One-fifth of the world's population at high risk from climate-related hazards

doi:10.21203/rs.3.rs-5302969/v1

Download PDF

Physical Sciences - Article

One-fifth of the world's population at high risk from climate-related hazards

https://doi.org/10.21203/rs.3.rs-5302969/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Climate change is increasing the frequency and intensity of extreme weather events. When these events occur, they threaten lives and livelihoods. Here we estimate the global population at high risk from climate-related hazards by examining household level vulnerability and local exposure to four types of events: droughts, floods, heatwaves and cyclones. Under current climate conditions, 4.5 billion people are expected to experience these hazards at intensities exceeding critical thresholds within their lifetime. One third of this population is considered highly vulnerable, based on seven dimensions that influence potential welfare impacts: income, education, access to finance, social protection, drinking water, electricity, and access to services and markets. Overall, we estimate that 1 in 5 people globally are at high risk from these hazards meaning that they are likely to experience them and would struggle to recover from their impacts. While the proportion of the population at high risk has nearly halved since 2010 due to decreased vulnerability, the number of people exposed has increased, and progress has been uneven across regions. This study demonstrates how household survey and high-resolution spatial data can be used to consistently track climate risks across countries and over time. The findings are relevant for policymakers working on climate change adaptation strategies.

Scientific community and society/Developing world

Scientific community and society/Social sciences/Climate change/Climate-change impacts

Scientific community and society/Social sciences/Economics

At any time, some households are experiencing welfare gains and moving out of poverty, whilst others are experiencing setbacks and falling into poverty¹. Poverty reduction, and development more broadly, requires both advancing welfare and protecting households from setbacks. Climate-related hazards are one major cause of setbacks increasing in frequency and intensity with climate change^2,3. Not only do they increase poverty when they occur, they cast a long shadow on welfare, as they can result in losses to assets, health and natural capital that limit welfare gains for many years to come. Vulnerable households are sometimes forced to adopt costly coping strategies, such as reducing spending on health and education, or selling productive assets, which can affect people’s – and especially children’s – prospects for decades⁴.

Sustainable development and poverty reduction requires reducing the number of people who are at risk from extreme weather events. Tracking global progress on this metric demands an indicator that can be measured consistently across countries and over time. Existing approaches to measure climate risk or vulnerability typically model aggregate economic losses or aggregate country level metrics in an index^5,6,7,8,9,10. Instead, we use a granular approach to count people at high risk¹¹, similar to what is done to monitor global poverty (the number of people living on less than $2.15 per day and the number of people multidimensionally poor). This new indicator is now being used by the World Bank as one measure of progress towards ending poverty on a livable planet [1]. It will be regularly updated and publicly available.

In this study, we present the indicator methodology and estimate the number of people at high risk from climate-related hazards globally around 2021, and its trend since 2010. We follow the traditional framework in which risk is the combination of hazard, exposure, and vulnerability¹². Hazard is the potential occurrence of an extreme event; exposure is the presence of people in places or settings that could be adversely affected; and vulnerability is the propensity or predisposition of these people to be adversely affected when an extreme event occurs.

Everybody faces some level of risk from extreme weather events, and everyone is vulnerable to some extent. To provide a meaningful metric, we consider thresholds that focus on exposure to severe events and high levels of vulnerability. We identify people at high risk from climate-related hazards as those who are both exposed and likely to suffer severe consequences (i.e., who are highly vulnerable). Figure 1 summarizes our approach. The exposed population is estimated using gridded spatial data derived from remote sensing and hazard modeling, whereas vulnerability is largely measured from household surveys (see Methods). The result is an estimate of the number of people likely to experience substantial welfare losses due to climate hazards in their lifetime.

Global exposure to extreme weather events

We examine exposure to four types of climate-related hazards: agricultural droughts, floods, heatwaves, and tropical cyclones. The exposed population is defined as those living in areas where events of a given return period, or frequency, exceed a critical intensity threshold (Table 1). As a benchmark, we consider 100-year return period events, which correspond with a 1% annual probability of occurrence, or 52% probability over the average life expectancy of 72 years. People are more likely than not to experience these events in their lifetime. Each type of hazard has distinct impacts, affecting lives, productivity, or assets in different ways. This makes it difficult to select equally severe exposure thresholds across different contexts. We identify critical thresholds for each type of hazard from existing literature.

Locations exposed to severe drought are based on FAO’s Agricultural Stress Index (ASI) and restricted to rural areas¹³. Agricultural production losses are expected when the vegetation health index falls below 35% over a growing season and this threshold is used by the ASI to define areas affected by severe drought¹⁴. Considering the challenges in modeling the probability of drought events, we define exposed locations as those where more than 30% of the land area experienced severe drought (ASI > 30) for any season over the last 39 years with observations¹⁵. We consider pluvial, fluvial and coastal flooding using modelled probabilistic inundation maps from Fathom and Deltares^16,17. Locations are considered exposed to severe flooding if the maximum inundation depth exceeds 0.5 m from any type of flooding. Floods of this depth are associated with disruptions to livelihoods and economic activities¹⁸. Estimates of the share of residential assets lost from a 0.5 m fluvial flood range between 22-49% across regions¹⁹. Exposure to severe tropical storms is based on the modelled STORM wind speed data for different return periods^20,21. We use a wind speed threshold corresponding with a Category 2 tropical cyclone on the Saffir-Simpson scale to identify exposed locations. Cyclones of this intensity are associated with damages to buildings and infrastructure^22,23. For heatwaves, we use a 5-day average maximum Wet-Bulb Globe Temperature (WBGT) of 33°C as the threshold. This corresponds with the reference upper limit for healthy, acclimatized humans at rest to keep a normal core temperature²⁴. At this temperature, heat-related mortality and hospital visits increase, and physical work capacity is reduced to around 50%^25,26. WBGT is approximated using the Environmental Stress Index (ESI), derived from hourly ERA5 reanalysis^27,28.

Table 1: Hazard exposure thresholds

Hazard	Return period	Intensity threshold defining an exposed location
Agricultural drought	Historical (39 years)	> 30% cropland or pasture affected and rural
Flood	100 years	> 0.5 m inundation depth
Heatwave	100 years	> 33°C 5-day maximum Wet-Bulb Globe Temperature
Tropical cyclone	100 years	≥ Category 2 wind speed

We estimate that 4.5 billion people, or 57% of the global population, are exposed to at least one hazard exceeding intensity thresholds. Figure 2a maps the share of population exposed to any hazard. By relying on spatial data with global coverage we can determine the exposure status of every person in every country, whereas when we consider vulnerability using household surveys, coverage is limited by data availability. The share exposed is higher in South Asia (87%) and East Asia and the Pacific (69%), but local hotspots exist within every region and country (Extended Data Table 1). The global population exposed to any hazard decreases to 47% when considering more frequent events, with a 20-year return period and the same intensity thresholds (Extended Data Table 2).

