Estimation of Soil Infiltration in Agricultural and Pasture lands Using Artificial Neural Networks and Multiple Regressions

Common methods to determine the soil infiltration would need lots of time as well as costs. On the other hand، the existence of non linear behaviors in soil infiltration would make it difficult to be modeled. Regard to the difficulties of direct measurement of soil infiltration، using indirect methods for estimation of this parameter، has received attention in recent years. In spite of the existence of various theoretic as well as experimental equations، some other indirect methods such as artificial neural networks are used to estimate this soil phenomenon. Now a day، artificial neural networks are shown high efficiency in modeling non linear equations. In the present study، 200 soil samples were collected from Ghoshe location in Semnan Province. Half of samples were collected from agricultural lands and the other half were collected from pastures nearby lands. Some soil chemical as well as physical properties such as electrical conductivity (EC)، soil texture، lime percentage، sodium adsorption ration (SAR) and bulk density were considered as easy and fast obtainable features and soil permeability as difficult and time consuming feature. The collected data randomly divided in two categories of training and testing and they used for train and test of two artificial neural networks، multi-layer perception using back-propagation algorithm (MLP/BP) and Radial basis functions (RBF) and nonlinear regression model. Results of this research show high efficiency of artificial neural network compare with multiple regression and also MLP network was better than RBF network. Sensitive analysis was used to determine correlation between independed parameters and permeability.