Prediction of Dexter’s Soil Quality Index Using Soil Physical, Chemical, and Geophysical Properties

Abstract

Abstract Dexter’s soil physical quality index (the S index), which is defined as the slope of the soil-water characteristic curve at the inflection point, can properly describe the soil physical quality. This study was conducted to derive a model to predict the S index using soil physical, chemical, and geophysical properties. For this, 72 soil samples were collected, and some of the physical, geophysical, and chemical properties of the soil samples were measured. The soil water content was measured using sand-box and pressure plate apparatus. RETC software was also used to determine the parameters of the van Genuchten equation. These parameters were employed to find the slope of the moisture curve at the inflection point as an indicator of soil physical quality. The decision tree analysis was applied to derive the proper model for predicting the S index using different combinations of chemical, physical and geophysical soil properties in 4 scenarios and the most important predictor(s) were determined in each scenario. The effective factors on the S index in the first and third scenarios included the moisture and the mean weight diameter of aggregates; while in the second and fourth scenarios in which some of chemical properties were replaced by geophysical and structural properties, the dielectric constant and mean weight diameter of aggregate were the most influential factors on the S index. The coefficient of determination of 0.86 was obtained as a result of correlation between the measured and predicted data in the first and third scenarios, while in the second and fourth scenarios, it was obtained as 0.87. Considering the values of %RMSE, the modeling in all four scenarios can be regarded as successful. However, the % RMSE value in the second and fourth scenarios was less than that in the first and third scenarios.