Probabilistic evaluation of agricultural drought using meteorological drought propagation mechanisms and Vine Copula in a compound drought framework

The propagation of meteorological drought (MD) to agricultural drought (AD) is crucial for early drought warning and mitigation, but its spatiotemporal propagation characteristics have not been fully explored. In this study, a framework is presented which adopted a three-dimensional identification method by accounting for spatiotemporal continuity to identify MD and AD and quantify their intensity. The compound droughts are detected through spatiotemporal matching and their spatiotemporal propagation potential is characterized by defining the compound drought duration ( CD ) and the compound drought affected area ( CA ). Then, a four-dimensional C-vine copula with three conditional factors, namely meteorological drought intensity ( I MD ), CD , and CA is constructed to assess the triggering probability of AD by MD. The results for the framework applied in Naoli River Basin, Northeast China, showed that 45 compound drought events were detected among all MD and AD events from 1982 to 2017. These compound droughts were characterized by strong spatiotemporal coupling between MD and AD. CD quantified drought persistence, whereas CA represented its spatial extent. Both CD and CA are essential variables for understanding the propagation of MD into AD. Under specific I MD , CA is positively correlated with triggering probability of AD, while CD exhibits a nonlinear hierarchical phenomena: CD < 3 months, the probability of extreme AD is close to 0; CD ≥ 5 months, the probability of extreme AD remains significantly high even with lower I MD . Consequently, proactive interventions during the initial stages of drought propagation is critical for effective drought management. The framework incorporating CD and CA can effectively capture the complex spatiotemporal processes of drought propagation and reduce biases associated with single conditional factor assessments. The framework could provide valuable guidance for AD probability assessment and disaster management in other regions. • Drought probability estimates lacking spatiotemporal information may be biased. • Quantifying drought propagation potential with compound drought duration and area. • Assessing the probability of agricultural drought intensity with a 4D Vine copula.