Our method assigns every population grid cell (approximately 90 x 90 m) to one of 16 exposure categories representing all possible combinations of hazards. This prevents double counting and permits analysis of exposure to multiple hazard types, as shown in Figure 2b. Heatwaves are the hazard with highest share of the population exposed (40%) and dominate in regions closer to the equator (Extended Data Figure 1). Around three quarters of those exposed to cyclones (8% of the total population) and more than half of those exposed to floods (10%) are also exposed to heatwaves, influenced by tropical weather patterns, topography and urbanization in these locations. A substantial share of the population is exposed to severe droughts (18%), considering this only counts people in rural areas, of which more than a third are exposed to another type of hazard. Overall, 16% of the population are exposed to more than one above-threshold hazard but only 1.6% are exposed to more than two. Figure 2b does not visualize very small intersections such as the population exposed to all four hazards (0.04%), but there are some people exposed to all possible hazard combinations.

We also classify the population by degree of urbanization, which is necessary to merge exposure with household data on vulnerability specific to the rural and urban population. Almost three quarters of the rural population, defined by the degree of urbanization methodology²⁹, are exposed to severe drought. However, a relatively larger share of the urban population is exposed to cyclones, floods and heatwaves (Extended Data Table 3). Exposure to flooding is highest in peri-urban and suburban areas on the outskirts of cities (12% of the population), where formal land-use planning may be less common. Results highlight varying rates of population exposure across urbanization levels and the need for tailored risk management strategies.

Vulnerability to extreme weather events

To identify vulnerable households, we use seven indicators that proxy different dimensions of vulnerability (Figure 1). These are informed by a large literature identifying the type of households with a high probability of experiencing severe losses when exposed to extreme weather events. They capture both household’s physical propensity to experience loss of income, assets or health and their inability to cope with and recover from losses. A household is considered vulnerable if they are highly vulnerable on any dimension. This approach gives an equal importance to each dimension. In reality, certain dimensions may be more critical depending on the context or type of hazard, and the interaction between dimensions may be important. Our approach does not currently incorporate this complexity; however, we demonstrate the potential for such extensions by mapping the complete joint distribution of hazard exposure and vulnerability dimensions (Extended Data Figure 4).

The specific measures of vulnerability were selected because they have well-recognized global definitions and are widely available in survey data. The indicators are mostly derived from household surveys that are representative of the population at subnational rural-urban level, allowing global scale analysis of the population at risk with unprecedented granularity. Most indicators are derived from harmonized household survey microdata available from the World Bank’s Global Monitoring Database (GMD), while others are fused into the GMD household surveys using data from other sources including the Findex individual-level microdata for access to finance and ASPIRE for social protection (see Methods and the replication package for details)^30,31.

Physical propensity to experience severe loss is proxied using three indicators. Households are considered highly vulnerable if they lack access to improved drinking water. Improved sources of water can protect people from contamination when storms and floods occur or lessen the impacts of droughts and heatwaves³². Households are also considered highly vulnerable if they lack access to electricity. During heatwaves, households with electricity are much more likely to have assets such as fans and fridges that can alleviate impacts. Lastly, access to services and markets enhances resilience by providing access to healthcare and other support, and ensuring households can access alternate employment opportunities and markets for goods (including food when local production fails)³³. Households are considered highly vulnerable if they are rural and located more than 2 km from an all-season road, consistent with the Rural Access Index (SDG 9.1.1).

Inability to cope with losses is proxied using four indicators. First, households are considered highly vulnerable if they live below the international poverty line of $2.15 (2017 PPP) per person per day. Should a hazard occur, they would be unable to meet basic needs Second, households are considered highly vulnerable if no adult member has completed primary education. Educational attainment captures both the ability to understand and respond to information such as early warnings, as well as the ability to switch livelihoods when faced with climate-related hazards^34,35,36,37. Third, households are considered highly vulnerable if they do not have access to a bank or mobile money account. Borrowing money or accessing transfers from family members is a widespread and effective coping strategy in the aftermath of a disaster^38,39,40. Access to a bank or mobile money account increases the geographical reach of these transfers so that they can be used to manage climate shocks. Fourth, there is considerable evidence that public cash transfers and other social protection systems help households manage shocks^{41,42,43,44,45,46,47}. Ideally, to measure this dimension we would use data on whether a household can be covered by social protection should a crisis occur. The number of beneficiaries at one point in time may be a poor proxy for this potential coverage, especially with the development of adaptive social protection systems that can scale up in times of crisis. To better account for the effect of social protection systems, we assume that individuals contributing to or receiving social insurance and those currently receiving social protection transfers are not highly vulnerable^48,49,50.

The share of the population highly vulnerable on each dimension in 2021 ranges from 2% (access to services and markets) to 19% (access to social protection), based on data for 160 countries accounting for 94% of the global population (Extended Data Table 4). 14% of people lack access to finance and 6-10% are considered highly vulnerable on the remaining dimensions. The share of population vulnerable on any dimension is 36% in our sample. Figure 3a maps the stark differences in vulnerability across the globe. Sub-Saharan Africa has 90% of its population highly vulnerable on at least one dimension, whereas less than 3% are highly vulnerable in high income countries. The dimension on which the largest share of the population is vulnerable varies substantially across countries (Extended Data Figure 2).

While one could expect income to be a good proxy for other dimensions of vulnerability, the correlation across dimensions is not particularly high (Extended Data Figure 3). This is underscored by the fact that although two-fifths of those highly vulnerable are vulnerable on multiple dimensions (16% of the global population), three fifths are vulnerable on one dimension (Figure 3b). This emphasizes the different nature of vulnerability across households, and the complexity of resilience-enhancing policies, which require different interventions in different places, even within countries. Lack of access to social protection and lack of financial inclusion are the dimensions on which people that are not vulnerable on other dimensions are most likely to be vulnerable (Figure 3c).

