Prediction of Cation Exchange Capacity in the Soils of Gilan Province Using Intelligent Models

Cation exchange capacity (CEC) is one of the most important characteristics of soils in relation to nutrient elements and water storage in the soil, as well as soil pollution management. CEC measurement is difficult and time-consuming. So, estimating it by use of soil readily available properties is good. In this study, intelligent model was employed and the parameters used were the physical and chemical properties of the soil such as particle size distribution, organic carbon, clay and sands content, phosphorus, nitrogen, PH and EC. The methods of artificial neural network (MLP), (RBF) and Adaptive-network-based fuzzy inference system (ANFIS) were used to assess CEC. Then, the ability of this method to predict CEC was investigated by using 250 soil samples in two groups: 80 percent for training and 20 percent for validation. To determine the accuracy of the model prediction of CEC, statistical indices including Mean Absolute Error (MAE), the coefficient of determination (R2), and Root Mean Square error (RMSE) were evaluated. The results showed higher efficiency of artificial neural network MLP compared to the other models with the values of MAE, RMSE, R2equal to 1.79, 2.54, and 0.8, respectively1. The sensitivity analysis performed on the input data to the model showed that organic carbon and the pH had the highest and lowest correlation with the cation exchange capacity. The results show that use of artificial neural network to estimate the soil cation exchange capacity is possible and can be used to facilitate the measurement, lower economic cost, and save time.