Comparison of Random Forest and Kriging Models for Soil Organic Carbon Mapping in the Himalayan Region of Kashmir

The knowledge about the spatial distribution of soil organic carbon stock (SOCS) helps in sustainable land-use management and ecosystem functioning. No such study has been attempted complex topography land use Himalayas, which is associated with great heterogeneity uncertainties. Therefore, this digital mapping (DSM) was used to predict evaluate SOCS using advanced geostatistical methods a machine learning algorithm Himalayan region Jammu Kashmir, India. Eighty-three samples were collected across different uses. Auxiliary variables (spectral indices topographic parameters) derived from satellite data as predictors. Geostatistical methods—ordinary kriging (OK) regression (RK)—and method—random forest (RF)—were for assessing variability inter-comparison models their prediction performance. best fit model validation criteria coefficient determination (R2) root mean square error (RMSE) resulting maps validated by cross-validation. concentration varied 1.12 Mg/ha 70.60 Mg/ha. semivariogram analysis OK RK indicated moderate dependence. RF (RMSE = 8.21) performed better than 15.60) 17.73) while RK. it may be concluded that provides estimation SOCS; however, further selection choice auxiliary higher sampling density could improve accuracy prediction.