Rainfall intensities determine accuracy of canopy interception simulation using the Revised Gash model

Rainfall canopy interception plays a crucial role in rainfall redistribution and hydrological processes in forests. While previous studies have often focused on monthly or yearly time scales, the responses of forest canopy interception to different rainfall magnitudes, frequencies and intensities, particularly under changing climate conditions have been less explored. In addition, the performance of canopy interception models that capture the dynamics of rainfall interception under changing climate remains largely unknown. In this study, we conducted field observations across various tree species and used the Revised Gash model to evaluate the canopy interception under different rainfall intensities. Our findings revealed that the observed interception loss of gross precipitation were 26.1 %, 42.1 %, and 41.6 % for Pinus tabuliformis (PT), Quercus wutaishanica (QW), and Betula platyphylla (BP), respectively. The Revised Gash model accurately estimated canopy interception, with percentage errors of 0.4 %, 5.6 %, and 22.3 % for PT, QW, and BP, respectively. Interestingly, the model performed better for PT, especially under light to moderate rain, while its applicability for QW and BP were diminished under moderate to heavy rain. Overall, the Revised Gash model underestimated interception loss across different rainfall intensities, with more pronounced underestimations observed at higher rainfall intensities. Evaporation during and after rainfall contributed significantly to over 85.3 % of interception loss across three tree species. Sensitivity analysis highlighted that parameters including mean rainfall intensity, mean wet canopy evaporation rate, and canopy storage capacity were critical in influencing canopy interception simulation. These findings highlight the influence of rainfall intensity on the model's reliability in simulating interception loss and provide insights for forest hydrology research in semi-arid regions.