Agricultural flash drought risk assessment under climate change based on crop growth stages

Study region Bengbu City, located in the middle reaches of the Huai River Basin, China. Study focus Flash droughts develop rapidly and pose serious threats to agricultural production. However, conventional drought indices based on fixed temporal scales are often insufficient to capture crop water supply–demand dynamics during different growth stages. This study proposed a crop-growth-process-based framework for agricultural flash drought (AFD) risk assessment by integrating DSSAT-CERES-Maize simulations, BCC-CSM2-MR climate projections, and growth-stage-specific water deficit identification. The framework was used to quantify AFD duration, severity, joint occurrence risk, and yield impacts under historical and future climate scenarios. New hydrological insights (1) Historical AFD events showed evident stage-dependent variation, with a mean annual cumulative severity of 9.54 (dimensionless, summed across drought-affected DITs) during 1960–2019 and the highest occurrence in the milking stage (S4). Under SSP5–8.5, AFD duration during S3 decreased by 8.33%, while severity increased by 3.44% around 2080. (2) Yield losses differed markedly among AFD types, and events involving the reproductive stages (S3 and S4) produced the highest yield-loss risk, with historical reductions of 13.98%-39.80% and future maximum reductions exceeding 80% under SSP5–8.5. (3) Pathway analysis identified maximum temperature (T max ) as the main climatic driver of yield reduction, with the duration of S3 and S4 acting as key mediating variables.