Improving Infiltration Prediction by Point-based PTFs for Some Soils in Southern of Iran

Abstract

Modeling soil water infiltration at the field scale with ruler of calcareous, saline and sodic conditions is important for a better understanding of infiltration processes in these soils and future of infiltration modeling. The aim of the present study was to derive and evaluate soil water infiltration models for some calcareous, saline and sodic soils in Marvdasht plain, southern of Iran. The infiltration data was measured in 72 locations at the regional scale with 3 replications. In each location, the basic soil properties were also measured. The multiple linear regression (MLR) and feed-forward multilayer perceptron artificial neural networks (ANN) model were used to estimate cumulative soil water infiltration at different time. The results performance of water infiltration models such as Kostiakov, Kostiakov–Lewis, USDA-NRCS, Philip, Horton and Green-Ampt models according to the mean R2, ME, RMSE and SDRMSE indices for all soils showed the Kostiakov–Lewis model provided the most accurate predictions. Moreover, the results showed that the derived ANN models at different times with a R2 of 0.438-0.661 and a RMSE of 0.977-17.111 performed better than MLR model. There would be great interest to improve the cumulative soil water infiltration in site-specific soil utilization, management and protection of the environment by MLR and ANN methods.