Greatly enhanced risk to humans as a consequence of empirically determined lower moist heat stress tolerance

The magnitude and spatial extent of uncompensable heat stress increases as expected from theory (16, 17, 27, 28) with every degree of global mean surface temperature (GMST) warming or “global warming levels.” Regional emergence of dangerous moist heat is heterogeneous in relation to future warming levels as semiequatorial regions first experience some hours of threshold exceedance at 1.5 °C warming above preindustrial levels (and likely before), while other parts of the globe remain safe until the GMST increase reaches 4 °C (Fig. 1 A–D). With increased global warming, the regions that will experience the first moist heat waves and subsequent substantial increases in accumulated hot-hours per year are also the regions with the largest concentrations of the world’s population, specifically, those in India and the Indus River Valley (population: 2.2 billion), eastern China (population: 1.0 billion), and sub-Saharan Africa (population: 0.8 billion) (Fig. 1E). These also are generally home to lower-to-middle income countries and hence represent some of the world’s most vulnerable populations.

Extended Data, SI Appendix, Fig. S1A shows the global extent of threshold exceedance at the previous theoretical upper limit of T_w = 35 °C at 4° GMST warming. A minimal amount of annual hot-hours are confined to sub-Saharan Africa and the Indus Valley at T_w = 35 °C, comprising a small fraction of the spatial extent of substantial annual hot-hour accumulation when applying the empirical thresholds at 4 °C GMST warming (Fig. 1D). As a result, the number of person-hours of exposure to dangerous heat-health conditions using a T_w = 35 °C threshold is severely underestimated (Table 1).

In climate change scenarios of 2 °C warming and below, conditions associated with threshold exceedance are limited to the Indus River Valley, east China, the Persian Gulf coastline, and sub-Saharan Africa. Increases in the number of hours over the threshold in these regions are mild in transitioning from 1.5 °C to 2 °C (Fig. 1 A and B). Most of these regions are doubly exposed to extreme heat environments because of high incoming solar radiation causing high land temperatures in conjunction with proximity to water bodies with climatologically high sea surface temperatures. Monsoon dynamics are also likely to exacerbate such conditions, such as in South Asia and east China, and should continue to do so in the future with a projection of increased associated moisture in the seasonal events (29).

In future climates characterized by 3 °C and 4 °C GMST increases (Fig. 1 C and D), moist heat stress deepens, and these regions see significant increases in the number of annual hours above the heat stress thresholds. In addition, the affected area broadens, and regions of North and South America begin to exhibit an accumulation of hot-hours in the +3 °C warmer world (Fig. 1C), while northern and central Australia begin to experience an accumulation of hours above the threshold once GMSTs reach 4 °C above the preindustrial baseline (Fig. 1D).

These large-scale patterns are strongly modulated by regional and local scale climate processes and geography. To estimate risk to people requires further consideration of population distributions. Regionally, southwest Asia dominates the population-based distribution of hot-hours across the four warming scenarios presented (Fig. 2 A–D). While impacts are fairly contained to eastern Pakistan and the Indus River Valley in northern India in the 1.5 °C and 2 °C warming scenarios, they expand greatly in 3 °C and 4 °C warmer worlds with substantial accumulation of annual hot-hours along the Indian coast of the Bay of Bengal as well as into the countries of Bangladesh and parts of Myanmar. Heat stress also dominates the highly populated cities of East Asia (Fig. 2 E–H). As in southwest Asia, dangerous heat-health conditions stay clustered up until 4 °C, primarily in the highly populated cities of eastern China such as Shanghai, Nanjing, and Wuhan. However, with future warming, the coastal regions of Japan, Korea, Cambodia, Laos, and Thailand also begin to experience increased accumulations of hot-hours with cities such as Tokyo, Seoul, Bangkok, and Ho Chi Minh City affected (Table 2). In Africa, the heat stress signal increases in magnitude and extent with every degree of GMST increase, spreading radially outward from the middle of the Sahel to the Atlantic Ocean and Gulf of Guinea coastlines as GMST increases (Fig. 2 I–L).

