SENSOR BASED CHARACTERIZATION OF SOIL PROPERTIES FOR SUNDARBAN

Sensor-based characterization of soil properties is inevitable due to the tedious, expensive, and time-consuming nature of traditional laboratory-based soil testing. This study focused on characterizing the soil properties of Indian Sundarban at depths of 0cm–20cm, 20cm–50cm, and 50cm–100cm using a PXRF, a Nix Pro color sensor, and a USB soil microscope. The soil samples belonged to mudflats in the intertidal zone and abandoned shrimp ponds. The soil samples were analyzed via Wet Digestion and Dry combustion for organic carbon (OC), total carbon (TC), and bulk density (BD). The average OC, TC, and BD at 0cm–20cm were 7.69 t/ha, 8.74 t/ha, and 1150 kg/m3, respectively. At 20cm–50cm, it was 11.73 t/ha, 13.08 t/ha, and 1239 kg/m3, respectively; 17.733 t/ha, 20.083 t/ha, and 1196 kg/m3, respectively, at 50cm–100cm. Modeling of OC, TC, and BD via multiple linear regressions was validated using R2 and RMSE. PXRF produced satisfactory results, with BD (R2 = 0.928; 0.9431; 0.998), SOC (R2 = 0.9785; 0.9297; 0.9726), and TC (R2 = 0.9938; 0.9619; 0.987). The Nix Pro color sensor with RGB, CIEL*A*b, and CMYK color spaces BD (R2 = 0.3769; 0.4786; 0.3894), SOC (R2 = 0.649; 0.3029; 0.4314), and TC (R2 = 0.4932; 0.2603; 0.5518). USB microscope with RGB color space underperformed, while similar performance sequence was observed using RMSE. Also, soil data was spatially analyzed to assess variability across three depths.