Heat Risk Is More Than a Temperature Reading
This series examines the global intensification of human heat stress from 1950 to 2024 and the expanding number of people exposed to it. Its focus is not simply whether the air is warmer, but whether the combination of weather, time of day, duration and location is placing a greater physiological load on human bodies. The practical judgment it supports is straightforward but consequential: a heat signal is useful only when it tells us something about the conditions people must endure, recover from and plan around.
That distinction matters because “heat” can describe several different things. A weather station may report air temperature. A thermal index may estimate how the body responds to temperature, humidity, wind and radiation. An exposure measure may count how many people live through a threshold, and for how many days. A health assessment must then consider age, illness, medication, work, housing, cooling access and social isolation. These are connected, but they are not interchangeable. A hotter map is not automatically a map of illness, and a larger exposed population is not the same as a larger hazard.
The number on the thermometer is not the whole exposure
Imagine two places with the same air temperature. In one, dry air and moving wind allow sweat to evaporate and carry heat away. In the other, humid air slows evaporation, weak wind reduces convective cooling, and radiation adds heat to the body. The thermometer can show the same value while the thermal load is different. The difference is not a matter of subjective discomfort alone. It affects the physical conditions under which the body tries to maintain a safe internal temperature.
The global study uses the Universal Thermal Climate Index, or UTCI, to capture that broader load. UTCI combines air temperature, humidity represented through dewpoint temperature, wind speed and radiation. It then expresses the modeled physiological response as a “feels-like” temperature in degrees Celsius. The measure is built around a standardized reference person, so it is not a diagnosis of what every individual will experience. It is a way to compare thermal stress consistently across places and years while keeping the body’s response in view.
The study uses four established UTCI bands: at least moderate heat stress begins at 26 °C, strong heat stress at 32 °C, very strong heat stress at 38 °C and extreme heat stress at 46 °C. These thresholds are useful because they turn a continuous environmental signal into categories that can be counted and compared. They also make a central feature of heat risk visible: crossing a high threshold can matter more than a small change in a mild range. Thermal indices are nonlinear, so equal warming at already high temperatures can produce a larger change in the index than the same warming at lower temperatures.
That does not mean a UTCI value predicts a fixed outcome for a named person. The model does not account for every difference in age, body composition, acclimatization or health. Nor does a threshold crossing by itself establish that someone will become ill. The measure answers a narrower, important question: how the physical environment is changing in relation to a modeled human heat load. Health guidance supplies the additional context about who has less capacity to cool or recover.
The diagram’s logic is the foundation for everything that follows. Air temperature remains an important input, but it is not the endpoint. Heat stress becomes a public-health concern through the interaction of the physiological load with time, place and people. The same index can reveal a changing hazard; it cannot, on its own, tell us how a particular person will respond.
A global heat-stress transition
With that distinction in place, the study asks how human-experienced heat stress has changed across the world. It uses ERA5-HEAT reanalysis data to examine land areas from 1950 through 2024. Reanalysis combines observations with modeled information to provide a consistent record where measurements are sparse or uneven. The authors compare the 1970s with the most recent decade, 2015–2024, because the 1970s mark the beginning of a clear and sustained rise in the indicators they examine.
The result is not one number but a pattern across several dimensions. The average of the ten hottest daily maximum UTCI values each year has increased by 0.27 ± 0.05 °C per decade since the 1970s. The ten hottest nighttime minima have risen faster, by 0.32 ± 0.05 °C per decade. The global record also adds roughly two days per decade with at least strong heat stress, one day per decade with at least very strong heat stress and two tropical nights per decade. These are global averages over land; they do not imply that every region follows the same path.
Frequency changes make the shift especially clear at the dangerous end of the scale. Comparing 2015–2024 with the 1970s, the study finds that extreme heat stress occurred about 2.5 times as often in Europe and South America, twice as often in North America, and 1.8, 1.7 and 1.2 times as often in Africa, Oceania and Asia. A relative increase can be large even when an extreme remains rare. In Europe, for example, extreme heat stress affected only a small fraction of days and locations in the recent decade, but its frequency was still 138% higher than in the 1970s. Relative and absolute measures answer different questions and should not be confused.
