The Geography of Human Survivability Under Climate Stress

The body sheds heat mainly by sweating, and sweat cools only if it can evaporate. Wet-bulb temperature combines heat and humidity into the lowest temperature evaporation can reach. Sherwood and Huber proposed in 2010 that a sustained wet-bulb temperature of 35 °C marks a theoretical limit to human adaptability, beyond which even a resting, well-ventilated person cannot lose heat. Laboratory work by Vecellio and colleagues at Pennsylvania State University found that young, healthy adults reach uncompensable heat stress well before that: the critical wet-bulb temperature averaged about 30.5 °C in warm, humid conditions, and lower still in hot, dry air. Older people, people with chronic illness and anyone working hard reach their limits earlier.

Daily maximum wet-bulb temperatures from the ERA5 reanalysis show how close large cities came in 2024. Dubai peaked at 32.1 °C and Muscat at 31.0 °C; Kolkata, Jeddah, Chennai, Lahore and Chongqing passed 30 °C. Thirty-nine of the 149 cities had at least one day at 28 °C or more, and together they hold 44% of the sampled cities' population. Kolkata had 85 days at or above 28 °C, Dhaka 64. Lower thresholds are crossed for most of the year in the humid tropics: Bangkok had 252 days at or above 26 °C, Jakarta 223 and Phnom Penh 215.

These are grid-cell values that smooth out urban heat islands and coastal humidity, so local peaks were higher. The point is not that cities passed a single lethal line — survival also depends on wind, sun, clothing, exertion and how long the exposure lasts — but that the margin between ordinary summer conditions and the physiological edge is now narrow for hundreds of millions of people. Where that margin is thin, survival depends on cooling, shade, water and rest, which is to say on infrastructure and on rules that let people stop working.

The long record is one of the great achievements of development. EM-DAT recorded an average of about 66,000 deaths a year from floods, storms, droughts, landslides and wildfires in the 1980s, dominated by the Sahel and Ethiopian famines; about 32,000 a year in the 1990s; 24,000 in the 2000s, including Cyclone Nargis, which killed some 138,000 people in Myanmar in 2008; and 11,000 in the 2010s. The World Meteorological Organization, compiling the same database, counts just over two million deaths from weather, climate and water hazards in 1970–2021, more than 90% of them in developing countries, and credits better early warning and disaster management for the decline.

The decline has not been even. Over 2000–2024, low- and lower-middle-income countries, home to 45% of humanity, recorded 71% of deaths from floods, storms, droughts, landslides and wildfires. Population-weighted, low-income countries recorded 4.1 such deaths per million people a year and lower-middle-income countries 3.6, against 1.5 in upper-middle-income and 0.6 in high-income countries. The IPCC's synthesis reached the same conclusion from a different angle: between 2010 and 2020, mortality from floods, droughts and storms was 15 times higher in highly vulnerable regions than in those with very low vulnerability.

Heat breaks the pattern, and the reason is measurement. EM-DAT recorded about 325,000 extreme-temperature deaths in high-income countries in 2000–2024, most of them in European heatwaves estimated from excess-mortality studies, against about 3,000 in low-income countries. That reflects where deaths are counted, not where heat kills: the Lancet Countdown estimates 546,000 heat-related deaths a year worldwide in 2012–2021. Recorded heat deaths rose to 37,000 a year in 2020–2024 as more studies were done. The analysis in the following sections therefore uses floods, storms, droughts, landslides and wildfires, and treats heat separately.

To separate hazard from capacity, 191 countries are split into thirds twice: by their exposure to climate hazards — the highest of their INFORM scores for river floods, coastal floods, tropical cyclones and drought — and by INFORM's lack of coping capacity, which combines institutions, disaster risk reduction, governance, communication, physical infrastructure and access to health care. Each cell of the grid shows the population-weighted rate of recorded deaths from floods, storms, droughts, landslides and wildfires per million people a year over 2000–2024.

The left column is the striking one. Among countries with strong coping capacity, the death rate is about 0.6 per million a year at low, middle and high hazard alike — 0.61, 0.64 and 0.63. Japan, the United States and China sit among the highly exposed countries in that column. Among countries with weak capacity, the rate climbs to 1.8 at low hazard, 2.3 at middle hazard and 8.7 at high hazard: in the most exposed third, weak capacity is associated with a death rate 14 times higher. Coping capacity, in other words, does not just reduce deaths — it largely decouples them from hazard.

The scatter shows the same relationship country by country. Across all countries the rank correlation between recorded death rates and lack of coping capacity is 0.41, against 0.23 for climate-hazard exposure. The countries with the highest recorded rates — Myanmar, Libya, Somalia, Haiti and small Caribbean islands struck by hurricanes — combine high exposure with fragile states or narrow economies. Single catastrophic events dominate some of these rates: Myanmar's is almost entirely Cyclone Nargis in 2008, Libya's Storm Daniel and the Derna dam collapse of 2023.

Survival through a climate hazard depends on a short list of services. Electricity powers cooling, pumps, refrigerated medicines and phones. Safe water and sanitation prevent the diarrhoeal disease that follows floods and droughts. Connectivity carries warnings. And health workers and hospital beds treat heatstroke, injuries and infections. In 2024, 0.66 billion people lacked electricity, 0.69 billion basic drinking water and 1.46 billion basic sanitation, and 2.2 billion did not use the internet. In low-income countries, population-weighted, only 49% had electricity, 35% basic sanitation and 20% used the internet.

Health capacity is thinner still. The median low-income country has 0.14 physicians and 0.7 hospital beds per 1,000 people, against 3.6 and 3.4 in the median high-income country; median health spending per person is about $130 at purchasing-power parity, against more than $4,000. These are the systems that absorb the casualties of a flood or a heatwave, and the gap is 25-fold for doctors alone.

