2.1 Data

We used bias-corrected and downscaled daily temperature and precipitation data from the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) (Thrasher et al., 2012, 2022) for the period 1976–2100. Thirty-two NEX-GDDP-CMIP6 Global Climate Models (GCM) were used, driven by historical forcings (1976–2014) and the SSP5-8.5 scenario (2015–2100) (O’Neill et al., 2016). While this high-emission scenario is widely used in climate studies, it is worth noting that its future plausibility has been questioned (Hausfather & Peters, 2020). A single ensemble member was selected from each of the 32 GCMs. In addition, we used 10 CMIP6 GCMs covering the same historical and SSP5-8.5 periods. These data provided the higher temporal resolution (3 hourly) necessary to compute daily maximum wet-bulb temperatures. Detailed information about the 32 NEX-GDDP-CMIP6 GCMs and 10 CMIP6 GCMs is presented in Table S1 of Supporting Information S1.

2.2 Temperature-Sensitive Phenomena

3.1 Theory

Here, we propose a theory to explain how a relatively uniform global warming may result in large spatial variability of regional impacts. This theory, illustrated in Figure 1, applies to a wide range of climate-sensitive phenomena, including yield of food crops such as sorghum, transmission of infectious diseases such as malaria, prevalence of mild weather as characterized by the number of outdoor days, and deadly humid heat as measured by wet-bulb temperature conditions. For each of these phenomena and other similar ones, thresholds of temperature define optimal ranges that are ideal for occurrence, success, and dominance of the phenomena involved. Outside this optimal range the phenomena considered are limited by natural processes and fail to impress.

Figure 1 presents a schematic of an optimal range for a generic phenomenon in comparison to 4 examples of different regional climates. Each climate is characterized by the probability distribution of the spatiotemporal variability in surface temperature: C1, C2, C3, and C4. In the region characterized by C1, the temperature is too cold, much colder than the optimal zone. A significant increase in temperature may not bring C1 close enough to the optimal zone or result in any significant regional impact on the phenomenon considered. On the other hand, C4 is too warm, and any increase in temperature may have a minimal impact on this phenomenon. The story is different for C2 and C3. The regional climate C2 is slightly colder than the optimal zone. A significant warming of C2 pushes this regional climate into the optimal zone resulting in significant impacts on the phenomenon considered. Conversely, the regional climate C3 offers the ideal conditions for the phenomenon considered under current conditions. A significant warming of C3 pushes this regional climate out of the optimal zone resulting in significant impacts on the same phenomenon, just like C2 but of the opposite sign.

3.2 Examples of Regional Impacts

The spatially heterogeneous response of sorghum yield to global warming strongly supports the proposed theory of regional impacts (Figure 2). Figure 2 illustrates the changes in temperature distribution over the growing season for major sorghum-producing countries. Given the optimal temperature range (24–27°C) for sorghum growth, we project disproportionate yield changes using state-of-the-art CMIP6 GCMs under the high-emission scenario (SSP5-8.5). Historically, major sorghum-producing countries like Nigeria, India, Brazil, and Mexico experienced temperature distributions centered around the optimal range. The climatic conditions in these countries are thus favorable for sorghum growth, possibly contributing to high yields during this period (Khalifa & Eltahir, 2023). However, rapid increases in temperature, as a consequence of future climate change, are expected to result in temperature distributions falling outside the optimal range by the end of this century. As a result, significant reductions in sorghum yields are expected in many regions, with particularly steep declines anticipated in Sudan. Conversely, Ethiopia, historically characterized by its cooler conditions relative to other sorghum-producing regions, is expected to become more suitable for sorghum production as temperatures rise. Our findings indicate that the effects of climate change on sorghum vary considerably across regions, as illustrated by the marked disparity between Sudan and Ethiopia.

Malaria transmission further supports our hypothesis, as it is highly temperature-dependent, with optimal conditions occurring between 25°C and 28°C. Temperatures outside this range could reduce the malaria transmission rate. Figure 3 presents the changes in temperature distribution across malaria-endemic African countries. The current climatic conditions in West African countries, including Mali, Gambia, and the Central African Republic, are conducive to malaria transmission. This has led to a severe malaria burden in these countries, resulting in numerous cases and fatalities in the historical period (World Health Organization, 2014). However, rising temperatures due to climate change may offer a benefit to these regions. That is, the relatively uniform warming under the high-emission scenario could shift temperature distributions away from the optimal range for malaria transmission, potentially reducing future cases in West Africa. In contrast, East African countries such as Ethiopia, Uganda, Burundi, and Malawi, may face an escalation of malaria transmission as temperatures converge toward the optimal range.

The anticipated warming is projected to have disparate impacts on outdoor days worldwide (Figure 4). Figure 4 depicts the impact of relatively uniform global warming on the number of annual outdoor days. Developed countries located in the mid-latitudes and high latitudes, including major greenhouse gas emitters like the US, European countries, and Russia, are projected to see little change or a slight increase in outdoor days by the end of this century under the high-emissions scenario. In contrast, developing countries situated near the equator, such as Brazil, Nigeria, and India, which contribute relatively less to global greenhouse gas emissions, are projected to undergo a substantial decline in outdoor days. This disparity can be attributed to differences in baseline climatic conditions between the Global North and Global South. That is, the Global North, with historically cooler climates, is expected to become more conducive to outdoor activities, such as walking, jogging, and cycling, as future temperatures over this region shift toward conditions more suitable for outdoor activities. Conversely, the Global South, with historically warmer climates, will become increasingly less suitable for outdoor activities as temperatures shift away from the optimal range. These results provide compelling evidence of climate-related disparities and exemplify the localized impacts of global climate change.

Humid heat extremes also serve as an example of the regional impacts of climate change. Our findings, as depicted in Figure 5, reveal that the shift toward warmer temperatures under the SSP5-8.5 scenario results in significant regional disparities in heatwave characteristics. Warmer regions will become more susceptible to heatwaves, while relatively cooler regions will remain less exposed to humid heat extremes. To underscore the heterogeneous nature of climate change impacts, we selected three densely populated regions in the tropics and three in the mid-to high-latitude Northern Hemisphere. Model projections suggest that tropical regions will experience more frequent and intense heatwaves. For instance, cities like Dubai, Dhaka, and Shanghai, currently with a few days or no days at all exceeding 31°C wet-bulb temperature, are projected to experience severe heatwaves. In contrast, mid-to high-latitude cities like Seattle, Paris, and Stockholm are expected to remain less exposed to humid heat extremes than low-latitude cities, even though temperatures will rise.