The geographic footprint is changing as well. Some subtropical areas of southern North America, southern Europe, northern and southern Africa and South America now experience up to 50 more days per year with at least strong heat stress. Locations that recorded no heat-stress days or tropical nights in the 1970s now record a non-zero annual average. This matters for two reasons. Regions already accustomed to heat face a more intense burden, while regions with less historical experience may have less prepared buildings, institutions, warning systems or social routines.
Seasonal timing adds another layer. In the northern extratropics, the period between the first and last day above a heat-stress threshold is now longer than in the 1970s—by about 15 days for moderate heat stress, 12 for strong, 9 for very strong and 6 for extreme heat stress on average. In Europe, moderate heat stress now begins on average in mid-May rather than early June, while strong heat stress begins in June rather than July. A longer season changes the window in which people need protection and reduces the usefulness of treating heat as a short, exceptional interruption.
These findings support a broad conclusion: the hazard is intensifying in its peaks, recurring more often, reaching farther and occupying more of the year in some regions. They do not support the simpler claim that every place is becoming uniformly hotter in every measure. Parts of India and Pakistan, for example, show decreases in daytime UTCI in the study even as nighttime minima rise. A credible global account must preserve such divergence rather than force every local pattern into one narrative.
From hot days to a burden people must carry
The daytime pattern is only the beginning of the human story. Nights are part of exposure because the body needs opportunities to release stored heat and recover. The study defines a tropical night using a minimum air temperature of at least 20 °C, then examines how much heat stress is present during those already-warm nights. The number of tropical nights and the severity of the heat stress on them are separate measures. A place can have more warm nights without every one imposing the same modeled thermal load.
The distinction becomes more important when hot days and warm nights occur in sequence. The study defines a compound event as consecutive days in which the daily maximum UTCI reaches at least strong heat stress while the same grid cell also experiences a tropical night. Globally, these events have become more frequent and prolonged, particularly in Europe, Africa and North America. In Europe, single-day compounds increased by 73% compared with the 1970s, while events lasting 15–30 days became 3.4 times more common. The pattern is evidence of changing persistence and reduced overnight relief; it is not, by itself, a new estimate of mortality.
Exposure then depends on who lives in these conditions. In the 1970s, 55.4% of the global population experienced at least 90 days of strong heat stress annually. In 2015–2024, the figure was 69.5%. The share experiencing at least one day of extreme heat stress rose from 15.9% to 21.9%, which the study translates into roughly one billion additional people when population growth is included. For most categories, the increase associated with changing heat stress was equal to or larger than the increase associated with population growth and redistribution. That is why hazard and exposure must remain separate: both are changing, and each contributes to the total burden.
Exposure is not the same as harm either. CDC guidance notes that hot days can affect physical and mental health and are associated with worse pregnancy and birth outcomes and more emergency visits and hospitalizations, including for heat illness, cardiovascular and respiratory disease, kidney disease, mental-health conditions and injuries. Risk is shaped by where people live, work, play and learn. Children, older adults, pregnant people, people with chronic conditions or disabilities, outdoor workers, people without reliable cooling and people taking medications that impair sweating or temperature regulation may have less margin for safe recovery. Urban heat islands and rural work conditions create different routes to risk.
The practical lesson is not that every heat number should be treated as an emergency for every person. It is that a responsible heat judgment must ask more than “How hot will it get?” It should ask which measure is being used, whether the heat occurs during the day or night, how long it persists, where people are located, and who has the fewest ways to cool down. The Nature study makes the changing physical pattern visible; health guidance explains why that pattern matters differently across settings and populations.
Heat risk is therefore a relationship, not a single reading. Weather conditions shape physiological load. Repeated thresholds and longer seasons shape exposure. Warm nights can remove recovery between hot days. Population distribution determines how many people encounter the hazard, while health, work, housing and cooling access influence what that encounter means. The central question is no longer just whether the air crossed a temperature line. It is whether the chosen signal captures the conditions people must withstand—and whether protection is designed around the full pattern.