Fig. 7. Temporal evolution of annual outdoor days derived from CMIP6 models. Time series of annual outdoor days (day) derived from 29 CMIP6 GCMs under the historical and SSP5-8.5 scenarios. Thick solid blue line indicates an ensemble mean of CMIP6 models. Horizontal black and blue lines denote the 1976-2005 mean and the 2071-2100 mean, respectively. Difference (2071–2100 minus 1976–2005) in the number of outdoor days is represented in each plot. The background image was obtained from NASA Visible Earth.
Underlying mechanisms responsible for the north-south disparity in climate hazard have been investigated by studying changes in the probability distribution of temperature over six selected countries across all climate zones (Fig. 8). Modeling results show an evident shift of the probability distribution of temperature toward warmer temperature across the countries (Fig. 8a). The warming shift of the probability distribution of temperature induces a significant increase in outdoor days in the European Union and to a less degree in the United States. The latter enjoys a broader range of climatic zones than the former. In both regions, climate change causes fewer outdoor days during the warm season. However, an increase in outdoor days in the cold season compensates for this decreasing trend (Fig. 9). A large fraction of the population in the northern high-latitude regions, such as Russia and Canada, will likely see large increases in outdoor days, benefiting significantly from climate change. In contrast, the projected probability distribution of temperature in Brazil, Nigeria, and India, are likely to move away from the conditions of thermal comfort, limiting outdoor activities significantly in these large population centers of the South (Fig. 8). The results of Figure 8 show clearly that the north-south disparity is rooted in the background climatology of temperature, mainly the position of the probability distribution of temperature relative to the range of values used to define an outdoor day. Similar results and conclusions are obtained by analyzing projections by CMIP5 models (Fig. S5).