The global population at high-risk

We estimate the population at risk from climate-related hazards by merging the population exposure estimates derived from high resolution spatial data with information on household vulnerability from surveys. People at high risk are defined as those exposed to any hazard and vulnerable on any dimension, based on specific thresholds (Figure 1). We find that almost 1 in 5 people (19%) are at high risk from climate-related hazards based on data from 160 countries for 2021 (Extended Data Table 5).

Figure 4a maps the share of population at high risk in 2021, highlighting spatial inequalities arising from the intersection of exposure and vulnerability. Nearly everyone who is exposed in Sub-Saharan Africa is considered highly vulnerable and therefore at high risk. Sub-Saharan Africa had the highest vulnerability rates for 2021 (8-71%) on each dimension except access to finance (30%), which is also high in the Middle East and North Africa (39%). In contrast, Asia has higher rates of population exposure on average, but lower vulnerability. Population exposure is similar in North America and Latin America and the Caribbean at the region level, but there is a considerable difference in the share at high risk because very few are counted as highly vulnerable in North America when using our measure, which is extreme by high-income country standards. There are 37 countries, many of them small islands, where the entire population is exposed to cyclones or heatwaves exceeding intensity thresholds, but among these the population at high-risk ranges from less than 1% to over 90% (Haiti) due to differences in vulnerability. The map also highlights significant variation within countries with subnational data.

8% of the population are exposed and vulnerable on more than one dimension. People in sub-Saharan Africa and those exposed to drought are particularly likely to be highly vulnerable on multiple dimensions. Of those exposed to drought, three fifths are highly vulnerable on multiple dimensions, while this is closer to one third for people exposed to other hazards (Figure 4b). These exposed populations facing multiple deprivations are more likely to require multiple interventions to become more resilient. We further decompose results to quantify the population exposed to each combination of hazards and facing each combination of vulnerabilities (Extended Data Figure 4). This detail is useful for informing context sensitive and multifaceted climate change adaptation strategies.

Historical trends

We estimate recent trends in those at-risk by examining how the number of people exposed and vulnerable have changed between 2010 and 2021. Given the short time trend we assume constant climate conditions between 2010 and 2021. Two dimensions of vulnerability (access to social protection and access to services and markets) are also assumed constant across time, due to lack of available time trend data. Access to social protection, services and markets has improved over the past decade, and additional work is ongoing to improve upon the time trend estimation.

With these simplifications, the share of the global population at risk to extreme weather events has decreased substantially from over 1 in 3 people in 2010 (2.5 billion) to 1 in 5 in 2021(1.5 billion) (Figure 5a). The trend can be attributed almost entirely to a decrease in severe levels of vulnerability among households in exposed locations. The share of people living in exposed locations did not change significantly at the global level as a result of net-migration and differences in population growth between exposed and unexposed places, however the absolute number of people exposed increased by 461 million. Trends in exposure for specific locations and hazards vary, for instance cities are growing faster in flood zones than in safe areas⁵¹.

Globally, the largest decrease in vulnerability was on the access to finance dimension. The share of the exposed population without a bank or mobile money account decreased from 21% in 2010 to 8% in 2021, highlighting rapid improvements in financial inclusion over the last decade (Extended Data Table 5). The share of exposed people vulnerable on income, education and access to infrastructure dimensions have each decreased by around half. The decrease in the population at high-risk over the last decade has been driven by development in some highly populated and exposed regions, with seemingly limited progress occurring in others. South Asia and East Asia & Pacific regions saw the largest decreases in their populations at high-risk, dropping from 66 to 32 percent and 32 to 11 percent respectively, due to significant improvements across most vulnerability dimensions (Figure 5b).

[1] https://scorecard.worldbank.org

Assessing climate risk is complex and data intensive, requiring information on hazards, the people and assets exposed, and the extent that those exposed would be adversely affected. A comprehensive risk assessment would ideally integrate probabilistic models to estimate welfare losses, beyond just monetary impacts. This exists for some losses from some types of events.6 However, the modelling involved is highly uncertain and remains challenging to communicate to a broad audience. The indicator proposed here instead follows a simplified approach, counting the number of people at high risk based on exposure thresholds and proxies for vulnerability. While it does not replace more sophisticated models, it serves as a complementary tool, offering a practical alternative for assessing climate risks, especially in data-limited contexts.

The indicator we present complements existing global risk and vulnerability indices such as the Global Climate Risk Index, the UN Multidimensional Vulnerability Index, INFORM, and ND GAIN by providing a more granular, people-focused metric. Compared to existing indices, our approach offers more spatially granular insights and integrates often overlooked dimensions of vulnerability by merging high resolution spatial and household survey data. This provides a more nuanced understanding of vulnerability in the population exposed to climate shocks within countries. By counting people at risk, the indicator is a good complement to alternatives that measure assets at risk, which tends to weight the potential losses to wealthier people more highly, and is easy to communicate to a general audience and non-experts. However, challenges with data availability and aggregation remain.

Nonetheless, several limitations exist. First, our approach only counts people exposed to severe weather events, such as intense cyclones or heatwaves. It does not account for low-intensity, high-frequency events that, while individually less severe, may cause significant cumulative damage over time. In addition, we focus solely on four types of climate-related hazards, excluding other environmental factors like air pollution. This means the numbers presented do not fully capture all climate-related risks to welfare. We consider the probability of hazards in the current climate, but do not project exposure to hazards in a future climate scenario. This can be done using our methodology and will be part of future work.

Our analysis focuses on direct exposure to climate-related events, considering who is in the path of an event, rather than who may experience indirect effects such as disruption to markets, price fluctuations, or the consequences of damage to infrastructure. Moreover, the calculation method does not consider the cascading effects that often follow disasters, like propagation of impacts through supply chains, migrations from affected areas, or the outbreak of diseases that can sometimes occur in the aftermath of such events. While this means that our results underestimate the total number of people at risk, they remain a reliable proxy, as indirect effects stem from direct population exposure.

In measuring vulnerability, our approach focuses on extreme levels of vulnerability, acknowledging that vulnerability can be defined differently in different contexts and in countries at different income levels, and that even people who are not considered at high risk in our methodology can be severely affected by an extreme event. Our method assigns equal importance to each vulnerability dimension, such as access to infrastructure or financial inclusion. However, the importance of each dimension can vary depending on the context and specific hazard. There are also important dimensions of vulnerability for which it was not possible to find an indicator with significant global coverage to use in this analysis. Access to health services and measures of social or political exclusion are two key dimensions of vulnerability that were not included for this reason. While the former could be addressed through data investments, the latter is more challenging to measure consistently across contexts.

