Behnaz Attaeian; Ali Badrestani; Saeid Khosrobeigi Bozchelui; Mohammad Mehdi Artimani
Abstract
Soil organic carbon as a key factor in soil stability and fertility is considered as one of the important environmental challenges in the context of climate change. The aim of this study was to determine soil organic carbon zonation in Gonbad paired-watershed, Hamedan province. In this research, the ...
Read More
Soil organic carbon as a key factor in soil stability and fertility is considered as one of the important environmental challenges in the context of climate change. The aim of this study was to determine soil organic carbon zonation in Gonbad paired-watershed, Hamedan province. In this research, the information of meteorology, soil science and erosion and sedimentation study of Gombad watershed was used, including the information of 49 profiles in the 0-15 cm soil layer. After collecting data, tests of normality (Shapiro-Wilkα test <0.05), homogeneity of variance, and then the relationship between independent variables and organic carbon were performed using Pearson's linear correlation in SAS software. Also, determining the most effective independent variable using multivariate analysis, PCA factor analysis was used in XlStat 2.1 software. In order to determine the distribution and amount of soil organic carbon in the Gonbad representative watershed, modeling using SVM support vector machine learning algorithms and RF random forest was used in R software.The results showed that 78.18% of soil organic carbon changes depend on four components. Clay and nitrogen percentage were selected as the most effective variables on soil organic carbon content, so that the first component of clay content explained 34% and the second component nitrogen explained 18% of variations. According to the results of the implementation of the SVM and RF Models, the SVM model with a CE factor of 0.86 and RMSE of 0.05 in the test stage is a more accurate model in this study.
