Fund for Agricultural Development (IFAD), Dhaka, Bangladesh, p. 24. BGS and DPHE, (2001) Arsenic Contamination of Groundwater in Bangladesh, Vol. 2. Final Report WC/00/19, in: Kinniburgh, D.G., Smedley, P.L. (Eds.), National Hydrochemical Survey. British Geological Survey (BGS) and Bangladesh Department of Public Health Engineering (DPHE), Keyworth. Bhattacharya, P., Hossain, M., Rahman, S.N., Robinson, C., Nath, B., Rahman, M., Islam, M.M., Brömssen, M.V., Ahmed, K.M., Jacks, G., Chowdhury, D., Rahman, M., Jakariya, M., Persson, L.Å., Vahter, M. (2011) Temporal and seasonal variability of arsenic in drinking water wells in Matlab, southeastern Bangladesh: A preliminary evaluation on the basis of a 4 year study. Journal of Environmental Science and Health, Part A 46, 1177-1184. Burgess, W.G., Hoque, M.A., Michael, H.A., Voss, C.I., Breit, G.N., Ahmed, K.M. (2010) Vulnerability of deep groundwater in the Bengal Aquifer System to contamination by arsenic. Nature Geoscience 3, 83-87. BWDB, (2013) Hydrogeological study and mathematical modelling to identify sites for installation of observation well nest, Final Report “Establishment of monitoring network and mathematical model study to assess salinity intrusion in groundwater in coastal area of Bangladesh due to climate change". Bangladesh Water Development Board (BWDB) and Institute of Water Modelling (IWM), Dhaka. Conti, K.I., Velis, M., Antoniou, A., Nijsten, G.-J., (2016) Groundwater in the Context of the Sustainable Development Goals: Fundamental Policy Considerations. International Groundwater Resources Assessment Centre (IGRAC), Delft. Dhar, R.K., Zheng, Y., Stute, M., van Geen, A., Cheng, Z., Shanewaz, M., Shamsudduha, M., Hoque, M.A., Rahman, M.W., Ahmed, K.M. (2008) Temporal variability of groundwater chemistry in shallow and deep aquifers of Araihazar, Bangladesh. Journal of Contaminant Hydrology 99, 97-111. DPHE, (2016) Circlewise watre source status and coverage - June 2016. Department of Public Health Engineering (DPHE), Dhaka, Bangladesh. DPHE/JICA, (2010) Situation Analysis of Arsenic Mitigation 2009. Department of Public Health Engineering, Japan International Cooperation Agency Bangladesh, Dhaka. EPA, (2018) 2018 Edition of the Drinking Water Standards and Health Advisories, EPA 822- F-18-001. U.S. Environmental Protection Agency, Washington, DC. 15    Hanasaki, N., Yoshikawa, S., Pokhrel, Y., Kanae, S. (2018) A global hydrological simulation to specify the sources of water used by humans. Hydrol. Earth Syst. Sci. 22, 789-817. Khan, A.E., Ireson, A., Kovats, S., Mojumder, S.K., Khusru, A., Rahman, A., Vineis, P. (2011) Drinking water salinity and maternal health in coastal Bangladesh: implications of climate change. Environmental Health Perspectives 119, 1328-1332. Khan, A.E., Scheelbeek, P.F.D., Shilpi, A.B., Chan, Q., Mojumder, S.K., Rahman, A., Haines, A., Vineis, P. (2014) Salinity in Drinking Water and the Risk of (Pre)Eclampsia and Gestational Hypertension in Coastal Bangladesh: A Case-Control Study. PLoS ONE 9, e108715. R Core Team, (2017) R: A language and environment for statistical computing (R version 3.4.3). R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R- project.org/. Rahman, A.A., Ravenscroft, P., (2003) Groundwater resources and development in Bangladesh. The University Press Limited, Dhaka, p. 446. Ravenscroft, P., (2003) Overview of the hydrogeology of Bangladesh, in: Rahman, A., Ravenscroft, P. (Eds.), Groundwater resources development in Bangladesh. The University Press, pp. 43-86. Ravenscroft, P., McArthur, J.M., Hoque, M.A. (2013) Stable groundwater quality in deep aquifers of Southern Bangladesh: The case against sustainable abstraction. Sci Total Environ. 