Initial soil moisture conditions dominate variation in event-scale propagation time from meteorological to agricultural drought

Agricultural drought, typically triggered by meteorological drought, poses a significant threat to crop production and regional water resources. Understanding the propagation from meteorological to agricultural drought is therefore crucial for improving drought early warning and agricultural water management. In this study, we investigated event-scale drought propagation in the Yellow River Basin using the Standardized Precipitation Evapotranspiration Index and Standardized Soil Moisture Index to characterize meteorological and agricultural droughts, respectively. Variations in drought characteristics (duration and intensity) across the entire drought event and during its development, persistence, and recovery stages were analyzed based on matched drought events. We further identified the dominant drivers and constructed predictive models of propagation time using the eXtreme Gradient Boosting (XGBoost) algorithm. The results indicate that agricultural droughts occur less frequently and with lower intensity but persist longer than meteorological droughts. Approximately 49.5 % of meteorological droughts propagate into agricultural droughts, with the one-to-one propagation type being dominant. Lengthening of duration and attenuation of intensity were observed during drought propagation across different drought stages. Initial soil moisture conditions emerged as the dominant driver of event-scale propagation time, followed by the timing of meteorological drought occurrence and its development duration. Based on the identified dominant influencing factors, a propagation time prediction model was constructed for each subregion using the XGBoost algorithm, enabling reliable prediction of propagation time. These findings underscore the critical role of initial soil moisture in regulating drought propagation, offering valuable insights for the development of agricultural drought early warning systems and the optimization of irrigation scheduling.