Unraveling nonlinearities in daily-scale SPEI construction for improving extreme drought detection across China

High-temporal-resolution drought monitoring is essential for identifying short-duration and rapidly developing drought events that may be obscured at the conventional monthly scale. However, extending the Standardized Precipitation Evapotranspiration Index (SPEI) from monthly to daily resolution is not simply a matter of increasing temporal resolution, because potential evapotranspiration (PET) estimated by the Penman–Monteith (PM) method responds nonlinearly to meteorological forcing. Consequently, PET calculated from daily meteorological inputs and subsequently aggregated is not equivalent to PET calculated directly from temporally aggregated inputs, potentially introducing systematic biases into daily-scale drought assessment. Here, we quantify this temporal-scale nonlinearity of PM-based PET across China and evaluate its influence on daily-scale SPEI construction and drought identification. The results show a pronounced seasonal dependence, with substantially stronger nonlinear effects during the cold season than in summer. This temporal-scale dependence propagates into SPEI and leads to clear seasonal variations in probability-equivalent drought thresholds, with lower thresholds in summer and higher thresholds in winter. Applying fixed thresholds across seasons therefore produces systematic false detections and missed detections, particularly for severe and extreme drought events. To address this problem, we develop a daily-scale SPEI framework that combines PET calculated from daily meteorological inputs with probability-equivalent threshold mapping, thereby preserving the nonlinear response of PM-based PET while maintaining statistical consistency with conventional monthly SPEI drought categories. The proposed framework increases the average drought-detection skill score from 45.2% for the conventional monthly-scale approach to 79.6% and provides a clearer representation of the spatial occurrence of high-frequency extreme drought events. These findings demonstrate that the transition from monthly to daily SPEI involves not only a change in temporal resolution, but also changes in the underlying PET response and drought-classification criteria. Recognizing this scale dependence is essential for developing physically informed and statistically comparable daily drought indices, particularly for monitoring short-duration and rapidly evolving drought events.