Temperature and humidity based projections of a rapid rise in global heat stress exposure during the 21st century

Abstract

As a result of global increases in both temperature and specific humidity, heat stress is projected to intensify throughout the 21^st century. Some of the regions most susceptible to dangerous heat and humidity combinations are also among the most densely populated. Consequently, there is the potential for widespread exposure to wet bulb temperatures that approach and in some cases exceed postulated theoretical limits of human tolerance by mid- to late-century. We project that by 2080 the relative frequency of present-day extreme wet bulb temperature events could rise by a factor of 100 - 250 (approximately double the frequency change projected for temperature alone) in the tropics and parts of the mid-latitudes, areas which are projected to contain approximately half the world's population. In addition, population exposure to wet bulb temperatures that exceed recent deadly heat waves may increase by a factor of five to ten, with 150 - 750 million person-days of exposure to wet bulb temperatures above those seen in today's most severe heat waves by 2070 - 2080. Under RCP 8.5, exposure to wet bulb temperatures above 35°C - the theoretical limit for human tolerance - could exceed a million person-days per year by 2080. Limiting emissions to follow RCP 4.5 entirely eliminates exposure to that extreme threshold. Some of the most affected regions, especially Northeast India and coastal West Africa, currently have scarce cooling infrastructure, relatively low adaptive capacity, and rapidly growing populations. In the coming decades heat stress may prove to be one of the most widely experienced and directly dangerous aspects of climate change, posing a severe threat to human health, energy infrastructure, and outdoor activities ranging from agricultural production to military training.

Figure 1:

Top panel (a-c): changes in annual maximum air temperature in 2060 – 2080 relative to 1985 – 2005 under RCP 4.5 (a) and RCP 8.5 (b). Panel (c) shows the range in projected annual maximum temperature increase spatially averaged over land for both emission scenarios over all 18 CMIP5 GCMs. Bottom panel (d-f): same as (a-c) except for annual maximum wet bulb temperature. Air temperatures increase at a faster rate and have more spatial variability than wet bulb temperatures, in part due to the dependence of wet bulb temperature on humidity.

Figure 2:

The number of days per year which exceed the historical (1985–2005) mean annual maximum temperature (top row) and wet bulb temperature (bottom row) in 2060 – 2080. Maps show results under RCP 8.5 (see Supplementary Figure 8 for maps under RCP 4.5), and (b, d) show the variation with latitude of the number of days per year under both RCP 4.5 and RCP 8.5, excluding water grid cells. Wet bulb temperatures exceed the historical mean annual maximum more frequently than air temperatures due to lower variability, especially in the tropics.

Figure 3:

Global population exposure to varying wet bulb temperature thresholds, in mean number of person-days per year. (a): Global mean annual exposure under RCP 4.5 and RCP 8.5 in 2070 – 2080 to wet bulb temperatures from 30 – 35°C. Error bars show the full range across 18 GCMs and five SSPs. Exposure to wet bulb temperatures above 30°C is reduced by several orders of magnitude in RCP 4.5 as compared to RCP 8.5. Right: mean global annual exposure to wet bulb temperatures exceeding 32°C, approximately the upper limit at which sustained physical labor is possible and above anything experienced in the historical climate. RCP 4.5 is shown on top (b), and RCP 8.5 on bottom (c). Exposure is separated into a population effect (constant climate but changing population), climate effect (constant population but changing climate), and a combined effect (result of changing population and changing climate). Total exposure is the sum of these three components. Error bars on total exposure show the 10^th – 90^th percentile range across 18 GCMs and five SSPs.