454-455, 627-638. Rusydi, A.F. (2018) Correlation between conductivity and total dissolved solid in various type of water: A review. IOP Conf. Series: Earth and Environmental Science 118, 012019. Shamsudduha, M., (2018) Impacts of Human Development and Climate Change on Groundwater Resources in Bangladesh, in: Mukherjee, A. (Ed.), Groundwater of South Asia. Springer Hydrogeology, Singapore, pp. 523-544. Shamsudduha, M., Chandler, R.E., Taylor, R.G., Ahmed, K.M. (2009) Recent trends in groundwater levels in a highly seasonal hydrological system: the Ganges-Brahmaputra- Meghna Delta. Hydrol. Earth Syst. Sci. 13, 2373-2385. Shamsudduha, M., Taylor, R.G., Longuevergne, L. (2012) Monitoring groundwater storage changes in the highly seasonal humid tropics: validation of GRACE measurements in the Bengal Basin. Water Resour. Res. 48, W02508. Shamsudduha, M., Zahid, A., Burgess, W.G. (2018) Security of deep groundwater against arsenic contamination in the Bengal Aquifer System: a numerical modeling study in southeast Bangladesh. Sustainable Water Resources Management, doi.org/10.1007/s40899-40018-40275-z. Smith, A.H., Lingas, E.O., Rahman, M. (2000) Contamination of drinking water by arsenic in Bangladesh: a public health emergency. Bulletin of the World Health Organization 78, 1093-1103. Taylor, R.G., Scanlon, B., Doll, P., Rodell, M., van Beek, R., Wada, Y., Longuevergne, L., Leblanc, M., Famiglietti, J.S., Edmunds, M., Konikow, L., Green, T.R., Chen, J., Taniguchi, M., Bierkens, M.F.P., MacDonald, A., Fan, Y., Maxwell, R.M., Yechieli, Y., Gurdak, J.J., Allen, D.M., Shamsudduha, M., Hiscock, K., Yeh, P.J.F., Holman, I., Treidel, H. (2013) Ground water and climate change. Nature Climate Change 3, 322-329. 16    UNICEF, (2010) WASH Technology Information Packages for UNICEF WASH Programme and Supply Personnel. UNICEF and Skat - Swiss Resource Centre and Consultancies for Development, Copenhagen. van Geen, A., Ahmed, K.M., Ahmed, E.B., Choudhury, I., Mozumder, M.R., Bostick, B.C., Mailloux, B.J. (2016) Inequitable allocation of deep community wells for reducing arsenic exposure in Bangladesh. J Water Sanit. Hyg. Dev. 6, 142-150. WHO, (2011) Guidelines for Drinking-water Quality. World Health Organization, Geneva, Switzerland, p. 564. World Bank, (2016) Bangladesh Interactive Poverty Maps. The World Bank Group, Washington D.C. Zahid, A., Jahan, K., Ali, M.H., Ahmed, N., Islam, M.K., Rahman, A., (2013) Distribution of Groundwater Salinity and Its Seasonal Variability in the Coastal Aquifers of Bengal Delta, in: Zahid, A., Hassan, M.Q., Islam, R., Samad, Q.A., Khan, M.S., Haque, R. (Eds.), Impact of Climate Change on Socio-economic Conditions of Bangladesh. Alumni Association of German Universities in Bangladesh, German Academic Exchange Service (DAAD), Dhaka, pp. 170-193. 17    Main Figures Figure 1. Groundwater withdrawal (in million cubic meters) combined estimated irrigation and domestic usages in Bangladesh. Groundwater use for irrigation is estimated using the information on number of irrigation wells (dry season of 2015-16), both shallow and deep and the irrigated area. Domestic water use is estimated using the number of population in each district (2011 census data) and applying a daily rate of 50 liters per capita usage. 18    Figure 2. Groundwater hydrological parameters used to develop groundwater risk map in Bangladesh: (a) arsenic concentrations in shallow groundwaters, (b) groundwater salinity (i.e., electrical conductivity), and (c) depth to dry-season groundwater levels. Hydrological parameters are interpolated at the national scale. The additional gray-color filled line graphs on top and right margins represent mean values of interpolated grids along the columns and rows of the raster datasets of the interpolated parameters. 19    Figure 3. A national-scale RGB (red-green-blue) color composite map of the three groundwater hazards: storage decline as indicated by the depth to dry-season groundwater levels (red), arsenic concentrations in shallow groundwaters (green), and groundwater salinity hazard as indicated by electrical conductivity (blue). 20    Figure 4. Groundwater risk maps at the national scale in Bangladesh: (a) groundwater risk map based on model 1 (arsenic, salinity and dry-season groundwater levels); (b) groundwater risk map based on model 2 (arsenic and salinity only). Both maps show four zones: extremely high, high, medium and low risks to shallow groundwater based on a combination of two or three parameters above some threshold values described in the method. Maps of higher resolution are provided in the supplementary information. 