Terrestrial laser scanning-derived canopy storage capacity improves the performance of the revised Gash model in temperate forests

Rainfall interception loss (I) by forest canopy is a crucial hydrological process in forest ecosystems, and thus its accurate modeling is essential for understanding water balance. The revised Gash model is commonly employed in I modeling; however, its performance is affected by the accuracy of canopy storage capacity (S), which is identified as one of the most sensitive parameters. Consequently, optimizing the estimation of S and then cascading application in the revised Gash model warrants further attention. In this study, we measured gross rainfall, throughfall, and stemflow for the larch (Larix kaempferi) plantation forest (LPF) and the Mongolian oak (Quercus mongolica) forest (MOF) in Northeast China in 2018 and 2019. Terrestrial laser scanning (TLS) was introduced to derive S (S_ex). S_ex was then compared with values calculated from two commonly regression-based methods (S_mean and S_mini). Finally, the revised Gash model was run using the three types of S, and the model performances were evaluated. As a result, I of LPF (27.9 %) was higher than that of MOF (20.1 %). For LPF and MOF, S calculated from S_ex was the largest (1.45 and 0.51 mm), followed in descending order by S_mean (0.98 and 0.32 mm) and S_mini (0.29 and 0.13 mm). Compared with models run with S_mean and S_mini, S_ex improved the model performance, regardless of whether the Penman-Monteith equation or a linear regression method was used to calculate the evaporation rate (another sensitive parameter of the revised Gash model). Moreover, the model using S_ex particularly enhanced the model's accuracy at middle and heavy rainfall levels. In conclusion, the TLS-derived S improves the model performance in temperate forests in Northeast China. Meanwhile, in contrast to previous studies, which emphasized the contribution of evaporation rate/rainfall intensity (E/R) in modelling larger rainfall events, this study suggests the role of S should not be overlooked.