World’s population susceptible to water risks is concentrated in two of the 58 IPCC reference regions

The use of climate reference regions is a common practice in climate change research and impact assessment studies^1,2,3,4, having been utilized especially within the context of the Intergovernmental Panel on Climate Change (IPCC)’s Assessment Reports (ARs) over the past 25 years. They have been developed and used for synthesizing the regional properties and historical trends of climate, and for climate change projections for the future⁴. Climate reference regions have shown their value and usefulness in providing consistent geographic delineation for showing a high number of interrelated and intricate natural phenomena with complex societal implications in climate change impact assessments^4,5,6.

Standardized geographical reference units facilitate the modelling and analysis of climate change impacts, but they mask the reality that human settlements, infrastructure, land use, economic activity and other attributes of human systems through which loss and damage are experienced are not evenly distributed. The identification of reference regions where such risks are most concentrated is beneficial across a variety of international climate policy initiatives, including United Nations Framework Convention on Climate Change (UNFCCC) loss and damage negotiations; the Santiago Network (www.https://santiago-network.org), which connects developing countries and communities with tailored, context-specific technical assistance; and a range of disaster risk-reduction initiatives under the United Nations Sendai Framework for Disaster Risk Reduction. We also observe the emergence of climate hazard impact scholarship making use of IPCC standardized climate regions to identify geographical distributions of risks, which in some cases includes demographic data to help further identify risk concentrations^7,8.

Here we identify results from a study investigating the distribution of population susceptible to selected water-related risks across the 58 geographical reference regions (Fig. 1) used in the 2022 IPCC Assessment (AR6)^5,6. For each land-based reference region we estimated water-related risks using an approach consistent with IPCC and Sendai Framework framings^5,6,9 in which risk is a multiplicative function of the probability of occurrence of hazard (or stressor), given exposure, and vulnerability, with adaptive capacity having the potential to moderate vulnerability. We used publicly available, georeferenced data for six water-related stressors (floods, droughts, groundwater depletion, overuse of available water supplied, eutrophication of surface water, and inadequate drinking water resources), global population estimates¹⁰ and, as indicators of relative vulnerability (expressed as 1—adaptive capacity), Governance Effectiveness Index¹¹ and Human Development Index (HDI)¹². These datasets were gridded, and water stressor data were converted to a numerical severity scale that accorded them standardized values of between 0 and 1.

The results are illustrated according to the IPCC reference regions (Figs. 2 and 3). Figure 2a–e reveals the specific geographic profile of each of the six water stressors. Because their impacts are typically more local relative to the geographical size of the IPCC reference areas, floods do not generate high stressor values using our approach as compared with other, less spatially concentrated stressors. Geographical reference regions containing large arid region components have high drought and overuse stressor values (e.g. reference areas 21 Sahara, 37 Arabian Peninsula, 41 Central Australia), while groundwater depletion and drinking water stressor scores are highest in sub-Saharan African reference regions (regions 22–28 in Fig. 1). Eutrophication scores are highest in reference regions that feature a combination of intensive agriculture and relatively high populations (region 36 East Asia and 18 Western/Central Europe).

Figure 2g shows the global human population distribution according to IPCC reference regions, with regions 36 East Asia and 38 South Asia each containing approximately 21% of the total human population. Figure 2h shows the vulnerability score, with reference regions in sub-Sharan Africa (22–28) and South Asia (38) having the highest relative vulnerability using these indicators. Figure 3a–f shows the results when we calculate risk for each water-related stressor for each reference region as being a product of the stressor, population, and vulnerability score for that region. With this measure we find, for example, that across all six categories of water-related stressors the risks are highest in reference regions 36 East Asia and 38 South Asia. These diverge vastly and should not be mixed with the per capita risk values of Fig. 3g–l with Africa standing out in contrast to the previous maps, where south and east Asia dominate.

In Fig. 4 we isolate the ten reference regions with the highest percentage of global population at risk and the aggregate population at risk for each of the six water-related stressors. Using regions 36 East Asia and 38 South Asia as references further emphasizes the disproportionate share of risk experienced in these regions. Together, these two regions account for 67% of populations globally susceptible to groundwater depletion, 60% for floods, 48% for eutrophication and overuse. Reference region 38 South Asia’s share of global population-adjusted water risk is especially pronounced, accounting for 50% of the global population susceptible to groundwater stress, 38% to floods, and 35% to overuse.