Coastal influences, be it ocean, sea, or river, are seen in the Middle East (Fig. 2 M–P) and North America (Fig. 2 Q–T). Cities along the west coast of Saudi Arabia (Jeddah and Mecca) and Yemen (Al Hudaydah) experience a substantial number of hours of threshold exceedance (Table 2), but the signal quickly diminishes off-shore. Land areas around the Persian Gulf experience heat stress with penetration deeper into the interior occurring in the United Arab Emirates, namely Dubai, and southeastern Saudi Arabia. In North America, heat stress signals begin to appear at 3 °C GMST warming but do not become substantial until the 4 °C level. In that future climate, hot-hours become concentrated along the Gulf of California coast in Mexico, in the Missouri and Mississippi River Valleys, the Gulf of Mexico coast, and the Atlantic seaboard with major cities like Chicago, St. Louis, Dallas, Houston, Washington, DC, and Philadelphia having considerable accumulations of hot-hours (Table 2).

The annual accumulation of hot-hours experienced by the world’s population begins to climb substantially in worlds warmer than 2 °C above the preindustrial baseline (Fig. 1F). In this study’s worst-case scenario of a 4 °C warmer world, around 2.7 billion persons will experience at least 1 wk of daytime (8 h) ambient conditions associated with uncompensable heat stress, 1.5 billion will experience a month under such conditions, and 363.7 million will be faced with an entire season (3 mo) of life-altering extreme heat (Fig. 1F). There is large variability in the annual accumulations of hot-hours within regions with geography as a controlling factor (Table 2). For example, in a 4 °C warmer world, Mecca, Saudi Arabia, only experiences a mean accumulation of 142.5 h of hot-hours while 500 miles to the south, Al Hudaydah, Yemen, is projected to experience 2,407 h. When divided into equal 8-h increments, this equates to 301 d of daytime threshold exceedance per year.

Based on the varying T_w thresholds described in the Introduction and Materials and Methods, we investigate threshold exceedance below and above T_a = 40 °C (humid vs. nonhumid) to delineate differences in environmental impacts on physiological heat strain. Globally, across all warming levels, the substantial majority of the person-hours of threshold exceedance are characterized by humid conditions. The number of global person-hours under uncompensable heat stress conditions experienced begins to increase rapidly at +2 °C GMST, going from 86 billion person-hours at that point to 349 billion person-hours at +3 °C and 1080 billion person-hours at +4 °C with the large majority being associated with humid heat (Fig. 3A). Again, the underestimation of heat stress exposure using the previous T_w = 35 °C upper limit is readily apparent (Fig. 3A vs. Extended Data Fig. 1B) as the number of global person-hours is two orders of magnitude higher at 4 °C warming when using the updated, empirical limits.

This also reflects the large populations in South and East Asia, which, in a world that is 4 °C warmer than the preindustrial period, is projected to experience around 608 and 190 billion person-hours of threshold exceedance in humid conditions, respectively (Fig. 3 B and C). These make up 84.5% and 90.5% of their total person-hours of threshold exceedance. While the magnitude of person-hours of threshold exceedance will not be as high in sub-Saharan Africa as in Asia, even at 4 °C warming when the region will experience around 88.6 billion person-hours, the number of hot-hours associated with T_a above 40 °C is negligible (Fig. 3D), confirming the future impact of increasing humidity on heat stress in the region (30).

However, nonhumid conditions leading to threshold exceedance are more significant in other parts of the world. The Arabian Peninsula is projected to experience an increase in nonhumid hours of threshold exceedance in a 2 °C warmer world. By the time of a projected GMST increase of 4 °C, nonhumid person-hours of threshold exceedance are expected to be greater than 11 billion, making up more than one-quarter of all person-hours of threshold exceedance (Fig. 3E). North America, which does not experience significant exceedance of the heat stress threshold until 3 °C GMST increase, experiences the largest proportion of nonhumid days of uncompensability with 23.8% of person hours at 3 °C and 31.8% at 4 °C (Fig. 3F).

These results indicate that a significant portion of the world’s population will experience—for the first time in human history—prolonged exposures to uncompensable extreme moist heat. Humans will struggle to adapt to these conditions in a warmer world as they will present widespread challenges across many aspects of food–energy–water security, human health, and economic development including in the world’s most populous and most vulnerable regions. Past research focusing on the T_w = 35 °C upper limit to human adaptability consistently highlighted the Middle East and South Asia as regions that would experience the brunt of deadly or intolerable conditions (23), as our study does. We also find that sub-Saharan Africa may be the true hotspot of T_w ≥ 35 °C in the future using our bias-corrected projections.

