Proximal sensor integration for land use classification and soil analysis in a coastal environment

Abstract

Portable X-ray fluorescence (PXRF) and Nix Pro sensors are efficient tools for rapid in-situ soil analysis. This study combined PXRF and Nix Pro to classify land use and characterize soils from Sandwip Island, Bangladesh. Soil samples from agricultural, abandoned, and seashore areas were analyzed for EC, pH, organic carbon, and texture. Random forest model achieved 84 % classification accuracy, outperforming support vector machines (72 %). Significant soil salinity and management variations were noted, particularly in seashore areas. The findings highlight the potential of these sensors for sustainable soil monitoring, with future work needed to expand applicability to diverse regions and soil types.