Fig. 2: Heterogeneity of power outages caused by natural disasters.

Notes: In a and b, we show the impacts of Natural disaster, Strong wind, Rainstorm, Cold wave, Geo hazard, Wildfire and Heatwave on power outages. In c and d, we show the heterogeneity impact of Natural disaster on power outages. The solid dots denote the point estimation coefficients and the lines refer to the 95% confidence interval of the coefficients in a–d. In e and f, we show the decomposition results of natural disaster impacts on power outages based on Shapley Values. The definition of summer is from May to October. The definition of winter is the remaining months. 0-10 represents counties with fewer than 10 natural disaster occurrences, whereas >10 indicates counties experiencing more than 10 natural disasters during the study period. The 1 and 0 in the X-axis in e and f represent the poverty and non-poverty counties, respectively. Our regression analysis dataset comprises data from 2245 counties across China. The significance of the regression coefficients is assessed using a two-sided t-test, which evaluates whether the coefficients are statistically different from zero. Source data can be found in sheets Supplementary Data 2–7 of file Supplementary Data 1–7.