Prediction of drought trigger thresholds for future winter wheat yield losses in China based on the DSSAT-CERES-Wheat model and Copula conditional probabilities

Predicting the risk of diminished wheat yields caused by drought under future climate change climate is essential for the long-term sustainability of agriculture. Although studies have explored the relationship between drought and crop yield loss, the precise thresholds triggering yield losses in the future remain unclear. In this study, we established a conditional probability framework for drought trigger thresholds at various yield loss levels in China’s winter wheat regions in the future based on copula functions. The primary drivers influencing the dynamics of drought thresholds were evaluated using a random forest model. The results revealed that the projected drought thresholds for the baseline period (1981–2020), near future (2021–2060), and far future (2061–2100) ranged from –2.1 to –1.2, –0.8 to –0.6, and –1.2 to –1.0, respectively, implying that the drought thresholds for winter wheat yield loss in the future firstly rises and then declines. This trend was primarily due to the increased contribution of precipitation (P_re) (from 24.0% to 31.5%) to the drought threshold in the far future, coupled with a decrease in the contribution of temperature (T_mean) (from 37.1% to 30.4%). This shift suggested that the increased P_re might alleviate the adverse effect of high temperature on yield in the future. The average drought thresholds for yield loss were higher in the Southwest (–1.0 to –0.6) and Xinjiang (–1.1 to –0.7) winter wheat regions, where mild drought occurrences led to a 30% yield loss (70th percentile). T_mean was the primary driving factor for the dynamic changes in future drought thresholds. The research findings provide scientific guidance for future agricultural water resource allocation and drought risk management.