%0 Journal Article
%T Spatial Prediction of Soil Surface Organic Carbon Using Spectral and Non-Spectral Factors (Case Study; Asuran Summer Rangeland, Semnan Province)
%J Journal of Range and Watershed Managment
%I دانشکده منابع طبیعی دانشگاه تهران
%Z 5044-2008
%A Nateghi, saeedeh
%A Khalifehzadeh, Rostam
%A Souri, Mahshid
%A Khodagholi, Morteza
%D 2021
%\ 05/22/2021
%V 74
%N 1
%P 177-188
%! Spatial Prediction of Soil Surface Organic Carbon Using Spectral and Non-Spectral Factors (Case Study; Asuran Summer Rangeland, Semnan Province)
%K Organic carbon
%K Topsoil
%K Landsat 8
%K factor analysis
%K multiple linear regressions
%R 10.22059/jrwm.2021.313256.1547
%X Soil organic carbon is one of the most important indicators of soil quality. The purpose of this study is to study the spectral and non-spectral behaviors of soil in order to estimate the organic carbon of topsoil using factor analysis and multiple regression methods in the semi-steppe rangelands of Asuran, Semnan province. Soil sampling was performed using stratified random sampling method. After creating a map of homogeneous units in the area, in each homogeneous unit according to its area, several sampling points were selected completely randomly. A total of 145 sampling points were collected. At each sampling point, a composite soil sample (a mixture of 9 observations) was taken. Soil organic carbon was measured using Valkyli-Block titration method. Data of 114 samples were used to calibrate the model and data of 31 samples were used to validate it. The results showed that the correlation of spectral variables obtained from Landsat OLI sensor with surface soil organic carbon is higher than non-spectral variables obtained from 1: 25000 topographic maps. Also, the results of factor analysis by principal component analysis with eigenvalues greater than one showed that the total cumulative variance explained by 14 variables was equal to 90.2%, which was explained by three factors. The regression equation generated by the three extracted factors had suitable potential for predicting surface soil organic carbon (R2 = 0.59). The root mean square error (RMSE) of the proposed model was calculated to be 0.3.
%U https://jrwm.ut.ac.ir/article_82314_28d2ae5d156bb209cbf4b9c34233f68d.pdf