Prediction of soil saturated water content using evolutionary polynomial regression (EPR)

Abstract

The measurement of saturated water content (SWC) is necessary in the estimation of soil water retention and unsaturated hydraulic conductivity curves. In several studies, pedotransfer functions (PTFs) were developed to predict SWC. Among them, evolutionary polynomial regression (EPR) is one that can operate on large quantities of data in order to capture nonlinear and complex interactions between the variables of the system. In this study, the evolutionary data-mining technique was used to derive new PTFs and different methods were evaluated, such as the soil porosity method, Rosetta method, and others, for the estimation of SWC. For this purpose, 270 soil samples (3:1 ratio for development and validation) from three data sets were used. Among 190 PTFs provided by EPR, one equation with the highest accuracy and the least number of inputs was selected. The EPR predictions were compared with the experimental results as well as the PTFs proposed in previous studies. Comparison of the statistical indicators showed that the ‘proposed PTF’ and ‘porosity method’ are the best and worst methods for the prediction of SWC, respectively. Also, good predictions were achieved from the proposed approaches by the groups of Scheinost, Vereecken, and Williams.

Keywords:

• evolutionary polynomial regression
• Rosetta
• soil textural data
• pedotransfer functions