Wet-Bulb Temperatures and the Emerging Geography of Unsurvivable Heat

Sherwood and Huber's 35 °C limit follows from physics. Skin must stay at about 35 °C for the body to dump the heat its metabolism produces, and heat can only flow from warm to cool. Once the wet-bulb temperature reaches skin temperature, even perfect sweating in a gale cannot remove heat, and core temperature climbs until it becomes fatal, typically within hours. The limit assumes an idealised person: naked, at rest, in shade and wind, with unlimited water and fully functioning sweat glands. Everyone falls short of that ideal, which makes 35 °C a ceiling rather than a threshold.

In 2021–22 the PSU HEAT Project tested the ceiling directly. Young adults aged 18 to 34 sat quietly or walked slowly in an environmental chamber while temperature or humidity was raised step by step until their core temperature could no longer stabilise. No participant reached 35 °C. In warm-humid conditions (36–40 °C air), their critical wet-bulb temperature averaged 30.55 ± 0.98 °C; in hotter, drier air it was lower still, because the air itself was heating the skin faster than sweat could cool it. The authors' conclusion was that a single wet-bulb threshold cannot apply across all climates, and that where it does apply, in humid heat, it is well below 35 °C.

These were healthy young volunteers doing almost nothing. Older adults sweat less and their hearts struggle to move blood to the skin; infants, pregnant women and people with heart, lung or kidney disease, or on some medications, are more vulnerable still. Outdoor workers generate several times the metabolic heat of a person at rest. For all of them the practical limit is lower, and heat kills long before uncompensable conditions arrive, through heart attacks, strokes and kidney failure brought on by strain. The thresholds in this dashboard are therefore not lines below which heat is safe, but markers of a gradient of danger.

The hottest wet-bulb hours of 2024 cluster in one place. Nine of the eleven sites that reached 30.5 °C lie on the Persian Gulf, where very warm, shallow sea water feeds humid sea breezes into air already above 35 °C: Abu Dhabi peaked at 32.3 °C, Ras al Khaimah at 32.2 °C, Bandar Abbas at 31.6 °C and Doha, Kuwait City and Manama at 31.4 °C. Dubai, the largest city in the sample to cross the line, reached 32.1 °C, and Muscat on the Gulf of Oman 31.0 °C. The one site outside the Arabian region was Kolkata, at 30.9 °C during its pre-monsoon and monsoon heat.

Just below the line sits a second tier. Chennai, Jeddah, Jacobabad in Pakistan's Sindh province, Lahore, Visakhapatnam and Chongqing all peaked between 30.2 °C and 30.4 °C, and Dhaka, Wuhan, Hong Kong, Chattogram and Karachi between 29.5 °C and 29.8 °C. Twenty-three of the 149 sampled cities reached 29 °C, and 64 — home to 412 million people — had at least one hour above the 27 °C mark of extreme humid heat. The world's largest humid megacities are within one to two degrees of the measured physiological limit at their hottest hour.

These are reanalysis values, which average conditions over grid cells tens of kilometres across and so smooth out the sharpest local peaks. Weather-station records are higher: Raymond, Matthews and Horton found that a few coastal subtropical stations had already reported wet-bulb temperatures of 35 °C, briefly, and that extreme humid heat above 27 °C had more than doubled in frequency since 1979. The values here are best read as a conservative, consistent comparison across places, and the true local extremes — in a dense neighbourhood, a factory floor or a tin-roofed home — as higher.

Wet-bulb temperature rises with both heat and humidity, so very different weather can produce the same number. A wet-bulb temperature of 30 °C can come from 32 °C air at 90% humidity or from 45 °C air at about 30%. The physiological consequences are not the same. In humid air the skin gains little heat from the air but cannot shed heat by sweating; in hot, dry air sweat evaporates readily, but the air itself pours heat into the body, and sweat rates cannot rise enough to compensate. That is why the PSU experiments found critical wet-bulb temperatures well below 30.5 °C in hot-dry conditions.

