Figure 3: Geographical distribution of deadly climatic conditions under different emission scenarios.

a–d, Number of days per year exceeding the threshold of temperature and humidity beyond which climatic conditions become deadly (Fig. 1b), averaged between 1995 and 2005 (a, historical experiment), and between 2090 and 2100 under RCP 2.6 (b), RCP 4.5 (c) and RCP 8.5 (d). Results are based on multimodel medians. Grey areas indicate locations with high uncertainty (that is, the multimodel standard deviation was larger than the projected mean; coefficient of variance >1). The expected lower number of deadly days at higher latitudes (Fig. 4) may help explain the large variability among Earth System Models in the projected number of deadly days at higher latitudes³¹ (for example, in the case for New York (illustrated in Fig. 4j) the one model projects nine deadly days by 2100; yet any other model projecting 18 days will double the variability). The uncertainty presented in this figure should be interpreted with that caution in mind.