Site-Dependence Performance of Infiltration Models

Quantitative description of infiltration process is crucial for modeling soil water status as well as many other applications related to surface-subsurface hydrology. The importance of infiltration process has lead to develop several theoretical and empirical infiltration models. However, the applicability of these models is strongly subjected to soil spatial variability. In this study, the performance of Kostiakov, Mezencev, Horton and Philip infiltration models under different study sites were evaluated. For this purpose, the infiltration data of 123 sites with different soil series were obtained. The parameters of these models were then obtained, using sum of least squares error method. To quantify prediction accuracy of the models, the mean error (ME), root mean square error (RMSE), mean absolute of mean error (MAME) and Pearson correlation coefficient (r) were calculated. The results indicated that the performance of Kostiakov, Horton and Philip models were highly related to the location in which the data were collected. While the performance of Mezencev model was not site-dependent, the performances of all other models were varied with different soil textures. It was then concluded that Mezencev model can provide the best site-independent performance compares to Kostiakov and Philip models. Copyright Springer Science+Business Media B.V. 2009