21    Figure 5. Groundwater risk map and exposure of arsenic-affected population in Bangladesh. Upazila-wise number of arsenic-affected patients (i.e., cases of skin lesions) are mapped as solid colors and various groundwater risk zones (under model 1) are superimposed to show the spatial association between the two datasets. Upazilas without any reported arsenic- affected patients or data are left blanked (i.e., white). 22    Table 1. Summary statistics number of grids used in the groundwater risk mapping at the national-scale in Bangladesh; area covered under four risk zones under each risk model. Groundwater Risk Map (model 1: arsenic, salinity and groundwater storage) Proportion of 2 Cumulative Risk Zone No of grids Area (km ) cumulative area (km2) land area (%) Extremely high risk 27 208 208 0.15 High risk 955 7,146 7,354 5.26 Medium risk 2,299 10,350 17,704 12.66 Low risk 5,951 28,123 45,826 32.78 Groundwater Risk Map (model 2: arsenic and salinity only) Proportion of Cumulative Risk Zone No of grids Area (km2) cumulative area (km2) land area (%) Extremely high risk 2,376 18,297 18,297 13.09 High risk 4,343 15,147 33,444 23.92 Medium risk 6,052 13,160 46,604 33.33 Low risk 8,493 18,797 65,401 46.78 23    Table 2. Summary statistics of demographics of the population, poverty and social vulnerability levels, and household and irrigation access to water supplies in Bangladesh that are under various degrees of groundwater risks due to elevated arsenic and salinity contamination and depletion of groundwater storage. Statistics of the same variables are reported under the two groundwater risk models developed in this study. Note that figures under each risk class only refer to exposure within each risk zone and not cumulative. Groundwater Risk Map (model 1: arsenic, salinity and groundwater storage) Extremely Variables at risk High Risk Medium Risk Low Risk High Risk Total Population 219,193 6,361,209 9,453,381 21,291,790 (% of country total) (0.15%) (4.4%) (6.6%) (14.8%) Poor Population 86,640 2,101,933 2,677,818 6,443,670 Access to tubewells 41,232 1,252,477 1,763,131 4,141,321 (No of households) Access to tapwater 777 79,289 157,063 311,157 (No of households) No of people under 112,231 2,206,011 4,288,679 10,569,615 irrigation water risks No of underweight 9,740 229,192 547,423 1,374,227 children No of severely 2,438 53,563 127,925 327,032 underweight children No of stunted 10,206 274,545 659,847 1,629,770 children No of severely 5,735 155,620 373,739 939,453 stunted children Groundwater Risk Map (model 2: arsenic and salinity only) Extremely Variables at risk High Risk Medium Risk Low Risk High Risk Total Population 12,766,085 11,646,293 8,958,893 13,431,311 (% of country total) (8.9%) (8.1%) (6.2%) (9.3%) Poor Population 4,800,111 3,838,106 2,276,558 3,852,640 Access to tubewells 2,460,107 2,185,143 1,585,841 2,710,915 (No of households) Access to tapwater 88,852 122,535 206,509 209,565 (No of households) No of people under 1,623,303 4,049,400 5,278,533 10,834,711 irrigation water risks No of underweight 475,340 921,537 1,205,462 1,707,839 children No of severely 113,746 221,445 286,543 405,216 underweight children No of stunted 540,627 1,062,574 1,413,476 2,019,439 children No of severely 307,608 606,262 805,423 1,161,879 stunted children 24    Supplementary Information Supplementary Texts Arsenic concentrations in groundwater The first national-scale survey of groundwater in Bangladesh (National Hydrochemical Survey) was conducted jointly by the British Geological Survey (BGS) and the Department of Public Health Engineering (DPHE) of Bangladesh during a 1998–1999 campaign that sampled a total of 3,534 well points throughout Bangladesh except for some parts of Chittagong Hill districts and the Sundarbans.1 According to the national hydrochemical survey, nearly 90% of these surveyed wells have an intake depth of <150 mbgl and are considered shallow wells. Approximately 25% of these