Modeling and Mapping of Salt-Affected Soils through Spectral Indices in Inland Plains of Semi-arid Agro-Ecological Region

Abstract

Mapping the salt-affected soils (SAS) through spectral indices derived from satellite imagery is essential for its monitoring at a frequent interval and for the management. Several salinity indices have been developed based on the combination of spectral bands for predicting soil salinity in the arid environment. Present study attempted to identify and map the salt-affected soils (SAS) in inland plains of semi-arid agro-ecological region of Tamil Nadu, India using satellite image (Resourcesat-1(IRS-P6) LISS IV). Vertisols and Inceptisols are the major soil orders cultivated under paddy, sugarcane and pulses. An integrated approach of spectral bands and indices along with analytical data of samples collected from the field was used to develop multiple regression equations for prediction. The best linear regression model was selected based on adjusted R2 and MSE values for the prediction of pH, EC, and ESP. The variation observed between salt-affected soils predicted by multiple regression equations and measured data were 15, 10, and 5 percent for pH, EC, and ESP, respectively, due to weak variation in soil properties, sample density, better vegetation cover by management practices.