The chart places each site's hottest hour of 2024 by air temperature and wet-bulb temperature, with faint lines of constant relative humidity. The highest wet-bulb hours cluster between 32 °C and 41 °C of air temperature at roughly 50–85% humidity: humid heat, similar to the 36–40 °C humid environments in which the 30.55 °C average was measured. Bandar Abbas and Doha reached their peaks at about 34 °C and 83–85% humidity; Dubai and Jacobabad at 39–40 °C and just over 50%. For these places the 30.5 °C marker is the relevant line.

The dry end tells a different story. Desert and Sahelian cities such as Khartoum, Niamey and Baghdad rarely reach high wet-bulb temperatures because their air is dry, but their air temperatures reached 44–50 °C in 2024, and there a lower wet-bulb value already marks uncompensable heat. A wet-bulb map alone therefore understates the danger in hot-dry regions. Combined heat indices, or limits defined for both temperature and humidity, as the PSU group and later climate studies do, are better guides there; this dashboard flags the issue rather than resolving it.

How fast does humid heat rise with global warming? A useful result from atmospheric physics is that extreme wet-bulb temperatures in the tropics are tied to temperatures in the free atmosphere above, which warms nearly uniformly across the tropics. Zhang, Held and Fueglistaler (2021) showed that extreme tropical wet-bulb temperature rises about 1 °C for each degree of tropical mean warming, and projected a rise of 1.33–1.49 °C (66% range) at 1.5 °C of global warming relative to pre-industrial. This analysis uses that ratio, 0.94 °C per degree of global warming (0.89–0.99), and applies it to every hour of 2024, a year the World Meteorological Organization put at 1.55 °C above the 1850–1900 average.

The result is a ladder. At today's level, the hottest hour reached 30.5 °C in 3 sampled cities with 9 million people. At 2 °C, the upper Paris goal, Chennai, Lahore, Jeddah and Chongqing join them: 7 cities, 54 million people. At the 2.3–2.5 °C implied by current national pledges, Dhaka and Wuhan follow: 9 cities, 78 million. At the 2.8 °C to which current policies lead, according to UNEP's 2025 Emissions Gap Report, 16 cities with 161 million people cross — adding Karachi, Chattogram, Shanghai, Guangzhou, Tianjin, Hong Kong and Luanda — almost a quarter of the population of all sampled cities. Using the 3-hour peak instead of the single hottest hour halves these numbers, to 87 million at 2.8 °C.

The 35 °C ceiling, by contrast, stays out of reach on this evidence. The first site to reach it, Abu Dhabi, would do so only at about 4.4 °C of warming, and Dubai at 4.6 °C. That broadly agrees with Vecellio, Kong, Kenney and Huber (2023), who found that limiting warming to below 2 °C nearly eliminates exposure to widespread uncompensable humid heat, with brief exceedances at 1.5 °C confined to parts of the Middle East and the Indus valley, more widespread brief exceedances at 2 °C including eastern China and sub-Saharan Africa, and a sharp rise at 3 °C. The ladder shown here is a lower bound: it uses smoothed reanalysis data, healthy young adults' limits and no change in humidity patterns.

A single hour above a threshold is rarely fatal. The PSU limits describe conditions in which core temperature keeps rising, and how dangerous that is depends on how long it lasts and whether nights bring relief. The hourly record matters as much as the peak. In 2024 Kolkata spent 1,685 hours — 70 full days — with a wet-bulb temperature above 27 °C, and 14 hours above 30.5 °C. At 2.8 °C of warming the same weather would give it 3,374 hours above 27 °C and 106 above 30.5 °C. Dubai's hours above the measured limit would rise from 59 to 284; Abu Dhabi's from 247 to 723.

The biggest proportional changes come where the peaks stay below the line. Lagos had 123 hours above 27 °C in 2024; at 2.8 °C it would have about 2,050. Bangkok's would rise from 905 to about 3,120, Mumbai's from 332 to about 1,400 and Jakarta's from 113 to about 1,300. Across all sampled cities, person-days with extreme humid heat would rise from 19 billion to 43 billion a year. These places are not on the map of the 30.5 °C line, but they are where most hot-humid exposure — and most heat-sensitive work, sleep and illness — will be.

