Assessing the Efficiency of Pedotransfer Functions for Estimating CEC in some Calcareous and Gypsiferous Soils of Yazd Province

CEC can be measured directly but its measurement is especially difficult and expensive in the Aridisols of Iran because of the large amounts of carbonates and gypsum in these soils. Pedotransfer functions (PTFs) provide an alternative to direct CEC measurement by estimating CEC from more readily available soil data. Estimation of costly measured properties from widely available or more easily measured basic soil properties by means of regression equations is referred to pedotransfer functions. To develop pedotransfer functions, we can use regression PTFs and neural networks. In the present study for estimating CEC, 60 surface soil samples (0-30 cm) were collected in 250×250 m grid from calcareous and gypsiferous soils of Yazd-Ardakan plain. Input parameters for estimating CEC in both mentioned models were: soil texture class and soil organic carbon content. Finally, result of prediction of CEC by regression PTFs and neural networks compare with empiric pedotransfer functions. In general, the neural network-based models were provided more reliable predictions (RMSE=1.04 and coefficient of determination=0.97) than the regression-based PTFs. Higher accuracy of ANN is because of non-linear correlation between dependent and independent variables.