Monitoring Ecosystem Dynamics Using Machine Learning: Random Forest-Based Land Use Land Cover Analysis in Dinder National Park, Sudan

Dinder National Park (DNP) is one of Sudan’s most significant protected areas, playing a critical role in biodiversity conservation and ecosystem services. However, like many protected areas, DNP faces growing challenges from climate change, human activities, and land use pressures, necessitating detailed and continuous monitoring to ensure its sustainability. This study explores land use and land cover (LULC) changes in Dinder National Park over the period from 2014 to 2024. Utilizing Sentinel-1 and Sentinel-2 imagery processed in Google Earth Engine (GEE), the study assesses vegetation health using indices such as NDVI, EVI, and RVI. A Random Forest classifier was employed to delineate key LULC classes, including trees, cropland, water bodies, grasslands, flooded vegetation, shrubland, built areas, and bare land. The analysis revealed a significant increase in tree cover by 18.3%, while cropland and shrubland decreased by 8.85% and 4.2%, respectively. These shifts, influenced by both natural and anthropogenic factors, reflect critical changes in the park's ecosystem. The findings offer valuable insights for sustainable land management and emphasize the necessity of continuous monitoring in this ecologically sensitive region.