Finally, data limitations remain a challenge even for the dimensions of vulnerability included. We identify three main challenges and areas for future work. First, our approach requires data on multiple dimensions for the same household. Since data on all dimensions is not available from the same household survey, indicators from different sources must be fused together. Work to further develop data fusion methods is warranted to better capture the overlap between vulnerability dimensions. Ongoing efforts to improve and standardize household surveys are also critical so that indicators measuring different dimensions of vulnerability are collected in the same survey.

Second, vulnerability data are generally available at the subnational level, a much larger geographical area than the grid-level exposure estimates. When overlaying the two, we assume that the share of highly vulnerable households is the same in exposed locations as in areas not exposed to hazards. This assumption is more plausible for smaller areas with similar households but may not hold for larger regions, especially in heterogenous urban areas. To assess the size of potential bias from making this assumption, we estimate upper and lower bounds by instead assuming maximum and minimum overlap between exposed and vulnerable populations within subnational areas. The difference between estimates at these bounds was less than three percentage points in subnational areas accounting for most of the global population. However, the potential bias is larger when surveys represent populations dispersed over large regions, and more spatially disaggregated data is needed to reduce this uncertainty (see Methods and Supplementary Information)

Third, in many cases key data – such as coverage by social protection systems or intra-household disparities – are either incomplete or unavailable for all countries. For instance, data on access to basic water services is not consistently included in surveys, and information on social protection beneficiaries is not available for all countries and often relies on modeled estimates rather than direct reporting. In many cases data on the different dimensions of vulnerability are not up to date, with information from different years often combined, which can lead to inconsistencies. Continued investments are needed in household survey data to ensure this measure becomes more accurate over time.

We follow three steps to identify those at high risk from climate-related hazards. First, the number of people living in locations exposed to extreme weather events is calculated by overlaying hazard maps and gridded population data. Second, the share of people highly vulnerable across several dimensions is calculated by fusing indicators derived from household surveys and other sources. These estimates are representative of the population in a specific geographic region (typically subnational) and can therefore be linked to the exposure data. The final step calculates the number of people who are at high risk by combining exposure and vulnerability estimates from the previous steps for the same location and population group. The following sections set out the methods and data used in each of these steps. A replication package containing all code used for analysis and data availability statements is available.

Step 1 - Determining who is exposed

The exposed population is estimated by combining global gridded population, degree of urbanization and hazard spatial data. The hazard data is used to identify exposed locations where hazard intensity thresholds are exceeded for a given return period event (Table 1). The degree of urbanization data is used to classify the population so that we can later merge vulnerability information specific to the rural or urban population. The hazard and degree of urbanization spatial data is first resampled so that grid cells align with the 3 arcsecond (approximately 90 m) resolution population data. We classify every population grid cell covering the globe (approximately 90 billion) by exposure to any combination of the four hazards, and by eight degree of urbanization categories. As a result, the global population is assigned to one of 128 possible exposure-urbanization categories at a very fine spatial scale (see Supplementary Information). The population in each exposure-urbanization category is aggregated from grid level to larger spatial units using boundary data corresponding with the level of representative survey data used to assess vulnerability. When statistical regions partially cover a grid cell, the population assigned to that region is weighted by the cell coverage fraction.

The following sections describe the spatial population, hazard and boundary data used.

Population data

To map the population, we use GHS-POP data from the Global Human Settlement Layer (GHSL)^52. The dataset depicts the residential population at a resolution of 3 arcseconds, derived from the Gridded Population of the World version 4.11 (GPWv4.11), and disaggregated to grid cells informed by the distribution, classification, and volume of built-up areas also mapped by the GHSL⁵³. The populations of GHSL are well suited for merging with survey data since they are closer to statistical values than alternatives such as WorldPop because the spatialization method maintains the population in the administrative or census region⁵⁴. We use the 2010 and 2020 global data files.

To classify the population so that we can merge vulnerability information specific to the rural or urban population, we use GHS-SMOD gridded data from GHSL which applies the Degree of Urbanisation methodology recommended by the UN Statistical Commission55. Settlements are classified according to population cluster sizes, population densities and built-up area densities, as defined by the stage I of the Degree of Urbanisation, at a spatial resolution of 1km globally29. Although the vulnerability data is only available for rural and urban subgroups, we classify the exposed population by seven degree of urbanization subcategories depicted by GHS-SMOD to improve our ability to match the definition used by surveys which varies across countries (described in Step 3).

Population estimates derived from GPWv4.11 are based on outdated census in some countries. For consistency with the population used for global poverty monitoring and other statistics, we use country population data from the World Bank’s World Development Indicators (WDI) to rescale estimates extracted from gridded data56.

Hazard data

For agricultural drought, we use Historic Agricultural Drought Frequency data from the FAO. The data depicts the annual frequency of severe drought in up to two growing seasons where more than 30% of cropland or grass-land is affected, according to the Agricultural Stress Index (ASI), with a spatial resolution of 1 km15. The ASI is based on remote sensing vegetation (NDVI) and land surface temperature (BT4) data, combined with information on agricultural cropping cycles derived from historical data and a global crop mask. Specifically, a vegetation health index (VHI) is defined as a weighted combination of deviations of NDVI and BT4 from the historical range of NDVI and BT4. Grid cells with a VHI value below 35% over a growing season are considered as experiencing severe drought. The final ASI value is calculated as the percentage of crop or grassland pixels within each administrative unit affected by severe drought. Unlike the other hazard data used, these drought maps are not produced using extreme value modeling but calculated directly from the 39-year time series: a location recording at least one severe drought (as defined by the ASI) between 1984 and 2022 is considered exposed to a 39-year return period event.

For pluvial and fluvial floods, we use the Fathom Global 2.0 flood hazard data16. The Fathom data depicts the maximum inundation depth of fluvial and pluvial floods at a resolution of 3 arcseconds for return periods ranging from 5 to 1,000 years. We use the undefended flood maps. While Fathom provides a defended option for fluvial flooding, it uses GDP as a proxy to set defense standards rather than the actual presence of flood defense structure. There is evidence that many low and middle-income countries do not have effective flood protection systems6^,57,18. The Fathom data have global coverage between 56°S and 60°N. For coastal floods, we use data from Deltares since the Fathom data does not represent coastal flooding17. Coastal flooding is modelled using the same digital elevation model (MERIT DEM) at the same 3 arcsecond spatial resolution as Fathom 2.0, but forced by tide and storm surges (e.g., during a tropical cyclone). The Deltares dataset depicts the maximum depth of coastal flooding for return periods ranging from 0 to 250 years, with global coverage. A combined global flood hazard map was created for return periods ranging from 5 to 100 years by taking the maximum inundation depth of any flood type.

