Tracking shifts in European drought hotspots

Abstract Drought is a complex hazard with far-reaching socio-economic and ecological impacts that are often inadequately captured by conventional indices based solely on hydroclimatic anomalies. To bridge this gap, this study develops novel impact-based combined drought indices (iCDIs) that directly relate hydroclimatic drivers to remotely sensed vegetation stress, here used as a proxy for observed agricultural impacts. The proposed framework employs a machine learning framework to optimize the selection of hydroclimatic predictors including precipitation, temperature, and soil moisture across distinct hydrological clusters of the European domain. This bottom-up machine learning-based approach outperforms traditional indices in reproducing observed vegetation responses across European river basins. Crucially, future projections of iCDIs under warming scenarios unveil a distinct northward shift in drought impacts, with Central Europe emerging as an increasingly vulnerable hotspot. Notably, this contrasts with patterns suggested by standardized indices, which typically emphasize Mediterranean regions as the primary areas of concern. This spatial reconfiguration of risk underscores the urgency of adopting impact-oriented drought diagnostics to support climate-resilient planning and adaptive risk governance in a rapidly changing climate.