Selection of a Suitable Soft Computing Model for Estimation of Soil Cation Exchange Capacity

Measuring of soil cation exchange capacity (CEC) is difficult and time-consuming. Therefore, it is essential to develop an indirect approach such as pedotransfer functions (PTFs) to predict this property from more readily available soil data. The aim of this study was to compare multiple linear and nonlinear regression, adaptive neurofuzzy inference system, and an artificial neural network (ANN) model to develop PTFs for predicting soil CEC. One hundred and seventy-one soil samples were used into two subsets for training and testing of the models. The model's prediction capability was evaluated by statistical indicators that include RMSE, R2, MBE, and RI. Results showed that the ANN model had the most reliable prediction when compared with other models. This study provides a strong basis for predicting soil CEC and identifying the most determinant properties influencing soil CEC in the north regions of Iran. Analytical framework results could be applied to other parts of the world with similar challenges.Abbreviations: ANFIS: Adaptive Neuro-Fuzzy Inference System; ANN: Artificial Neural Network; CEC: Cation Exchange Capacity; CV: Coefficient of Variation; FFBP: Feed-Forward Back-Propagation; FIS: Fuzzy Inference System; MBE: Mean Bias Error; MF: Membership Function; MLR: Multiple Linear Regressions; MNLR: Multiple Non-Linear Regressions; MLP: Multi-layer Perceptron; OC: Organic Carbon; PTFs: Pedotransfer Functions; R2: Determination Coefficient; RI: Relative Improvement; RMSE: Root Mean Square Error; SD: Standard Deviation