Our analysis suggests that the geographical reference regions being used by the IPCC would benefit from further spatial disaggregation in highly populated parts of the planet, especially South and East Asia. Current geographical boundaries being used are suitable for climate hazard identification as is done by IPCC Working Group I but are too coarse spatially when climate change impacts are viewed from a population-at-risk perspective as is done in Working Group II. Although our focus here has been on water-related climate risks, we suspect a similar disconnect may exist for other types of climate hazards assessed by the IPCC.

With loss and damage becoming a growing preoccupation of the UNFCCC process, we suspect policymakers will increasingly turn to the IPCC and particularly researchers whose work contributes to Working Group II assessments to generate projections of where future loss and damage potential is greatest. Here again, we suggest that reconsideration of the number of geographical reference regions in heavily populated regions and their boundaries is warranted. Having a larger number of geographical reference units with higher spatial resolution for heavily populated parts of the world would also support more detailed risk assessments in future Working Group II regional chapters, with particular relevance to Asia.

We use a multiplicative approach for calculating water related risks as promoted by IPCC⁶ and the United Nations Sendai Framework⁹. Risks are calculated as products of stress, exposure, and vulnerability, an approach widely used in climate change risk assessments and in more general statistical analyses of probabilistic risk in which stressor scores (i.e. the probability of an individual being subject to a stressor in each geographical unit) are assumed to be interpretable as probabilities of occurrence^13,14. We use published data that have been subjected to rigorous review processes and quality checks, meaning that the accuracy of our method is steered by the accuracy of the underlying datasets.

The data on stressors, vulnerability, and exposure, and their components were calculated from global published geospatial datasets and were harmonized to 30 arc-sec spatial resolution (~1 km × 1 km at the Equator). The calculations were performed by pixels and then aggregated by the IPCC AR6 reference regions. Most of the reference regions have a surface area between one and nine million km², meaning that each is represented by several million grid cells, an approach taken in previous, related studies^15,16.

Six stressors are included in the analysis (Table 1). They were selected to cover many of the multiple aspects of entities that are relevant to water resources management and policy in the realm of climate related risks. UNDRR has recently published a comprehensive list of climate related hazard information profiles¹⁷ that includes 281 hazards. The stressors we selected do not align fully with that document given the other long-term variables in addition to hazard events used in risk analyses in water policymaking. The data representing hydrometeorological and water quality stressors represent periods of multiple decades, while data for other risk components are from 2015.

The flood stressor score (FloS) was obtained from the EU FloodMapGL dataset¹⁸ by using the 100-year return period data linearly interpolated into the World Bank flood stress categories¹⁹. The classification’s interpolation points are shown in Table 2.

For drought DroS, the Global Aridity Index and Potential Evapotranspiration Database—Version 3²⁰ was used. We scaled it between [0, 1] with the following equation:

The scores for overuse (OvrS) and groundwater (GrwS) were obtained from the Aqueduct database²¹, from the indicators of baseline water stress, and groundwater table decline, respectively. OvrS stands for the use-availability ratio, i.e., the ratio of total water withdrawals to available renewable surface and groundwater supplies. GrwS measures the average decline of groundwater table over the study period. Eutrophication score (EutS) was obtained from Varis & Zhao (2026)²². The data for inadequate drinking water stress (DrwS) were also extracted from Aqueduct²¹.

For human exposure to water stressors, we used gridded world population data for 2015¹⁰. Vulnerability was considered as the lack of societal adaptive capacity, as the geometric average of government effectiveness (GOV)¹¹ and Human Development Index (HDI)¹²:

We use “government effectiveness” as an indicator of a country’s capability to manage water-related risk; it is based on indicators of the capacity and performance of government institutions published annually by The World Bank Group, which defines it as “… the quality of public and civil services and the degree of their independence from political pressures, together with the quality of policy formulation, implementation, and the credibility of the government’s commitment to them”¹¹. We use gridded HDI data to capture the spatial heterogeneity of the capacity of societies to address water related risks across the world¹².

Risks were calculated as the pixelwise product of population with the scores for stressors and vulnerability.

The unit for risk and exposure is population. For a pixel with 100 people, and the score for a stressor is 0.1 and the vulnerability score is 0.5, then the risk for being susceptible to that stressor within that pixel is 5 people.