19 - Prediction of Soil Organic Carbon and its Mapping Using Regression Analyses and Remote Sensing Data in GIS and R

Abstract

Soil organic carbon (SOC) is a dynamic property of soil that epitomizes a vital part of forest and agriculture ecosystems. In this research, an integrated approach using satellite data, laboratory data, and field data was applied for digital SOC mapping (DSOCM) in Dulung river basin (India). A total of 48 soil samples were collected randomly from five soil cores of undisturbed topsoil samples at depths of 0–15cm for testing and a validating dataset. Remote sensing (RS)-derived vegetation indices (normalized difference vegetation index, renormalized difference vegetation index, modified soil-adjusted vegetation index, and modified nonlinear vegetation index) and field-measured SOC using linear and multivariate regression (MR) models were used to predict DSOCM. The statistical relationship between observed SOC and vegetation indices was presented to predict SOC in R software. Concentration of SOC has been observed to be more in dense forest areas and vegetated lands. The results showed that the MR model developed from RS data and field-observed data yielded a lower root-mean-square error (RMSE=0.244) and higher R2 (0.71) than the individual model developed through vegetation indices. The outcome of this study indicated that vegetation indices measured with satellite data are good predictors of SOC for river basin areas, especially for tropical and subtropical areas. Therefore, this study provides a promising way for increasing the efficacy and accuracy of DSOCM.