Geomodeling soil quality using remote sensing and land use metrics in Abeokuta, Nigeria

Abstract

Remote sensing techniques offer potent and accurate geospatial modeling of Soil quality (SQ) for sustainable food security and development. The study aims to model SQ and landscape changes in Federal University of Agriculture Abeokuta, Nigeria using Landsat images of 2014 and 2024. We applied support vector machine and land use cover metrics like Number patches (NP), Largest-path (LP), and Effective—MESH combined with spectra Bare-soil (BSI), Normalized Difference Vegetation (NDVI), and Soil Adjusted Vegetation (SAVI) Indexes to assess land use and land cover, focusing on built-up, vegetation and wetlands of the area. A composite of 70 soil samples (0 to 30 cm) were analyzed for their physical, chemical, and biological soil properties. The additive, weighted-additive, and proposed models were used to estimate SQ of the studied area. Results showed that NP, LP, and MESH revealed substantial discontinuity and landscape fragmentation especially in the built-up areas. At the same time, BSI, NDVI, and SAVI signify shifts in wetland areas and significant decrease in vegetation cover of the area respectively. The models ranked SQ of the area as high (199 ha), moderate (1333 ha), and low (126 ha) hectares, respectively. The findings suggest valuable ways for SQ improvement and could contribute to efforts on achieving food security, well-being and meet sustainable development goal 2.