Rethinking standardized drought indices for critical drought evaluation

ABSTRACT Drought evaluation is a key component of water resources management, commonly performed using standardized drought indices (SDIs) based on probabilistic standardization of different input variables. While widely accepted for their simplicity, SDIs involve uncertainties, particularly related to the selection of the observation period and the parameters of the probability distribution function. One often overlooked methodological aspect is the fitting process: it can be applied to the entire dataset (classical model – C) or separately for each month (dynamic model – D), resulting in 1 or 12 models, respectively. This distinction has not been thoroughly addressed in the literature. To fill this gap, this study investigates the differences between classical and dynamic SPI models across various timescales using statistical metrics, drought classification, drought characteristics, and drought events. The analysis was applied to the Antalya Basin in Türkiye using monthly precipitation data from 1969 to 2022. Results reveal that the dynamic model produces significantly longer drought durations at short and medium timescales. For instance, drought duration for C-SPI 6 ranged between 4 and 6 months, while D-SPI 6 ranged between 6 and 8 months – indicating up to a 50% increase. These findings highlight the importance of model choice for accurate drought assessment.