Spatiotemporal evolution of rain-spell characteristics and multivariate extremes

Rainfall impacts are often governed not by single-day extremes but by the persistence, accumulation, and peak intensity of rainfall within continuous wet periods. To address this, an event-based framework is applied to analyze rain-spells across six rainfall thresholds (2.5,5, 10,25,50 and 100 mm) using three indicators: duration, magnitude, and peak rainfall. Their joint behaviour is investigated using a trivariate copula framework to estimate joint exceedance probabilities and return periods. The framework is applied to a multi-decadal rainfall dataset (1981–2024) for the Rupnarayan River Basin in eastern India, a monsoon-dominated basin characterized by strong spatial rainfall variability. The results reveal a distinct post-2000 hydroclimatic intensification, marked by longer and more intense rainfall spells. Rain-spell duration and magnitude increase across most parts of the basin, particularly at low-to-moderate thresholds (2.5–25 mm), while peak rainfall intensifies mainly in the southern and southeastern regions. Spatial variability also increases, especially for moderate thresholds, indicating greater temporal fluctuations in rainfall behaviour. The copula analysis further reveals a systematic reduction in joint return periods, indicating more frequent co-occurrence of long-duration, high-magnitude, and high-peak rainfall events. The proposed framework provides a transferable approach for multivariate rainfall analysis across diverse climatic regions.