A novel method of estimating dynamic partial contributing area for integrating subsurface flow layer into GIUH model

Abstract

The partial contributing area (PCA) is a conceptual parameter proposed to approximately quantify the effective surface-runoff region which is directly resulting from the excess rainfall. Previous studies further applied the ratio of PCA, which can determine the separation between the surface-and subsurface-flow regions, to introduce the subsurface-flow mechanism in a geomorphology-based IUH model. The temporal distribution of the simulated hydrograph was found to be sensitive to the ratio of PCA, especially for the stage of flow recession. Therefore, this study focuses on devising a method to estimate the dynamic PCA depending on the initial streamflow, antecedent precipitation, and current infiltration rate, in which the soil type is considered. An artificial neural network model, called long short-term memory (LSTM), is established to provide adequate values of PCA by considering the aforementioned factors associated with the hydrological conditions. A geomorphology-based IUH model is subsequently implemented to demonstrate the importance of using the proposed methodology to seek a reliable PCA by comparing simulated hydrographs with observed discharges. The proposed methodology could be a promising way to avoid the assumption of constant PCA or the complex process of deriving time-varying PCA. Additionally, the results of this study showed that it can significantly ameliorate the performance in terms of relative error of simulated hydrograph as well as the overall similarity compared to the flow records of floods.