For heatwaves, we use modelled data prepared by the World Bank Climate Change Knowledge Portal (CCKP) based on hourly ERA5 climate reanalysis data27. The probabilistic heatwave data depicts the 5-day average of the daily maximum Environmental Stress Index (ESI) at a resolution of 0.25 degrees for return periods ranging between 5 and 100 years. The ESI is an approximation for the Wet Bulb Globe Temperature (WBGT), derived from temperature, relative humidity and solar radiation58. The ESI formula was adjusted to correct for underestimation from solar radiation59.

For tropical cyclones, we use modelled data generated using the STORM (Synthetic Tropical cyclOne geneRation Model) synthetic resampling algorithm20^,21. STORM is applied to 38 years of historical cyclone track data from the International Best Track Archive for Climate Stewardship (IBTrACS) to statistically extend the dataset to 10,000 years of cyclone activity. The dataset covers all tropical cyclone basins except the South Atlantic which has too few historical cyclone formations in this basin for adequate model fitting. The cyclone data depicts the maximum 10-minute average sustained wind speed at a resolution of 0.1 degrees for return periods ranging from 10 to 10,000 years.

Boundary data

To aggregate gridded exposure data, we use boundary data corresponding with the smallest representative spatial units in household surveys used to derive vulnerability indicators. In many countries, these are admin-1 boundaries, but survey statistical regions are not necessarily aligned with administrative boundaries and vary between admin-2 and national level. Subnational boundary data sources include Global Administrative Unit Layers (GAUL) 2015, Nomenclature of Territorial Units for Statistics (NUTS), GADM (v4.1), United Nations Common Operational Datasets, and National Statistical Offices (NSOs)60^,61. The boundary data allows estimates aggregated from gridded spatial data to be merged with survey-based population vulnerability estimates.

Step 2 – Determining who is vulnerable

The share of the population considered highly vulnerable is derived from both household surveys and spatial data. We identify those vulnerable based on these different types of data separately. The following sections detail the sources of data used to calculate survey-based vulnerability indicators and how they are fused together to calculate the share of the population vulnerable on any dimension (or combination of dimensions), typically at subnational level. The final section of this step describes the Rural Access Index spatial data and how it is used to estimate the population vulnerable in terms of access to services and markets.

Survey-based vulnerability data

Six of the seven dimensions of vulnerability we consider are estimated using data from household surveys or similar population-based sources. For income, educational attainment, access to improved drinking water and access to electricity, we use household survey microdata from the World Bank’s Global Monitoring Database (GMD), which collates and harmonizes variables across countries and over time to allow comparability. The GMD data contains household level income or consumption aggregates used for global poverty monitoring as well as information on the highest education level completed by adults and access to infrastructure.

The income and consumption data are extrapolated or interpolated to a common reference year (e.g., 2021) following the method used by the World Bank’s Poverty and Inequality Platform (PIP) for global poverty monitoring 62. The extrapolation uses growth rates from national accounts, either real GDP per capita or real Household Final Consumption Expenditure per capita and a passthrough rate of 0.7. This allows the share of the population falling beneath a given poverty line to be calculated between and beyond the survey years which can be infrequent. For other indicators, we use survey data collected closest to the reference year as comparable lineup methods do not yet exist.

In a small number of countries where GMD surveys are missing variables on education, access water or access to electricity, we use estimates from other data sources for subgroups of the population. These are fused into the GMD household surveys using the simulation method described in the next section. For primary education completion, we use UNESCO Institute for Statistics (UIS) estimates based on national population censuses and household and labor force surveys, which are reported at the urban/rural level63. For access to improved drinking water, we use estimates from WHO/UNICEF Joint Monitoring Programme (JMP), which are reported at rural/urban level and by subnational region and wealth quintile when available64. For access to electricity, we use the SDG indicator 7.1.1 estimates from the World Bank’s Global Electrification Database (GED), reported at rural/urban level65. To improve country coverage, we assume universal access to improved water and electricity (i.e. no one is counted as highly vulnerable) if secondary data sources indicate more than 97% of the population have access.

For access to social protection, we use data from The Atlas for Social Protection (ASPIRE)31. The data includes the share of households that either receive social transfers or contribute to social insurance programs, since we want to capture those who will receive support in the event of a shock, even if they are not current beneficiaries. This indicator is based on household surveys when available, and administrative data or modelled estimates when survey data is incomplete. Estimates are reported for rural/urban and income quintile subgroups of the population, which we fuse into the GMD household surveys using the simulation method described in the next section. We assume no households are vulnerable on this dimension in high-income countries, which tend to have sophisticated social protection systems and are not included in ASPIRE.

For access to finance, we use individual level microdata from the Global Financial Inclusion Database (Findex)30. A household is considered vulnerable on this dimension if no household members have a bank or mobile money account. This indicator is estimated for rural/urban and income quintile subgroups of the population. Since Findex data is collected at individual level, we adjust estimates to household level to be consistent with other indicators. The adjustment is based on the number of adults in a household and a factor derived from examining countries where data on both household and individual ownership of accounts is available (see Supplementary Information). Household level estimates for each population subgroup are fused into GMD surveys using the simulation method described in the following section.

Fusing different sources of vulnerability data

Estimating the share of households vulnerable on any dimension requires “fusing” data from different sources, since not all dimensions are available from the same GMD household survey. For instance, the social protection and financial inclusion dimensions are based on ASPIRE and Findex, respectively. These indicators are available for population subgroups including rural/urban households, for households in each welfare quintile and in some cases at subnational level. Based on these subgroup statistics, we use a simulation method to fuse indicators derived from other data sources into the household level GMD data. This allows us to estimate the share of the population vulnerable on any dimension (or a combination of multiple dimensions).

For each dimension not available in GMD, households are randomly assigned a vulnerability status (0/1) based on the share of households that are vulnerable, estimated from other surveys, in the population subgroup they belong to. This preserves summary statistics reported for each subset of the population, for example, the share of the poorest rural quintile without access to social protection in ASPIRE. The share of households vulnerable on any dimension (or a combination of dimensions) is then calculated at the level of the representative statistical unit. This random assignment and aggregation process is repeated 100 times to account for household heterogeneity within each subgroup. The average share of households vulnerable on any and each combination of dimensions is calculated across all simulations and used for the analysis.

