Identification of Soil Properties Influencing Some Soil Physical Quality Indicators Using Hybrid PSO-ICA-SVR Algorithm in Some Agricultural Land Uses of Kerman Province, Iran

This research was conducted in order to determine the importance of the effect of some soil properties on some soil physical quality indicators (SPQIs) in southeast Iran. To this end, 169 points from different locations in Kerman province were selected which had different agricultural land uses, and disturbed and undisturbed soil samples were taken from a depth of 0-20 cm. Soil properties such as soil pH, soil texture, field capacity (FC), permanent wilting point (PWP), soil bulk density (BD), electrical conductivity (ECe), calcium carbonate equivalent (CCE), and soil organic matter (SOM), and soil physical quality indicators (SPQIs) such as field capacity (FC), mean weight diameter (MWD), air capacity (AC), and relative field capacity (RFC) were determined. Subsequently, soil properties affecting SPQIs were ascertained using Particle Swarm Optimization-Imperialist Competitive Algorithm-Support Vector Regression (PSO-ICA-SVR) hybrid algorithm, and after the sensitivity analysis, the importance of each selected property in terms of its impact on SPQIs was recognized. The results indicated OM, BD, clay content, and CaCO3 in general, affect soil physical quality and this selection was carefully made by the hybrid algorithm. Furthermore, after modeling using the features selected by the SVR method and performing the sensitivity analysis, clay content and BD, among the selected properties, had the most significant effects on SPQIs. The highest value of the coefficient of determination was associated with MWD index (R-2 = 83.33) and the lowest error was related to AC and RFC indicators (RMSE = 0.031).