Our research shows that the footprint of life-altering heat using updated, empirically derived heat stress limits is vastly expanded. The additional regions most significantly affected are projected to be the equatorial and Sahel regions of Africa and eastern China given future warming scenarios that reach upward to 2 °C, a viable outcome by the end of the century, perhaps sooner, without drastic reductions in greenhouse gas emissions (31). Continued warming above 3 °C and 4 °C, respectively, causes North and South America, as well as northern Australia, to experience extended periods of dangerous heat. These results are another indication of the importance of restricting warming to 2 °C as suggested in the ratification of the Paris Agreement in 2015. Here, we note the differences between our study and that of a recent publication by Freychet et al. (32) and why this manuscript improves upon the former in terms of the analysis of T_w thresholds. Freychet et al. (32) found that the exposure to T_w of 31 °C (a statistically derived threshold in their study) was much more widespread than in this study, even at 1.5° and 2 °C in the Western Hemisphere and at 3 °C in parts of Europe. Overestimations in T_w are likely accrued through three main factors: 1) the use of non-bias-corrected CMIP6 data, 2) the use of daily averaged T_a and RH to calculate T_w, and 3) the use of Stull’s (33) equation to calculate T_w instead of Davies-Jones (34). The latter two methodological choices can each impart ∼1 °C errors in the calculation of T_w (35), easily large enough to inflate values to unrealistic values given certain amounts of warming.

It is important to note that these T_w thresholds were created for a young and healthy population, albeit unacclimatized, doing minimal activity (21). This means that thresholds are certainly lower for those who are more vulnerable to the heat, either through age, body size, and morphology, comorbidities, medication use, or any physiological or behavioral impairment to thermoregulatory capacity (10). Additionally, these thresholds did not take the impact of solar radiation on human thermal balance into account which will likely also modify thresholds (36). Thus, some components of the real impact on populations are likely to be underestimated in this study. The increases in heat stress evidenced in this study must lead to changes to daily life and the need for large-scale adaptation efforts to mitigate heat stress risk.

Nevertheless, T_w above the limits found by Vecellio et al. have already been observed in regions such as South Asia, sometimes for multiple hours at a time, with lesser effects on health (37). Longer-term exposure to more intense warm season humid heat would lend itself to higher critical wet-bulb limits due to physiological, thermoregulatory adaptation (which should be confirmed with future empirical testing). Hence, in tropical and subtropical regions of the globe, where people are acclimated, our exposure metrics may overestimate vulnerability for acclimated healthy, hydrated young adults doing no work, but further study is required to establish the risk profile for other individuals and populations. The outlook in these regions is further dimmed by the fact that higher, less conservative thresholds are still breached with each degree of warming.

However, unacclimated and unacclimatized populations bear larger risks during acute extreme heat events because of their vulnerability to unfamiliar ambient conditions. Physiological acclimation to heat stress takes around 2 wk and is lost within a week after the stimulus is removed (38–40), so the infrequent heat waves which characterize mid-latitude climate dynamical regimes are unlikely to drive the kind of physiological adaptations supported by the tropical dynamical regimes. Behavioral, technological, and societal adaptation mechanisms are more likely solutions in those regions and also in higher-income tropical regions such as Singapore. This study does not directly address such adaptation and vulnerability reduction measures, but this analysis can help identify the scope and magnitude of the need for such measures.

At 4 °C warming, our worst case scenario, Al Hudaydah, Yemen, would experience threshold exceedance for 300 d a year assuming 8 h a day of the worst heat. Again, assuming even spread, every day in the year would experience at least 6 h of threshold exceedance. This would obviously be nonconducive to outdoor activities, occupational or recreational, for any extended period of time. Reliance on indoor air conditioning would become necessary to remain heat-healthy. These periods of time correlate with projected core temperatures increases associated with the clinical definition of heat stroke (25, 41). If mitigation efforts are not taken, it is likely that heat-related deaths will significantly increase due to both direct and indirect effects of the heat. However, even if indoor cooling is accessible, a quality-of-life shift would occur since a majority of a person’s time will have to be spent inside for the sake of their health. Physiological adaptation is likely in particular geographic regions up to certain points given the outcomes of recent heat wave events. During an extended period of extreme heat in Pakistan in the spring of 2022, T_w values reached as high as 33 °C in Jacobabad, sometimes referred to as the “hottest place on Earth,” on multiple days. Extreme dry-bulb temperatures of 50 °C also affected the Indian subcontinent during this extended heatwave. However, only 90 deaths were reported in the immediate aftermath, indicating that complex relationships between climate, health, and other socioeconomic factors must be taken in concert to develop a fuller understanding of heat-health (42).

