Evaluating Efficiency of Some Artificial Intelligence Techniques for Modeling Soil Wind Erodibility in Part of Eastern Land of Urmia Lake

Prediction of soil wind erodibility through soil characteristics is an important aspect for modeling soil wind erosion. This study was conducted to compare the efficiency of multiple linear regression (MLR), artificial neural network (MLP), artificial neural network based on genetic algorithm (MLP-GA) and artificial neural network based on whale optimization algorithm (MLP-WOA) for prediction of soil wind erodibility in part of eastern land of Urmia Lake. In this research, 96 soil samples were collected based on a stratified random sampling method and their physicochemical properties were measured. Additionally, the wind erodibility of soil samples was measured using a wind tunnel. Among the 32 measured soil properties, four properties including the percentages of fine sand, size classes of 1.7-2.0, and 0.1-0.25 mm (secondary particles) and organic carbon were selected as the model inputs by stepwise regression. Result showed that the MLP-WOA was the most effective method for predicting soil wind erodibility in the study area regarding to the lowest RMSE (2.9) and ME (-0.11), and the highest R2 (0.87) and NSE (0.87) values; followed by MLP-GA, MLP, and MLR. Considering the high efficiency of MLP-WOA, This method can be used as a promising method for determination of soil wind erodibility in the study area.