To summarise them, a survival-infrastructure index averages each country's percentile rank on seven indicators: electricity, basic water, basic sanitation, internet use, physicians, hospital beds and health spending. Its rank correlation with recorded climate-disaster death rates is −0.52, stronger than that of income per person (−0.47) or of hazard exposure (+0.23). Infrastructure is not the only thing that matters — governance, conflict and land use all shape outcomes — but it is the part of coping capacity that can be built, measured and financed.

The benchmark is simple: within each hazard third, the population-weighted death rate of the countries with the strongest coping capacity. Because that rate is about 0.6 per million a year at every level of hazard, it is a demanding but demonstrated standard. Deaths above it are counted as avoidable. Over 2000–2024 they averaged about 14,400 a year, 88% of the 16,300 recorded each year from floods, storms, droughts, landslides and wildfires. Myanmar's toll from Cyclone Nargis dominates the total; excluding Myanmar, avoidable deaths average about 8,300 a year, 78% of the rest.

The avoidable deaths concentrate in a short list of countries. After Myanmar come Somalia, whose drought of 2010–2011 turned into famine; the Philippines, struck by more typhoons than almost anywhere; India; Libya; Pakistan, Haiti and Bangladesh. Bangladesh's position illustrates both the problem and the solution: its cyclone deaths have fallen from hundreds of thousands in 1970 and 1991 to a few dozen in recent storms through warnings, shelters and volunteers, yet its exposure is so high that it still sits above the benchmark.

These are lower bounds. They count only recorded deaths from five hazards and leave out heat, disease after disasters, and deaths in countries where events go unrecorded. They also take coping capacity as fixed at its 2026 level, so countries that improved over the period are judged by their present strength. The calculator lets readers choose the benchmark — the strongest-capacity third or the middle third — and the share of the gap closed, and see the deaths that would be avoided each year and over a decade.

The map brings the layers together. Recorded death rates are highest across a belt of fragile and highly exposed states — Somalia, Myanmar, Haiti, Libya, Afghanistan — and on small islands where a single hurricane can kill a measurable share of the population. Climate-hazard exposure is highest across South and Southeast Asia, the Horn of Africa and the Sahel, the Caribbean and parts of China and the Americas, many of them places where capacity is strong enough to keep deaths low.

Coping capacity and survival infrastructure are weakest across sub-Saharan Africa, parts of South Asia and in countries affected by conflict. Where weak capacity overlaps high hazard — the Horn of Africa, the Sahel, Bangladesh, Pakistan, Myanmar, Haiti — avoidable deaths per person are highest. The country card shows each country's recorded losses, hazard scores, infrastructure and, where it has a sampled city, its humid-heat peak in 2024.

Two cautions apply. Recorded deaths are lumpy, so one catastrophe can dominate a 25-year rate. And absence of recorded deaths is not proof of safety: some of the countries with the lowest recorded rates have the weakest systems for recording them. The map is best read alongside the infrastructure and capacity layers, which do not depend on how deaths are counted.

Warnings first. The 2024 global status report on early warning systems by UNDRR and the WMO finds that countries with limited multi-hazard early warning coverage have disaster mortality about six times higher than those with substantial to comprehensive coverage. Yet fewer than half of the least developed countries and about a third of small island developing states have such systems. The UN's Early Warnings for All initiative aims to close that gap; its success depends on the last mile — sirens, phones, radio, volunteers and shelters — as much as on forecasts.

Then the survival infrastructure. Reliable electricity for cooling and pumps; water and sanitation that keep working through floods and droughts; health facilities with power, water and staff; and cyclone and flood shelters within reach. For humid heat specifically, survivability depends on access to cool spaces, on rest and water for outdoor workers, and on heat-health warning systems that trigger action. These investments serve development in ordinary years, which is why they are among the most cost-effective forms of adaptation.

And count the dead. The gap between 325,000 recorded heat deaths in high-income countries and 3,000 in low-income ones is a gap in civil registration and mortality surveillance. Without counting, heat remains a silent killer, budgets follow visible disasters, and progress cannot be measured. Strengthening vital statistics and excess-mortality monitoring is part of adaptation, not an afterthought. The aim of all three is the same: to move every country into the left-hand column of the survivability grid, where hazard no longer decides who dies.

Disaster deaths are from EM-DAT (CRED, UCLouvain) as processed by Our World in Data: deaths and missing people by country, year and hazard. Climate-related deaths here are floods, storms, droughts, landslides and wildfires; extreme-temperature deaths are reported separately because their recording is concentrated in high-income countries; earthquakes and volcanoes are excluded. Rates are deaths in 2000–2024 divided by the sum of each year's population (World Bank). Income groups use the current World Bank classification; group totals in section 2 are Our World in Data's aggregates.

Hazard exposure is the highest of each country's INFORM Risk Mid-2026 scores for river flood, coastal flood, tropical cyclone and drought; coping capacity is INFORM's lack of coping capacity. Countries are split into thirds on each. The avoidable-death benchmark in each hazard third is the population-weighted death rate of the strongest-coping third. The survival-infrastructure index is the mean percentile rank, among 191 countries, of electricity access, basic drinking water, basic sanitation, internet use, physicians and hospital beds per 1,000 and current health expenditure per person (at least five of seven required), from the World Development Indicators.

City heat uses daily maximum 2-metre wet-bulb and apparent temperature for 2024 from the ERA5 reanalysis via Open-Meteo for 149 large cities chosen as in this series' heat and air-pollution report. All results are associations based on recorded data, which understate losses where recording is weakest. Related reports: global survivability and avoidable losses and the geography of global vulnerability.