shallow tubewells were drawing groundwater with arsenic (As) concentrations >50 µg/L (Bangladesh drinking water standard); nearly 42% of these wells were recorded to contain As concentrations >10 µg/L (WHO drinking water standard). The distribution of observed As concentrations in Bangladesh is highly (positively) skewed, with values ranging from <0.5 to 1660 µg/L, and the spatial distribution is also highly variable.2-5 High As concentrations (>50 µg/L) are observed in most parts of southern Bangladesh at depths <100 mbgl1. There is no systematic, long-term monitoring of As concentrations in groundwater anywhere in Bangladesh. Patchy and short-term, time-series monitoring of groundwater As concentrations suggests that there is no discernible long-term trend in As in groundwater but seasonal variations in As concentrations are reported by some studies.6 Our analysis applies interpolated As concentrations (Figure 2a) for developing groundwater risk maps at the national scale in Bangladesh. Arsenic concentrations at geographic points (n=3207) were interpolated using the Inverse Distance Weighting (IDW) algorithm and resampled at 2.5-km × 2.5-km grid resolution using the ‘raster’ package in the ‘R’ programming language.7 Groundwater salinity: Electrical conductivity Elevated groundwater salinity is common at shallow depth (<150 mbgl) in coastal aquifers of southern Bangladesh and is generally defined by total dissolved solids (TDS), or Electrical Conductivity (EC) or chemical constituent such as chloride (Cl).8 The Electrical Conductivity (EC) of a solution (e.g., groundwater) is a measure of its ability to carry an electric current. In this study, groundwater salinity is expressed by EC, which has a unit of micro-Siemens per centimeter or µS/cm. 25    There is no national database for time-series monitoring of groundwater salinity in Bangladesh. Over the years, a few studies have generated some contour maps of groundwater EC at the national scale but there is little detail in the data sets used for mapping. One of these earlier maps on groundwater EC at the national scale was generated by the Master Plan Organization (MPO) under the Bangladesh National Water Plan Phase-I.9 Later on, Rahman and Ravenscroft 10 presented a groundwater EC map of shallow (<150 m bgl) groundwater in Bangladesh based on collated data on EC from various sources. Bangladesh Water Development Board (BWDB) has a monitoring network of some 118 stations where a set of chemical constituents of groundwater including chloride (Cl) concertation is monitored approximately once a year. Here, we have generated a groundwater EC map at the national scale using data primarily from two sources. First, we collated groundwater EC data from 461 boreholes that were installed recently under a regional-scale hydrogeological study conducted in 19 coastal districts by the Bangladesh Water Development Board (BWDB) between 2011 and 2013 funded by the Bangladesh Climate Change Trust.11 Secondly, we have digitized and georeferenced the contoured map of groundwater EC10 and extracted point data of EC at 102 locations throughout the country, predominantly in the northern part where there is limited groundwater EC measurements. Finally, we interpolated the point data (n=563) at the national-scale using the Inverse Distance Weighting (IDW) algorithm in ArcGIS environment. The interpolation error was 50 µS/cm compared to the data range of 27 to 43,950 µS/cm with a mean of 5,251 µS/cm. We then rasterized, using the ‘R’ programming language, the interpolated EC data at a grid resolution of 2.5-km × 2.5-km (Figure 2b). Depth to dry-season groundwater levels We use mean depth to dry-season (December–April) groundwater levels as a hydrological indicator for representing groundwater storage depletion in Bangladesh. This is an important measure of sustainability of shallow groundwater abstraction in Bangladesh using low-lift pumps, for example, the No. 6 hand-operated pump, which is widely used for drinking and domestic purposes. In the Indian Subcontinent, cast-iron made No. 6 suction pumps are