GIS-Based Multi-Temporal Land Use and Land Cover Classification for Urban Expansion Planning in the Federal Capital Territory (FCT), Nigeria

Abstract

Rapid urbanisation in the Federal Capital Territory (FCT) of Nigeria has led to increased demand on land resources; hence it is imperative to have reliable and up-to-date land use and land cover (LULC) information for planning and environmental management. This project established a cloud-based GIS procedure for mapping LULC in the FCT using the processed Sentinel-2 Level-2A imagery in Google Earth Engine. The Random Forest classifier was trained using stratified reference samples from the ESA WorldCover dataset. Six spectral bands of Sentinel-2 (B2, B3, B4, B8, B11 and B12), four spectral indices (NDVI, NDBI, MNDWI and BSI) and elevation from Shuttle Radar Topography Mission (SRTM) were used as the predictor variables. Classification performance was tested using an independent validation dataset and standard accuracy metrics, including the confusion matrix, Overall Accuracy, Kappa coefficient, Producer's Accuracy and User's Accuracy. The classification produced an Overall Accuracy of 75.43% and a Kappa coefficient of 0.6928 showing a strong agreement between the classified and the reference data. Vegetation was the dominant land-cover class (47.44%), followed by agriculture (41.33%), built-up land (7.80%), bare terrain (3.11%) and water bodies (0.34%). The spatial distribution of built-up areas indicated high urban concentration in the Abuja Municipal Area Council, while vegetation and agricultural land dominated the outer area councils. The suggested methodology demonstrates the usefulness of merging Sentinel-2 imagery, machine learning and cloud computing for rapid regional LULC mapping. The resulting dataset provides an important baseline for monitoring future urban expansion, analysing environmental change, and enabling evidence-based spatial planning and sustainable land management inside the FCT.