Evaluating Point and Parametric Spectral Transfer Functions for a Prediction of Soil Water Characteristics

Direct measurement of soil hydraulic properties is usually expensive and time consuming. Several attempets have been made to establish PedoTransfer Functions (PTFs)، that use readily available soil data، to provide soil hydraulic properties. The objective of the present study was to evaluate point and parametric Spectral Transfer Functions (STFs) as well as PTFs to estimate van Genuchten (vG) and Brooks-Corey (BC) parameters needed for predicting soil water characteristics’ curve. The derived STFs and PTFs were further evaluated with a number of available PTFs to find out if spectral variables can effectively improve the accuracy of soil water retention predictions. Consiquently، a number of 174 soil samples were collected and used to measure the spectral reflectance curves in visible، near-infrared and shortwave-infrared range، using a handheld spectroradiometer. Some physical soil properties and soil water contents at their specific matric potentials were recorded. Using a non-linear least square optimization method، the vG and BC parameters were then calculated. Stepwise multiple linear regression statistics was employed to derive point and parametric STFs and PTFs. The results indicated that the derived point and parametric STFs and PTFs are of higher accuracy (with mean RMSR of 0. 029 cm3 cm-3) in predicting soil water retention than the published PTFs (with mean RMSR of 0. 100 cm3 cm-3). Furtheremore، vG model was found to more accurately predict soil water content (mean RMSR of to 0. 034 cm3 cm-3) than BC model (mean RMSR of 0. 041 cm3 cm-3). The overall findings reveal that the derived STFs and PTFs provide almost similar results، but the STFs performance is somewhat more accurate at mid and high metric potentials.