Modeling mining-induced land degradation in Itagunmodi: A multi-temporal machine learning approach with random forest and gradient boosting

Mining activities significantly impact LULC (LULC) dynamics, often leading to environmental degradation and socio-economic consequences. This study employs a multi-temporal machine learning approach using Smile Random Forest (SRF) and Smile Gradient Tree Boost (SGTB) models to analyze and predict mining-induced LULC changes in Itagunmodi, Osun State, Nigeria, from 2002 to 2024, with projections for 2034. Multi-temporal satellite imagery from Landsat 7, 8, and 9 and auxiliary datasets such as digital elevation models (DEM), road networks, and precipitation data were utilized for classification and change detection. The classification accuracy for SRF was robust, achieving overall accuracies of 82 %, 81 %, and 85 % for 2002, 2016, and 2024, respectively. Results reveal significant LULC transitions, including a marked decline in vegetation cover (55.11 % in 2002 to 37.46 % in 2024), expansion of cultivated land (42.47 % to 55.51 %), and a rapid increase in mining sites (0 % to 4.59 %). Change detection analysis identified key transition pathways, with vegetation largely converted to cultivated land and mining sites, while abandoned mining pits contributed to increased water bodies. Driving factors such as geology, proximity to roads and streams, slope, and precipitation were analyzed, with geological formations and accessibility playing significant roles in LULC changes. Predictive modeling for 2034 indicates the continued expansion of mining activities (up to 5.15 %) and urbanization, with cultivated land remaining dominant. The findings emphasize the urgent need for sustainable land management strategies to mitigate environmental degradation while balancing socio-economic development. This study contributes to achieving Sustainable Development Goals (SDGs) 15 (Life on Land), 11 (Sustainable Cities and Communities), and 13 (Climate Action) by providing data-driven insights for informed policy-making and land-use planning. © 2025 Elsevier B.V., All rights reserved.