Background and Aim: Field capacity (FC) and permanent wilting point of the plant (PWP) are the most important points in the moisture characteristic curve. These parameters are very important for managing, planning and designing different irrigation systems and different
More
Background and Aim: Field capacity (FC) and permanent wilting point of the plant (PWP) are the most important points in the moisture characteristic curve. These parameters are very important for managing, planning and designing different irrigation systems and different agricultural management. The traditional methods to determine these properties are difficult and time-consuming. Pedotransfer functions (PTFs) have been widely used to predict soil hydraulic parameters in favor of expensive laboratory or field measurements. Rosetta is one of many PTFs and is based on artificial neural network (ANN) analysis which allows the estimation of van Genuchten water retention parametersand their uncertainties. This research was conducted with the aim of investigating the effect of different input levels of ROSETTA software in estimating FC and PWP.
Method: To conduct the present research, the data related to the physical characteristics of about 280 soil samples from different parts of Isfahan province were used. The studied soils classified into 8 textural classes including clay loam, silty clay loam, sandy loam, loam, clay, silty loam, sandy loam and sandy clay loam. In order to estimate the parameters of the van Genuchten equation three levels of the ROSETTA's inputs including 1) the use of textural class (TC), 2) the percentage of sand, clay and silt particles (SSC) and 3) the percentage of sand and clay particles and silt plus balk density (SSC+BD) were used. They were calculated by ROSETTA, through the van Gnochten equation and compared with the values measured in the laboratory using pressure plates. Accuracy and reliability of the estimated parameters were evaluated by, modified efficiency coefficient (E'), modified agreement index (d'), the mean and the standard deviation of the root of mean squared differences (RMSD), Normalized root of mean squared differences (NRMSD)and mean bias error (MBE).
Results: The results showed that in heavy textural soils such as clay (C), silty clay (SiC) and silty clay loam (SiCL), the first scenario (TC) is more accurate than the other two scenarios and FC values estimated with less error. On the other hand, in the loamy sand (LS) and sandy loam (SL) textures, the third scenario (SSC+BD) had the most accuracy in estimating the parameters of the van Genuchten model. Regarding the rest of the soil textures, the second scenario (SSC) had the highest accuracy in FC estimation. In terms of the correlation coefficient values between the measured and estimated FC values, respectively, the second scenario (SSC) (R2=0.5282), the first scenario (TC) (R2=0.3992) and the third scenario (SSC+BD) (R2= 0.4116) were placed. Unlike FC, the PWP values estimated in different scenarios did not show a clear dependence on the texture, but the highest correlation coefficient between the measured and estimated PWP values by ROSETTA software was observed in the second scenario (SSC), (R2=0.5185), followed by the third scenario (SSC+BD) (R2=0.4930) and the first scenario (TC) (R2=0.476).
Conclusion: In general, in the most of the studied soils, using the percentage of sand, clay and silt particles (SSC), as ROSETTA's input, had the lowest amount of error in the estimation of the parameters of the van Genuchten equation. Therefore, the estimated FC and PWP using second scenario (SSC) had acceptable accuracy compared to their measured values.
Manuscript profile
