Evaluation of Three Indirect Methods in Estimating Soil Water Characteristic Curve

In the present study, three methods of the inverse numerical solution, transfer function, and artificial neural network for estimating soil hydraulic parameters were evaluated. A Double-ring infiltration experiment was conducted in three sites with different soil textures with three replications. Disturbed and undisturbed soil samples were also collected from three depths (0−10, 10−30, and 30−60 cm) for each soil, and some soil physical properties were measured. In this study, HYDRUS- 2D/3D software was used for inverse estimating of hydraulic parameters. The accuracy and reliability of the predictions were evaluated by the mean difference (MSD, m3 m-3), the value of mean differences (MD, m3 m-3), the mean and the standard deviation of the root of mean squared differences (RMSD, m3 m-3) and the Pearson’s correlation coefficient (r). The results showed that inverse estimation of soil hydraulic parameters provided a reliable alternative method for determining the soil water retention curve at the field scale. The soil water retention curves obtained from the RETC fitting had very good correspondence with those derived from inverse modeling. The highest value of determination coefficient (R2) was observed between the measured and estimated volumetric moisture in the inverse numerical solution method (R2 = 0.9363). After that, the estimated volumetric moisture with Rosetta software (R2 = 0.8629) and the PTF of Dashtaki and Homayi (R2 = 0.8401) were respectively.