Spatial vulnerability data

For access to services and markets, we use gridded data produced to calculate the Rural Access Index (RAI) (SDG 9.1.1) for the Sustainable Development Report 202466. RAI is defined as the proportion of the rural population living within 2km of an all-season road. The dataset is derived from road network maps (Open Street Map, Global Roads Inventory Project Database, and Microsoft BING – Road Detection Project) and other geospatial data such as slope and rainfall which is used to assess the all-season status of uncategorized roads67. The data indicates the share of the population in each grid cell more than 2 km from an all-season road. We calculate the population more than 2km from an all-season road at grid level by multiplying the RAI inaccessibility data (resampled to align with the population grid) with the population data. The resulting grid level estimate of the number of people vulnerable is aggregated to the statistical regions corresponding with surveys in the same way as exposure estimates.

Step 3 - Determining who is at risk

To determine who is at risk, we combine exposure and vulnerability estimates from the previous steps. This requires first aggregating the exposure data to the same population units represented by surveys.

The grid level exposure-urbanization estimates are aggregated spatially to the level of the survey statistical regions by overlaying boundary data, as described in step 1. The exposure estimates by degree of urbanization (7 categories) for each spatial unit are then aggregated to the rural/urban classification used by the survey data by matching population shares. To estimate the rural exposed, for example, we take the population weighted exposure rate from the most rural degree of urbanization categories until we have accounted for the population share defined as rural according to the survey definition. Validation shows this method is effective at accounting for differences between rural/urban classifications (see Supplementary Information).

The spatially derived vulnerability data, measuring the population with low access to services and markets, is also classified by exposure status at grid level. This means we can directly estimate the population both exposed and vulnerable on this dimension, without making assumptions about the correlation between exposure and vulnerability within a larger geographic region. The grid level estimates are aggregated to the same population units as the survey-based vulnerability data in the same way as exposure, described above.

Once the exposure and vulnerability estimates represent the same spatial and rural/urban population units, the population at high risk in each unit (exposed to any hazard and vulnerable on any dimension) is calculated as:

where risk_RAI is the population exposed and vulnerable on the low access dimension (based on spatial data), exp is the population exposed, and vuln_survey is the share of the population vulnerable on any survey-based dimension. This calculation assumes that survey-based vulnerability dimensions are not correlated with exposure within population units (nor with spatially defined vulnerability). Estimates are aggregated to country, region and global level by taking the population weighted sum across all survey units.

Sensitivity of results

We check the sensitivity of results to several assumptions. The regional composition of the exposed population is similar when we use different intensity thresholds or return periods to define exposed locations (Extended Data Table 2). The trend in the population at high risk is robust to restricting the sample to countries with data on all dimensions from within 3 years of both 2010 and 2021. We also estimate lower and upper bounds by assuming the minimum and maximum possible overlap between the exposed and vulnerable population, rather than assuming the vulnerable population are just as likely to be exposed as those not vulnerable within a spatial unit. Estimates of the global population at high-risk range from 16% to 23% for 2021 in these extreme lower and upper bound scenarios, compared to 19% in our preferred specification (see Supplementary Information).

Data Availability

The datasets generated during and/or analysed during the current study are available at
https://doi.org/10.60572/9j17-9n52. Source data are provided with this paper.

Code Availability

The code used to generate the analysis is available at
https://doi.org/10.60572/9j17-9n52.

Acknowledgements

We are thankful for comments from an internal World Bank advisory group chaired by Luis Felipe Lopez-Calva and Jennifer Sara to develop the methodology. We are grateful for inputs from the CCKP and ASPIRE teams at the World Bank, and from the UN Sustainable Development Solutions Network. The findings, interpretations and conclusions expressed in this paper are entirely ours. These findings do not necessarily represent the views of the World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Author information

Authors and Affiliations

The World Bank, Washington, DC, USA

Ruth Hill, Nisan Gorgulu, Ben Brunckhorst, Minh Cong Nguyen, Stephane Hallegatte & Esther Naikal.

Contributions

The authors randomized the author ordering (R.H., N.G., B.B., M.C.N., S.H., E.N.). All authors contributed to the paper.

Corresponding authors

Correspondence to Ben Brunckhorst ([email protected]).

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Supplementary Information is available for this paper.

