مهندسی منابع آب (Apr 2015)
Accuracy of Different Methods in Determination of the Van Genuchten,s Retention Curve Coefficients
Abstract
Soil water retention curve is an indicator of one of the most important soil hydrological properties that have high applications in various soil water issues, and it is widely estimated by the van Genuchten,s model. This research was performed to estimate the soil water characteristic curve by method of van Genuchten,s model. Thus for determination of Inverse solution, Two points model H5 of Rosetta program and two point-Rosetta methods were used. Using these coefficients, the performance and accuracy of these models in retention curve estimation were evaluated by correlation coefficient (R2), magnitude of absolute error (MAE), root mean square error (RMSE), Akaike information coefficient (AIC) and model efficiency (EF). The overall results showed that the statistical parameters of R2, EF, MAE (cm3/cm3), RMSE (cm3/cm3) and AIC for inverse solution were 0.997, 0.981, 0.013, 0.020, and -7344, respectively for Two point-Rosetta model were equal 0.983, 0.898, 0.026, 0.047 and -5754, for Two-point model were equal 0.98, 0.879, 0.027, 0.051 and -5595, respectively and for Rosetta were equal 0.978, 0.859, 0.039, 0.055 and -5440, respectively. According to these results, inverse method, two point-Rosetta and two-point had the highest accuracy in retention curve estimation, respectively. The Rosetta program had the lowest accuracy in this case. Comparison of methods for different textures showed that the inverse solution had the lowest error in all textures. Rosetta program, Two point-Rosetta and Two-point are recommended for coarse, Medium and fine textures, respectively.
