Field investigation and numerical modeling of hydrological processes at a hillslope catchment in Hong Kong

Abstract

Motivated by mitigating flash flood-related damages, this study conducts a comprehensive exploration of hillslope hydrological processes, which includes data collection and experimental measurement, analysis of the rainfall-runoff features and modeling hydrological processes. To study the temporal and spatial variation of TF (throughfall), rainfall data recorded at eight rain gauges near and under a 6-m-tall multiple-layer deciduous tree (Macaranga tanarius) are analyzed. This study reveals that the leaf convergence and divergence can significantly influence TF temporal and spatial variations. For some rain gauges, the TF values can be larger than the GR (gross rainfall) for 30% of total rainfall events.

Using three classical baseflow separation methods (namely, the straight line method, inflection point method and one parameter filter method), this study computes the direct runoff volumes for 24 storms. Their runoff coefficients (which are the ratios of direct runoff volumes to the total rainfall volumes) are obtained. The analysis indicates that direct runoffs can contribute about 82% of flood peak discharges, but the average runoff coefficients are less than 4%.

To make use of the observed hydrological data, a numerical model, TOPMODEL, is used to simulate the hydrological processes. Further, to determine the model parameters, several field surveys have been conducted to collect land cover, soil properties, and hydraulic features of the catchment. Model parameters for the catchment are calibrated using a global optimization method, namely the SCE-UA (shuffled complex evolution-The University of Arizona), and Nash-Sutcliffe efficiency (E) as the objective function. To enhance TOPMODEL, this study confirms that the interflow process is important in improving runoff simulation.