Dang, H.A.H. and Dabalen, A.L. (2018). Is Poverty in Africa Mostly Chronic or Transient? Evidence from Synthetic Panel Data. Journal of Development Studies 55 (7): 1527–47.
Hallegatte, S., Bangalore, M., Bonzanigo, L., Fay, M., Kane, T., Narloch, U., Rozenberg, J., Treguer, D. & Vogt-Schilb, A. (2016). Shock Waves: Managing the Impacts of Climate Change on Poverty.Washington, DC: World Bank.
(2023). Climate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, H. Lee and J. Romero (eds.)]. IPCC, Geneva, Switzerland, 184 pp., doi: 10.59327/IPCC/AR6-9789291691647.
Hallegatte, S., Vogt-Schilb, A., Rozenberg, J. et al.(2020). From Poverty to Disaster and Back: a Review of the Literature. EconDisCliCha 4, 223–247. https://doi.org/10.1007/s41885-020-00060-5
(2023). Global Infrastructure Resilience: Capturing the Resilience Dividend – A Biennial Report from the Coalition for Disaster Resilient Infrastructure, New Delhi.
Hallegatte, S., Vogt-Schilb, A., Bangalore, M., & Rozenberg, J. (2017). Unbreakable: Building the Resilience of the Poor in the Face of Natural Disasters. Climate Change and Development. Washington, DC: World Bank. http://hdl.handle.net/10986/25335
Notre Dame Global Adaptation Initiative Country Index (ND-GAIN). (2024). University of Notre Dame.
Inter-Agency Standing Committee and the European Commission, INFORM REPORT 2024: 10 years of INFORM. (2024). Publications Office of the European Union, Luxembourg. https://data.europa.eu/doi/10.2760/555548, JRC136641.
Eckstein, D., Künzel, V. and Schäfer, L. (2021). The global climate risk index 2021. Bonn: Germanwatch.
Bündnis Entwicklung Hilft / IFHV (2024): WordRiskReport 2024. Berlin: Bündnis Entwicklung Hilft.
Doan,Miki Khanh ⓡ Hill,Ruth ⓡ Hallegatte,Stephane ⓡ Corral Rodas,Paul Andres ⓡ Brunckhorst,Ben James; Nguyen,Minh ⓡ Freije-Rodriguez,Samuel ⓡ Naikal,Esther G. (2023). Counting People Exposed to, Vulnerable to, or at High Risk From Climate Shocks - A Methodology (English). Policy Research working paper 10619. Washington, D.C. : World Bank Group.
(2022). Climate Change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [H.-O. Pörtner, D.C. Roberts, M. Tignor, E.S. Poloczanska, K. Mintenbeck, A. Alegría, M. Craig, S. Langsdorf, S. Löschke, V. Möller, A. Okem, B. Rama (eds.)]. Cambridge University Press. Cambridge University Press, Cambridge, UK and New York, NY, USA, 3056 pp., doi:10.1017/9781009325844.
Van Hoolst, R., Eerens, H., Haesen, D., Royer, A., Bydekerke, L., Rojas, O., Li, Y. and Racionzer, P. (2016). FAO’s AVHRR-based Agricultural Stress Index System (ASIS) for global drought monitoring. International Journal of Remote Sensing, 37(2), pp.418-439.
Kogan, F.N. (1995). Application of vegetation index and brightness temperature for drought detection. Advances in space research, 15(11), pp.91-100.
(2023). Agricultural Stress Index System (ASIS): Historic Agricultural Drought Frequency (Global - 1Km). http://www.fao.org/giews/earthobservation/
Sampson, C., Smith, A. M., Bates, P. D., Neal, J. C., Alfieri, L., & Freer, J. E. (2015). A High-resolution Global Flood Hazard Model. Water resources research, 51(9), 7358-7381.
(2021). Planetary computer and Deltares global data: Flood Hazard Maps. Deltares.
Rentschler, J., Salhab, M., & Jafino, B. A. (2022). Flood Exposure and Poverty in 188 Countries. Nature Communications, 13(1), 3527.
Huizinga, J., De Moel, H. and Szewczyk, W. (2017). Global flood depth-damage functions: Methodology and the database with guidelines(No. JRC105688). Joint Research Centre.
Bloemendaal, N., Haigh, I. D., de Moel, H., Muis, S., Haarsma, R. J., & Aerts, J. C. (2020). Generation of a Global Synthetic Tropical Cyclone Hazard Dataset Using STORM. Scientific data, 7(1), 40.
Bloemendaal, N., De Moel, H., Muis, S., Haigh, I. D., & Aerts, J. C. (2020). Estimation of Global Tropical Cyclone Wind Speed Probabilities Using the STORM Dataset. Scientific data, 7(1), 377.
Eberenz, S., Lüthi, S., & Bresch, D. N. (2021). Regional Tropical Cyclone Impact Functions for Globally Consistent Risk Assessments. Natural Hazards and Earth System Sciences, 21(1), 393-415.
Bloemendaal, N. and Koks, E.E. (2022). Current and Future Tropical Cyclone Wind Risk in the Small Island Developing States. In Hurricane Risk in a Changing Climate(pp. 121-142). Cham: Springer International Publishing.
International Organization for Standardization. (2017). Ergonomics of the thermal environment—Assessment of heat stress using the WBGT (wet bulb globe temperature) index (ISO Standard No. 7243:2017). https://www.iso.org/standard/67188.html
Foster, J., Smallcombe, J. W., Hodder, S., Jay, O., Flouris, A. D., Nybo, L., & Havenith, G. (2021). An advanced empirical model for quantifying the impact of heat and climate change on human physical work capacity. International Journal of Biometeorology, 65, 1215-1229. https://doi.org/10.1007/s00484-021-02105-0
Kjellstrom, T., Freyberg, C., Lemke, B., Otto, M., & Briggs, D. (2018). Estimating population heat exposure and impacts on working people in conjunction with climate change. International Journal of Biometeorology, 62(3), 291-306. https://doi.org/10.1007/s00484-017-1407-0
World Bank, Climate Change Knowledge Portal (2024). Environmental Stress Index. https://climateknowledgeportal.worldbank.org/.
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz‐Sabater, J., Nicolas, J., Peubey, C., Radu, R., Schepers, D. and Simmons, A., (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), pp.1999-2049.
Dijkstra, L., Florczyk, A.J., Freire, S., Kemper, T., Melchiorri, M., Pesaresi, M. and Schiavina, M., (2021). Applying the degree of urbanisation to the globe: A new harmonised definition reveals a different picture of global urbanisation. Journal of Urban Economics, 125, p.103312.
Demirgüç-Kunt, Asli, Leora Klapper, Dorothe Singer, Saniya Ansar. (2022). The Global Findex Database 2021: Financial Inclusion, Digital Payments, and Resilience in the Age of COVID-19. Washington, DC: World Bank.
World Bank. (2024). ASPIRE (Atlas of Social Protection: Indicators of Resilience and Equity). July 2024 update.
Hallegatte, S., J. Rentschler, and J. Rozenberg. (2019). Lifelines: The resilient infrastructure opportunity. Washington, DC: World Bank.
Burgess, Robin, and Dave Donaldson. (2010). "Can Openness Mitigate the Effects of Weather Shocks? Evidence from India's Famine Era." American Economic Review, 100 (2): 449–53.
Dobermann, T. (2023). How building up the human capital of the world's poor can help lessen the climate crisis. LSE Business Review.
