Evaluating the influential variables on rainfall interception at different rainfall amount levels in temperate forests

Rainfall interception (RI) by forest canopy is an important part of the ecohydrology cycle, influenced by rainfall characteristics, weather conditions, forest structures and their interactions. Rainfall events with different amounts are likely to change the ways that rainfall interacts with forest canopies, and consequently vary the effects of influential variables on RI markedly. Thus, analyzing the influential variables on RI separately ac-cording to rainfall amounts need further attention. In this study, rainfall partitioning was measured in 60 plots of four forest types during June to September from 2017 to 2019. Weather conditions were obtained from four weather stations spatially representing all the study plots. Detailed canopy structural variables were retrieved from the terrestrial laser scanning. Then, boosted regression trees (BRT) analyses were performed to evaluate the contributing variables of RI at different rainfall amounts. As a result, a total of 77 rainfall events were measured, and 43, 19 and 15 events were classifled as light rainfall events (LR, gross rainfall < 10 mm), middle rainfall events (MR, 10 mm <= gross rainfall < 25 mm) and heavy rainfall events (HR, gross rainfall >= 25 mm). The average values of RI amounts (the difference between gross rainfall and the sum of throughfall and stemflow) for LR, MR and LR were 2.0 mm, 3.5 mm and 6.6 mm, respectively. The average RI proportions (the proportion of RI amount to gross rainfall) for LR, MR and HR were 74.8 %, 20.2 % and 17.1 %, respectively. Both RI amount and RI proportion were signiflcantly different among the rainfall amount levels (LR, MR and HR). BRT results showed that, for all rainfall amount levels together, RI amount was governed by rainfall amount, followed by canopy interception index (CII) and humidity. Whereas RI proportion was most influenced by CII, followed by rainfall amount, humidity and average canopy height (ACH). As expected, for a given rainfall amount level, rainfall amount was the most influential variable on RI amount. Naturally, the most influential variable on RI proportion varied with rainfall amount levels. For LR and MR, the most influential variables were CII; while it was humidity for HR. Furthermore, a variety of signiflcant variables (with the relative influence value higher than the averaged value), such as rainfall intensity, wind speed and ACH, intricately and complexly affected RI amount and pro-portion at each rainfall amount level. In addition, the Rainfall amount exerted a stepwise positive correlation with RI amount. CII showed a stepwise positive and humidity a negative influence on RI proportion, respectively. A comprehensive understanding of the influence variable on RI at distinct rainfall amount levels provides valuable insights into how forest canopies intercept rainfall and greatly supports understanding of canopy hy-drological processes in temperate forests.