Multivariate linear modeling of soil properties using soil formation factors

Abstract

Parent material, climate, topography, time, and living organisms all play a role in soil formation. This study aimed to investigate the relationship between these factors and soil physicochemical properties for rapid management of watersheds and Multivariate linear modeling of soil properties using soil formation factors. Four watersheds of Qazvin province were studied, namely Juyank, Madan, Niyarak, and Plangeh. First, the working unit map was prepared using a geographical information system (GIS) on a scale of 1:25000 based on photogeological investigations and field visits. Then, 101 soil samples were collected and EC, pH, organic matter percentage, sand fraction, silt proportion, and clay percentages were measured. The relationship between soil formation factors and properties was investigated using linear multivariate regression in three methods of Enter, Forward, and Stepwise. Therefore, the 5 soil-forming factors were the independent variables, and the soil's physical and chemical characteristics were set as the dependent variables. The accuracy of these models was validated using two statistics of explanation coefficient and mean squared error (RMSE). The results demonstrated that the highest R2 value of 0.78 with an RMSE of 0.56 was associated with the relationship between pH and soil-forming factors (regression by the Enter method). There was less than a 50% correlation between silt, clay, and electrical conductivity with soil-forming factors.