A single year is a thin baseline, and 2024 was warm. For the 28 hottest sites, the analysis retrieved daily maximum wet-bulb temperatures for 2019–2023. 2024 was the hottest, or joint hottest, of the six years at 14 of them, and its peaks averaged 0.64 °C above the 2019–2023 median. Basing the ladder on each site's median annual peak and the long-term warming level of about 1.4 °C instead would push crossings about half a degree later on average. Luanda is the outlier: its 2024 peak of 29.4 °C was 1.5 °C above any of the five previous years.

The first lever is mitigation, and the ladder shows its value in people. Holding warming to 2 °C rather than the 2.8 °C of current policies keeps about 107 million people in the sampled cities alone on the safe side of the measured limit — two-thirds of the current-policies figure — and cuts person-hours above it by more than two-thirds, from 2.66 billion to 0.75 billion a year. Because extreme wet-bulb temperatures rise almost degree for degree with global warming, every tenth of a degree avoided moves the line back. No amount of adaptation recovers this margin as cheaply for outdoor life, farming and work.

The second lever is protection, and it already works. Ahmedabad's heat action plan, introduced after a 2010 heatwave that caused more than 1,300 excess deaths, combined early warnings, public cooling, hospital preparedness and outreach; Hess and colleagues estimated it avoided about 1,190 deaths a year. Yet protection is unevenly spread. The IEA estimates that about 3.5 billion people live in hot regions, and only about 15% of them own an air conditioner. The ILO puts the number of workers exposed to excessive heat at 2.41 billion, and the Lancet Countdown reports heat-related deaths up 63% since the 1990s.

The cities on the ladder are the priority. The Gulf states have the money to cool almost everyone indoors, so their risk is concentrated among outdoor and migrant workers and depends on enforced midday work bans and heat rules. Kolkata, Dhaka, Karachi, Lahore and Chennai combine humid heat with low air-conditioner ownership, unreliable power and large informal workforces; for them, warnings, cooling centres, reflective roofs, shaded and ventilated homes, and reliable electricity at the hottest hours are the difference between dangerous heat and a mass-casualty event. Sustainable cooling — efficient appliances, passive design and clean power — keeps adaptation from adding to the warming that drives the problem.

Hourly 2-metre wet-bulb and air temperatures for 2024 come from the ERA5 reanalysis via the Open-Meteo archive API, in local time, for 149 large cities chosen as in this series' heat and labour report (the largest city in every country of 5 million or more, plus up to seven others in the most populous countries; populations from GeoNames, 676 million in total), and for 12 hotspot sites chosen for their known humid heat: Jacobabad, Bandar Abbas, Dammam, Ras al Khaimah, Abu Dhabi, Doha, Kuwait City, Basra, Manama, Puerto Peñasco, Houston and Visakhapatnam. Hotspot sites carry no population. Daily maximum wet-bulb temperatures for 2019–2023 were retrieved for the 28 sites with the highest 2024 peaks.

For each site the analysis computes the highest hourly wet-bulb temperature, the highest 3- and 6-hour running means, hours and days above each threshold, and monthly maxima. Warming is represented by adding k × (GWL − 1.55) to every hour, where GWL is global warming above 1850–1900, 1.55 °C is the WMO estimate for 2024 (± 0.13 °C) and k is 0.94 °C per °C (0.89–0.99), derived from Zhang, Held and Fueglistaler's projected 1.33–1.49 °C rise in extreme tropical wet-bulb temperature at 1.5 °C. Thresholds are 27 °C, 29 °C, 30.5 °C (Vecellio et al. 2022), 33 °C and 35 °C (Sherwood and Huber 2010). The calculator uses Stull's (2011) empirical formula.

The main caveats are as follows. Reanalysis averages over grid cells tens of kilometres across and understates local and station extremes. A uniform shift ignores changes in humidity, circulation and day-to-day variability, and the ratio k is best constrained in the tropics; humid-heat projections for the extratropics, such as the US Midwest, depend on regional humidity changes this method cannot capture. 2024 was a warm year for many sites. City-proper populations understate metropolitan populations in some places and overstate them where municipalities are large. The thresholds come from young, healthy adults. Related reports: the geography of survivability under climate stress and the compounding burden of heat and air pollution.