While physiological adaptation to the heat will be one of the driving factors in the determination of the habitability of regions in the future, other adaptive measures will also need to be developed. Results showed that a large majority of the hours of uncompensable heat stress will occur at T_a less than 40 °C (50% RH or above), providing evidence that humid heatwaves will be an increasing danger in future climates (27, 43). However, as noted, climate models have a proclivity for underestimating extreme events, meaning extreme hot, dry heatwaves, perhaps exacerbated by compound local drought and subsequent land-atmosphere interactions, may be a larger factor in future heat stress than what current projections show (44, 45). Given that humid heat dominates projections and is an environmental limit to physiological adaptation, the proliferation of air conditioning, as discussed above, will be the most consequential path to ensuring livability (46, 47). This switch to widespread air conditioning will be further motivated by the fact that swamp coolers—the cooling method of choice for many lower-middle-income regions—become ineffective at high wet-bulb temperatures (48).

However, air conditioning use imparts its own contribution to exacerbating climate warming through its greenhouse gas contribution (49) and a direct sensible effluent heat flux into the local environment and an evaluation of the trade-offs will be an important future consideration. The design of future, more resilient, cities will need to take into account the type of heat it will face to maximize efficiency and utility (50, 51). While features such as green spaces, water features, and increased shade may work in regions where nonhumid threshold exceedance is more common (i.e., North America, Middle East, Australia), these factors may not be as effective or, when considering resultant added humidity from green and blue infrastructure, actually exacerbate heat stress in more humid environments (52).

Many other socioeconomic and cultural factors will also play a role in the future adaptation to heat as they have throughout history. Writing from an archaeological perspective, Rivera-Collazo (53) notes the spectrum in human responses to climate change in past civilizations, highlighting local, community-based perception to risk and subsequent adaptive capacity and mitigation efforts. They highlight specific examples such as food availability and shelter infrastructure as ways to measure social change in response to climate change. Differential vulnerability to climate change will exist (54), and gradients of adaptability (physiological, behavioral, and cultural), habitability, and survivability to extreme heat in a changing climate already do and will continue to exist on global, regional, and local scales (55) and are linked closely with socioeconomic status (56). It is important to note that a majority of the population projected to be significantly affected by the heat live in the Global South and other countries that have had the smallest effects on anthropogenic climate change but will be disproportionately harmed by its effects (57). Those who can afford to beat the heat will do so, and those who cannot will be forced to break with cultural customs to maintain a relative level of livability, but social change will be experienced by all exposed to these levels of dangerous heat stress. Hazard, risk, and vulnerability must be evaluated in tandem to understand specific health outcomes stemming from the more frequent crossing of critical environmental limits to heat stress compensability. However, it is indisputable that serious health complications on a large scale, up to and including death, will occur where environmental conditions associated with uncompensable heat stress become more frequent and longer-lasting unless behavioral or technological intervention is implemented. It is an inevitable consequence of energy balance that unless environmental conditions are changed, the positive internal heat balance must eventually lead to heat stroke. One does not need to run those physiological experiments to their unethical conclusion to understand that end result. Beyond the health impacts to the individual exposed to extraordinary heat, the magnitude to which spillover effects take place is unknown and questions as to whether these regional climate risks translate into a larger-scale climate-induced catastrophe (58). It must be the work of multidisciplinary teams of physical, health, and social scientists, especially those with specific, indigenous knowledge of those who will experience the highest risk of extreme heat due to climate change, to quantify these gradients of adaptability and survivability and provide context for just climate futures for all.

This research was supported by the National Institute on Aging Grant T32 AG049676 to The Pennsylvania State University, NIH Grant R01 AG067471; NASA FINESST Grant 80NSSC22K1544; NSF 1805808-CBET Innovations at the Nexus of Food, Energy, and Water Systems (INFEWS: US–China): A multiscale-integrated modeling approach to managing the transition to sustainability; NSF 1829764-OAC CyberTraining:CIU:Cross-disciplinary Training for Findable, Accessible, Interoperable, and Reusable science.