Field and laboratory data for assessing soil salinity and crop stress using Sentinel-2

Soil salinity and waterlogging pose significant threats to agricultural productivity in irrigated command areas, particularly in semi-arid regions like the Ghataprabha command area in Karnataka, India. This study leverages Sentinel-2 multispectral imagery from March 2024, processed via Google Earth Engine, to delineate salinity classes and assess their spectral signatures. Ground truth data from 120 soil samples (0-30 cm depth) were used to validate electrical conductivity (ECe) levels ranging from <3 to >12 dS m-1. Key spectral bands (Blue, Green, Red, NIR, SWIR1) and indices (NDVI, NDSI, RVI, SAVI, SASI, NDWI, EVI, SBI, SI, SI1, CSI, SI2) were analysed for correlations with ECe. Results reveal strong linear relationships in individual bands (r = 0.98-0.99, p < 0.01) and a multiple regression model (R2 = 1) incorporating all bands. Vegetation indices like EVI showed moderate negative correlations (r = -0.26, p < 0.01), while salinity-specific indices exhibited weaker associations. The approach achieved high accuracy in mapping salinity extents, highlighting Sentinel-2's efficacy for operational monitoring in tropical semi-arid agroecosystems.