Abstract
Abstract This study explores the rainfall and soil moisture thresholds for runoff using a classification and regression tree (CART) algorithm for a large dataset (146 events) of hydrologic variables collected from a humid hillslope. The common CART structure for the runoff coefficient successfully separated rainfall events with zero runoff generation and further identified five distinct clusters with distinct runoff generation and hysteretic loop features. The total amount of rainfall was the primary driver of runoff, with the highest relative importance in the CART model. In contrast, the antecedent soil moisture and rainfall intensity helped explain the hysteretic loop features as well as exceptional events through hysteretic loop analysis. The multinomial logistic regression model was introduced to determine the probability of different hysteretic loop variations according to the rainfall characteristics. By identifying the distinctive hysteretic loop patterns between upslope and downslope regions, these findings demonstrate the sequential development of hydrological processes from zero runoff to extreme runoff events.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call