traditionally used for the provision of drinking water and small-scale irrigation.12 More than 10 million tubewells, most of which are predominantly No. 6 pumps, are used to withdraw drinking water throughout Bangladesh13 and several other derivatives of this pump are available throughout Asia. When groundwater levels in the aquifer goes below 7-8 mbgl, which is the maximum suction limit of the No. 6 pump, these low-lift pumping technologies 26    are rendered unusable. Consequently, the domestic water supply in rural parts of Bangladesh is disrupted and the water supply becomes unsustainable. Mean depths to dry-season groundwater levels were estimated from weekly monitoring records from 236 boreholes from 2004 to 2013 (Figure 2c). These monitoring boreholes belong to a network of some 1,250 monitoring wells across the entire country that have been managed by the Bangladesh Water Development Board (BWDB) since the early 1960s. Depth from the well head to groundwater level at each station is referenced to a common horizontal datum known as the Public Works Datum (PWD), originally set approximately at the mean sea level (msl) with a vertical uncertainty of around ±0.45 m.14 We estimated depth to mean dry-season groundwater levels (i.e., maximum depth below ground level) using the ground surface as a reference level. The map of dry-season groundwater levels shows that the central, north-central, eastern and northwestern parts of Bangladesh have deepest water levels with mean depth acceding 15 mbgl that is twice more than the suction limit of hand pumps. Depths to water levels in much of the southern Bangladesh and along major river floodplains are generally shallower (<8 mbgl) than the maximum suction limit of the No. 6 pump. Demography, access to water supply, and social vulnerabilities Demographic datasets on population, poverty, tubewells, and access to pipe water supplies in Bangladesh at the upazila level are collated from a GIS database (The Bangladesh Interactive Poverty Maps) published by the World Bank.15 The country-level demographic database allows one to explore and visualize socioeconomic data at both Zila (district) and Upazila (sub-district) level. The online GIS-based mapping tool enables an easy access to different types of indicators including poverty, demographics of the population, children’s health and nutrition, education, employment, and household access to energy, water, and sanitation services.15 These maps (see maps in supplementary Figure S1) were constructed by combining three different data sources all of which are publicly available: (i) 2010 Bangladesh Poverty Maps, (ii) 2011 Bangladesh Census of Population and Housing, and (iii) 2012 Undernutrition Maps of Bangladesh. Upazila-level GIS data with a geographic projection (WGS 1984 Web Mercator coordinate system) have been used in this study. Upazila-level total population and percentage of poor population (i.e., percentage of the population that lives below the official national upper poverty line, which is based on household's poverty status assessed using per capita consumption) are shown in Figure S1. 27    According to the 2011 National Population Census, conducted by the Bangladesh Bureau of Statistics, the total population of the country was 144 million with 76% of the total population live in rural areas in Bangladesh. The spatial distribution of poor population shows that the north-central (Mymensingh to Rangpur), southwestern (Barisal to Khulna) and some parts of southeastern (Comilla) Bangladesh are generally poor. Upazila-level, percentage of households with access to tubewell and tapwater supplies in Bangladesh are shown in Figure S2. The data on access to water supply come from the 2011 Census of Population and Housing. The national average of households with an access to tubewell and tapwater supply (town or municipal water supply via piped network) is 82% and 10% respectively. Tapwater supply is limited to towns and large metropolitan cities such as Dhaka, Chittagong, Rajshahi, Sylhet, Barisal and Rangpur (Figure 1). The absence of tubewell-based