Combined effects of rainfall types and antecedent soil moisture on runoff generation at a hillslope of red soil region

Abstract

AbstractThe runoff threshold is one of the important parameters in hillslope hydrology. Information on the combined roles of rainfall types and antecedent soil moisture in the runoff generation threshold is still scarce. Therefore, we linked the runoff generation behaviours of a 150‐m hillslope with rainfall types and antecedent soil moisture conditions during a 4‐year observation period between January 2017 and December 2020 in the red soil region of China. We found that the surface runoff of a hillslope could be modelled with a relatively high accuracy using HYDRUS 1D. Rainfall event types, classified into four types (I, II, III and IV) using a machine learning based k‐means clustering algorithm, significantly affected the modelling accuracy of runoff. The prediction accuracy decreased as the level of rainfall type increased, with decreasing from 0.93 for type I to 0.41 for type IV and RMSE increasing from 0.0082 to 0.025 accordingly. The scope of the runoff threshold, represented by antecedent soil water storage of 0–30 cm and event rainfall amount (ASW+P), ranged between 97 and 139 mm. That is, when ASW+P was lower than the lower limit, no runoff occurred in all of the rainfall events, and runoff would be generated without considering the hydrological conditions if ASW+P was higher than 139 mm. The driving factors of runoff generation were strongly dependent on the classified rainfall types. Hourly maximum rainfall intensity was the main contributor to runoff generation for the relatively slight rainfall events (i.e., types I and II). The rainfall amount, however, played a dominant role for the rainfall types with larger amounts and intensities (i.e., types III and IV). These findings provide insights into the method of determining runoff thresholds as well as the inner mechanisms of runoff generation behaviours at the hillslope scale under regions with similar climatic and hydrological conditions to the present study.Highlights Heavy rainfall events were the main error sources of runoff modelling. The threshold of runoff generation was identified by the combination of antecedent soil moisture and rainfall. Threshold scope of runoff generation ranged between 97 and 139 mm. Rainfall classification is valuable in understanding runoff generation behaviours. Runoff driving factors depended on rainfall types.