Regional differences in the impact of climate extremes on future global rice yield variability

While much research has focused on the negative impacts of climate extremes on rice yields, understanding how different types of extremes affect future yield variability across regions remains limited. In this study, we first used extreme gradient boosting to predict global gridded rice yields under three Shared Socioeconomic Pathways (SSPs) (SSP126, SSP245, and SSP370) for 2030–2079. We then examined differences in rice yield changes across countries with different income levels. Furthermore, we quantified the relative contribution of five key climate extreme indices (CEIs) to rice yield variability using a stepwise nonlinear regression model. A comparison with random forest and generalized additive models demonstrated the effectiveness of the proposed model. Finally, yield responses to CEIs were discussed for six continents and each country. The results indicate that: (1) variability in the five CEIs will explain ~38% of the global rice yield variability on average; (2) drought and precipitation extremes are projected to have significant effects on rice yields in ~96% of the grid area; (3) extreme drought is projected to impact rice yield variability in key regions by more than −10%. These findings provide fresh insights into how climate extremes will impact future global rice yields in different regions.