Comparison of infiltration models to describe infiltration characteristics of bioretention

Bioretention is one of low-impact development measures, which widely used not only because it can reduce stormwater runoff total volume, decrease peak flow rate and delay peak flow time, but also can remove the runoff pollutants. Infiltration is an important hydrological process for bioretention to evaluate its runoff total volume reduction and pollutants removal. So, it is important to find an optimal infiltration model that can well describe the infiltration performance of bioretention. The Horton, Philip and Kostiakov infiltration models were selected to compare their accuracy when using for describe the infiltration characteristics of bioretention, and the errors between the different models simulate results and experiment results were assessed via the maximum absolute error (MAE), bias and coefficient of determination (R2). The experimental results showed that Horton model is fitting well and flexible under different experiment conditions, especially when the hydraulic head was 10 cm, with MAE of 0.50–0.81 cm/h, bias of 0.1–0.23 cm/h and R2 of 0.98–0.99. R2 of the Philip and Kostiakov models were all over than 0.87 at the initial infiltration period, but the model fitting accuracy decreased significantly with infiltration time elapse. Furthermore, the total runoff volume capture ratio and emptying time were advanced used to evaluate the flexibility of Horton model, and the Nash-Sutcliffe efficiency coefficients of them were over than 0.61 and 0.58, respectively. Therefore, the Horton model can be optimal selected to describe the infiltration process of bioretention and for its hydrological evaluation.