Prediction of Some Soil Hydraulic Properties Using Pedotransfer Functions

Field capacity (FC) and permanent wilting point (PWP) are efficacious in determining net irrigation water depth. However, direct measurement of these properties is tedious, time consuming and costly especially on large scale. Soil pedotransfer functions (PTFs) as the indirect methods can replace by the direct methods. In this study, performance of the six available pedotransfer functions on FC and PWP moisture content predicting was evaluated on 112 soil samples that were collected from the north and northeast regions of Iran. The Root Mean Square Error (RMSE) values of menioned available PTFs were changed between 0.05 to 0.17 and 0.03 to 0.13 in moisture prediction on FC and PWP points, respectively. Therefore new PTFs were developed by Multiple Linear Regression (MLR) and Artificial Neural Network (ANN) techniques based on soil properties (90 samples) and the results were validated on different soils (22 samples). The results showed that both MLR technique with assigning the RMSE values approximately 0.035, 0.01, 0.027 and 0.024 to predict soil moisture content on FC and PWP, total available water and specific yield and ANN technique with assigning the values approximately 0.013, 0.007, 0.015 and 0.013 to the same properties, evaluated in appropriate performance. The results also showed that using variables such as geometric mean and geometric standard deviation particle diameter, fractal dimension and air-entry suction, for the first one on input variables of PTFs, improved the accuracy of the results significantly, although accepting of this theory requires more studies.