water supply in these cities suggests that drinking water and domestic water come from municipal water supplies managed by the city authorities such as the Dhaka Water Supply and Sewerage Authority (DWASA) in Dhaka city. Access to groundwater-fed water supplies for irrigation is an important indicator for measuring food security in Bangladesh. Currently groundwater meets 80% of all irrigation water supplies and has been sustaining the dry-season “Boro” rice cultivation since the 1970s that has made the country self-sufficient in food production and has led to major economic development.16 Upazila-level, groundwater use for irrigation for the year of 2006–07 Boro rice season is shown in Figure S2c. Groundwater irrigation has been estimated using reported information on irrigated area and the number of irrigation pumps surveyed under the minor irrigation campaign by the Bangladesh Agricultural Development Corporation (BADC). Additional information on irrigation requirement for dry-season rice cultivation under various soil types and their infiltration capacity17 has been used to estimate groundwater irrigation. The social vulnerability is defined here by the lack of nutrition in children under the age of five. Four indicators measuring the level of under-nutrition in children: the number of underweight and severely underweight, and stunted and severely stunted children at the upazila level of Bangladesh (Figure S3) are used here to characterize social vulnerabilities. According to the 2012 Undernutrition Maps of Bangladesh, underweight and severely underweight children are those, whose standardized weight-for-age is more than two and three standard deviations, respectively, below the median for the international reference 28    following the WHO standard. Similarly, stunted and severely stunted children are those, whose standardized height-for-age is more than two and three standard deviations, respectively, below the median for the international reference following the WHO standard. The spatial distribution of both underweight and stunted children is generally higher in the northeastern and southeastern parts of Bangladesh except for hilly regions area where the population is the lowest in the country. 29    Supplementary References 1. BGS and DPHE. Arsenic Contamination of Groundwater in Bangladesh, Vol. 2. Final Report WC/00/19. Keyworth: British Geological Survey (BGS) and Bangladesh Department of Public Health Engineering (DPHE), 2001. 2. Gaus I, Kinniburgh DG, Talbot JC, Webster R. Geostatistical analysis of arsenic concentration in groundwater in Bangladesh using disjunctive kriging. Environ Geol 2003; 44: 939-48. 3. van Geen A, Zheng Y, Versteeg R, et al. Spatial variability of arsenic in 6000 tube wells in a 25 km2 area of Bangladesh. Wat Resour Res 2003; 39(5): 1140-55. 4. Shamsudduha M. Spatial Variability and Prediction Modeling of Groundwater Arsenic Distributions in the Shallowest Alluvial Aquifers in Bangladesh. J Spat Hydro 2007; 7(2): 33-46. 5. Yu WH, Harvey CM, Harvey CF. Arsenic in groundwater in Bangladesh: A geostatistical and epidemiological framework for evaluating health effects and potential remedies. Wat Resour Res 2003; 39: 1146. 6. Dhar RK, Zheng Y, Stute M, et al. Temporal variability of groundwater chemistry in shallow and deep aquifers of Araihazar, Bangladesh. J Contam Hydrol 2008; 99(1-4): 97-111. 7. R Core Team. R: A language and environment for statistical computing (R version 3.4.3). R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R- project.org/. 2017. 8. Zahid A, Jahan K, Ali MH, Ahmed N, Islam MK, Rahman A. Distribution of Groundwater Salinity and Its Seasonal Variability in the Coastal Aquifers of Bengal Delta. In: Zahid A, Hassan MQ, Islam R, Samad QA, Khan MS, Haque R, eds. Impact of Climate Change on Socio-economic Conditions of Bangladesh. Dhaka: Alumni Association of German Universities in Bangladesh, German Academic Exchange Service (DAAD); 2013: 170-93. 