Soil Organic Carbon Mapping Using Sentinel-1 And ALOS-2 SAR Data

Accurate Soil Organic Carbon (SOC) estimation over croplands is vital for sustainable land management and climate change mitigation. Remote sensing technologies offer a promising alternative by providing extensive spatial coverage and enabling frequent SOC monitoring across large areas. This study introduces a novel approach to estimating SOC using multi-frequency synthetic aperture radar (SAR) data and advanced machine learning techniques to monitor SOC spatial patterns over croplands during fallow and the effect of different cropping systems. A Random Forest (RF) model was employed to capture the complex relationships between SAR data and SOC. The study region, historically characterized by diverse agricultural practices, transitioned to a focus on rice, finger millet, Niger, maize, and vegetable cultivation in 2020. A correlation coefficient, r of 0.68, with a root mean square error (RMSE) of 0.19, was obtained, demonstrating its ability to predict SOC content accurately.