Hoffmann, R., & Muttarak, R. (2017). Learn from the Past, Prepare for the Future: Impacts of Education and Experience on Disaster Preparedness in the Philippines and Thailand. World Development, 96, 32-51.
Muttarak, R., & Lutz, W. (2014). Is Education a Key to Reducing Vulnerability to Natural Disasters and hence Unavoidable Climate Change? Ecology and Society, 19(1). https://www.jstor.org/stable/26269470
Muttarak, R., & Pothisiri, W. (2013). The role of education on disaster preparedness: case study of 2012 Indian Ocean earthquakes on Thailand’s Andaman Coast. Ecology and Society 18(4): 51. http://dx.doi.org/10.5751/ES-06101-180451
Jack, W., & Suri, T. (2014). Risk Sharing and Transactions Costs: Evidence from Kenya's Mobile Money Revolution. American Economic Review, 104 (1): 183-223.
Batista, C., & Vicente, P. C. (2023). Is mobile money changing rural Africa? Evidence from a field experiment. Review of Economics and Statistics, 1-29.
Riley, E. (2018). Mobile money and risk sharing against village shocks. Journal of Development Economics, 135, 43-58. https://doi.org/10.1016/j.jdeveco.2018.06.015
De Janvry, A., Finan, F., Sadoulet, E., & Vakis, R. (2006). Can conditional cash transfer programs serve as safety nets in keeping children at school and from working when exposed to shocks? Journal of Development Economics, 79(2), 349-373. https://doi.org/10.1016/j.jdeveco.2006.01.013
Abay, K. A., Yonzan, N., Kurdi, S., & Tafere, K. (2023). Revisiting Poverty Trends and the Role of Social Protection Systems in Africa during the COVID-19 Pandemic. Journal of African Economies, 32(Supplement_2), ii44-ii68. https://doi.org/10.1093/jae/ejac041
Gehrke, E. (2019). An Employment Guarantee as Risk Insurance? Assessing the Effects of the NREGS on Agricultural Production Decisions. The World Bank Economic Review, World Bank Group, vol. 33(2), pages 413-435.
Banerjee, A., Faye, M., Krueger, A., Niehaus, P., & Suri, T. (2020). Effects of a Universal Basic Income during the Pandemic. Working paper, University of California San Diego.
Ivaschenko, O., Doyle, J., Kim, J., Sibley, J., & Majoka, Z. (2020). Does ‘Manna from Heaven’ Help? The Role of Cash Transfers in Disaster Recovery—Lessons from Fiji after Tropical Cyclone Winston. Disasters 44 (3): 455–76.
Pople, A., Hill, R., Dercon, S., & Brunckhorst, B. (2021). Anticipatory Cash Transfers in Climate Disaster Response. Working Paper 6, Centre for Disaster Protection, London.
Londoño-Velez, J., & Querubin, P. (2022). The Impact of Emergency Cash Assistance in a Pandemic: Experimental Evidence from Colombia. The Review of Economics and Statistics, MIT Press, vol. 104(1), pages 157-165, March.
World Bank. (2022c). Poverty and Shared Prosperity 2022: Correcting Course. Washington, D.C.: The World Bank.
Gentilini et al. (2022). Social Protection and Jobs Responses to COVID-19: A Real-Time Review of Country Measures. World Bank.
Beazley, R., Marzi, M., & Steller, R. (2021). Drivers of Timely and Large-Scale Cash Responses to COVID19: what does the data say? Social Protection Approaches to COVID-19 Expert Advice Service (SPACE), DAI Global UK Ltd, United Kingdom.
Rentschler, J., Avner, P., Marconcini, M., Su, R., Strano, E., Vousdoukas, M., & Hallegatte, S. (2023). Global evidence of rapid urban growth in flood zones since 1985. Nature, 622(7981), 87-92.
Schiavina, M., Freire, S., Alessandra C., & MacManus, K. (2023). GHS-POP R20–3A - GHS Population Grid Multitemporal (1975-2030). European Commission, Joint Research Centre (JRC) [Dataset] https://doi.org/10.2905/2FF682-5B5B-4A22-8F40-C41DA8332CFE
Freire, S., MacManus, K., Pesaresi, M., Doxsey-Whitfield, E. and Mills, J., (2016). Development of new open and free multi-temporal global population grids at 250 m resolution. Population, 250, p.33.
Chen, R., Yan, H., Liu, F., Du, W., & Yang, Y. (2020). Multiple global population datasets: Differences and spatial distribution characteristics. ISPRS International Journal of Geo-Information, 9(11), 637. https://doi.org/10.3390/ijgi9110637
Schiavina, M., Melchiorri, M., & Pesaresi, M. (2023). GHS-SMOD R20–3A - GHS Settlement Layers, Application of the Degree of Urbanisation Methodology (stage I) to GHS-POP R2023A and GHSBUILT-S R2023A, multitemporal (1975-2030). European Commission, Joint Research Centre (JRC) [Dataset] https://doi.org/10.2905/A0DF7A6F-49DE-46EA-9BDE-563437A6E2BA
World Bank, World Development Indicators. (2024). “Population, total”. [Dataset]. http://data.worldbank.org/indicator/SP.POP.TOTL
Rozenberg, J., & Fay, M. (2019). Beyond the Gap: How Countries Can Afford the Infrastructure They Need while Protecting the Planet. Sustainable Infrastructure. Washington, D.C.: The World Bank.
Moran, D. S., Pandolf, K. B., Shapiro, Y., Heled, Y., Shani, Y., Mathew, W. T., & Gonzalez, R. R. (2001). An environmental stress index (ESI) as a substitute for the wet bulb globe temperature (WBGT). Journal of thermal biology, 26(4-5), 427-431.
Kong, Q., & Huber, M. (2022). Explicit calculations of wet‐bulb globe temperature compared with approximations and why it matters for labor productivity. Earth's Future, 10(3), e2021EF002334.
Food and Agricultural Organization of the United Nations. (2015). Global Administrative Unit Layers (GAUL). http://www.fao.org/geonetwork.
University of California, Berkeley. (2022). Global Administrative Areas Version 4.1. www.gadm.org.
Mahler, Daniel Gerszon, R Andrés Castañeda Aguilar, and David Newhouse. (2022). “Nowcasting Global Poverty.” The World Bank Economic Review36 (4): 835-56. https://doi.org/10.1093/wber/lhac017.
UNESCO Institute for Statistics (UIS). (2024). [Dataset]. https://uis.unesco.org/bdds. Accessed: April 2024.
United Nations Children’s Fund (UNICEF) and World Health Organization (WHO). (2024). Joint Monitoring Programme for Water Supply, Sanitation and Hygiene (JMP). https://washdata.org/data.
IEA, IRENA, UNSD, World Bank, WHO. (2023). Tracking SDG 7: The Energy Progress Report. World Bank, Washington DC.
Sachs, J. D., G. Lafortune, and G. Fuller. (2024). “The SDGs and the UN Summit of the Future. Sustainable Development Report 2024.” https://doi.org/10.25546/108572.
Iablonovski, G., Drumm, E., Fuller, G. and Lafortune, G. (2024). A global implementation of the rural access index. Frontiers in Remote Sensing, 5, p.1375476.

There is NO Competing Interest.

SupplementaryInformation.docx
Supplementary Information
Extendeddatafiguresandtables.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

One-fifth of the world's population at high risk from climate-related hazards

Status:

Version 1

Abstract

Figures

Main

Discussion

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1