9. MPO. The groundwater resources and its availability for development, Technical report no. 5, Master Plan Organization (MPO), Ministry of Water Resources, GoB. Dhaka: Harza Engineering USA in association with Sir MacDonald and Partners, UK, Met Consultant, USA and EPC Ltd., 1987. 10. Rahman AA, Ravenscroft P, editors. Groundwater resources and development in Bangladesh. Dhaka: The University Press Limited; 2003. 11. BWDB. Hydrogeological study and mathematical modelling to identify sites for installation of observation well nest. Dhaka: Bangladesh Water Development Board (BWDB) and Institute of Water Modelling (IWM), 2013. 12. Baumann E. Handpump technologies: Low cost hand pumps, Field Note No 2011-3. Gallen, Switzerland: The Rural Water supply Network (RWSN), Skat Foundation, 2011. 30    13. Danert K. Sustainable Groundwater Development: Manual Drilling Compendium, Publication 2015-2. Gallen, Switzerland: The Rural Water supply Network (RWSN), SKAT Foundation, 2015. 14. Shamsudduha M, Chandler RE, Taylor RG, Ahmed KM. Recent trends in groundwater levels in a highly seasonal hydrological system: the Ganges-Brahmaputra-Meghna Delta. Hydrol Earth Syst Sci 2009; 13(12): 2373-85. 15. World Bank. Bangladesh Interactive Poverty Maps. Washington D.C.: The World Bank Group; 2016. 16. Shamsudduha M. Impacts of Human Development and Climate Change on Groundwater Resources in Bangladesh. In: Mukherjee A, ed. Groundwater of South Asia. Singapore: Springer Hydrogeology; 2018: 523-44. 17. Ravenscroft P. Overview of the hydrogeology of Bangladesh. In: Rahman A, Ravenscroft P, eds. Groundwater resources development in Bangladesh: The University Press; 2003: 43-86.   Supplementary Figures Figure S1. (a) Total population of Bangladesh at the sub-district or upazila level, and (b) percentage of poor population at upazila level in Bangladesh (data source: World Bank, 2016). 31    32    (c) Figure S2. (a) Access to drinking water supplies through tubewells; (b) access to tap water through piped water networks in Bangladesh (data source: World Bank, 2016); and (c) access to groundwater-fed irrigation (mean water use per upazila) through various pumping technologies (data source: Bangladesh Agricultural Development Corporation, 2007). 33    (a) (b) Figure S3. (a) The upazila-wise number of underweight, and (b) severely underweight children under five years of age across Bangladesh (data source: World Bank, 2016). The total number of children under the age of 5 years was estimated using data from the 2011 Census of Population and Housing of Bangladesh. 34    (a) (b) Figure S4. (a) The upazila-wise number of stunted, and (b) severely stunted children under five years of age across Bangladesh (data source: World Bank, 2016). The total number of children under the age of 5 years was estimated using data from the 2011 Census of Population and Housing of Bangladesh. 35    Figure S5. Groundwater risk maps at the national scale in Bangladesh based on model 1 (arsenic, salinity and dry-season groundwater levels). The map shows four zones: extremely high, high, medium and low risks to shallow groundwater based on a combination of two or three parameters above some threshold values described in the method. 36    Figure S6. Groundwater risk maps at the national scale in Bangladesh based on model 2 (arsenic and salinity only). The map shows four zones: extremely high, high, medium and low risks to shallow groundwater based on a combination of two or three parameters above some threshold values described in the method. 37    Figure S7. Groundwater risk maps at the national scale in Bangladesh featuring risks imposed by groundwater arsenic alone. The map shows four zones: extremely high, high, medium and low risks to shallow groundwater based on concentrations of arsenic (>200, >100, >50 and >10 µg/L, respectively) in shallow groundwater. 38    Figure S8. Groundwater risk maps at the national scale in Bangladesh featuring risks imposed by groundwater salinity (EC: electrical conductivity) alone. The map shows four zones: extremely high, high, medium and low risks to shallow groundwater based on values of EC (>2000, >1500, >750 and >500 µS/cm, respectively) in shallow groundwater. 39    40