Abstract
Saturated hydraulic conductivity is one of the key parameters in soil physics and hydrological modeling. This study explores the use of Support Vector Machine (SVM) and a nonlinear statistical regression approach for the purpose of predicting the field saturated soil hydraulic conductivity (K
field
) of sandy soil based on basic soil properties of saline and alkaline soil data sets. Considering the significance of soil properties, both methods used the following levels of input soil data, which are easily measurable in the laboratory: hydraulic conductivity, clay/silt ratio, liquid limit, hydro carbonate anions, chloride ions, and calcium carbonate content. The influence of three kernel functions (linear, radial basis and sigmoid) on the performance of the SVM model was investigated. An adaptive genetic algorithm is used to determine the optimal free parameters of the SVM models. The results indicated that the SVM with the RBF model has better accuracy compared to the linear- and sigmoid-based models. The RBF model performed satisfactorily with a modeling efficiency of 0.972 and a correlation coefficient of 0.976. According to all of the performance measures, the different SVM models are a powerful tool and have better performance than statistical regression models. The excellent performance of the SVM with the RBF model demonstrated its potential to function as a useful tool for the indirect estimation of K
field
to assess maximum obtainable prediction accuracy.