Abstract

Rapid urban expansion poses significant challenges for sustainable land management in many cities across sub-Saharan Africa. This study assesses the spatial and temporal patterns of urban expansion in Benin City, Nigeria, between 2000 and 2024 using multi-temporal Landsat imagery and machine-learning classification. Landsat 5, 7, 8, and 9 surface reflectance data were processed and classified using a Random Forest algorithm into four land use/land cover classes: built-up, vegetation, water, and bare land. Classification accuracy was evaluated using independent validation samples, yielding an overall accuracy of approximately 71%. To reduce spectral ambiguity between built-up and bare surfaces and to address temporal inconsistencies common in pixel-based classification, built-up and bare land were aggregated into a single urban class, and urban expansion was analyzed using a cumulative urban footprint approach. Results show that cumulative urban area increased from approximately 1,584 km² in 2000 to 1,914 km² in 2024, representing a net increase of about 331 km² (20.9%). Urban expansion was spatially concentrated around the city core and along major transportation corridors, with notable outward growth between 2000 and 2020, followed by a slower expansion rate in recent years. The findings demonstrate the effectiveness of combining Landsat time-series data with machine-learning techniques for long-term urban expansion analysis and provide valuable insights for urban planning and sustainable development in rapidly growing Nigerian cities.

Keywords: Urban expansion, LULC classification, Landsat imagery, Random Forest classifier, Machine learning, Benin City, Nigeria