Evaluation of flood timing and regularity over hydrological regionalization in southern Brazil

Abstract

Extreme floods are often devastating, directly affecting millions of people worldwide. Because of the lack of infrastructure and proper hydrological monitoring, flood-related hazards in developing countries are magnified. Therefore, regionalization approaches associated with flood indices stand out for allowing reliable design flood estimates and enabling the identification of their driving mechanisms. In this context, the present study aims to evaluate flood timing and regularity using directional statistics for hydrological regionalization. A robust methodological framework was proposed combining fuzzy logic algorithms and the seasonality index, considering maximum annual streamflow data for Southern Brazil. It was found that major floods take place between midwinter and early spring with no well-defined spatial pattern; however, a significant decrease in their regularity was observed westward. Based on these characteristics, two clustering scenarios were proposed. According to the heterogeneity measure, hydrologically homogeneous regions were formed for the entire study area, except for the mountainous region in the northeast. Four main flood-driving mechanisms were identified: frontal systems, maritime air masses, consecutive rainfall days, and antecedent soil moisture.