Regionalization methods for PUB: a comprehensive review of progress after the PUB decade

Abstract

Abstract This paper presents an updated review of model-dependent regionalization methods in hydrology since the PUB decade, incorporating new regions and methodological advancements. Two categories of regionalization methods are discussed: distance-based and regression-based, with various modification approaches. Several factors affecting the accuracy of regionalization performance are identified, including hydrological models, climate characteristics, data availability, and regionalization techniques. The review concludes that distance-based regionalization methods with an output averaging option from multiple donor catchments are the most statistically reliable, and a threshold of 0.5–0.8 for donor selection is optimal for performance. The parsimonious hydrological model is also recommended, particularly in data-limited contexts. Other insights include the effectiveness of the ensemble concept and limited impact of prior classification. Additionally, it is found that the general performance difference between climatic classes is larger than between methods, and regression-based methods may have large uncertainties in tropical regions. Overall, this study provides practical guidance for improving regionalization studies and advancing the field of hydrology.