Spatiotemporal Analysis of Urban Expansion in Debark Town: Leveraging Machine Learning and Google Earth Engine with Multi-Temporal Satellite Imagery

Urban expansion is a pervasive global phenomenon with profound environmental, social, and economic implications, particularly in rapidly developing regions. This paper presents a spatiotemporal analysis of urban expansion in Debark Town, Ethiopia. In this study, machine learning algorithms, such as the random forest algorithm and support vector machine, were investigated for satellite image classification to observe spatiotemporal urban land cover changes. The Google Earth Engine has been utilized to pre-process the Landsat 7, 8, and Sentinel-2 Landsat imagery. Classification accuracy was assessed using an 80:20 training-to-testing split. RF demonstrated higher classification performance with overall accuracies of 95.86% (2000), 95.9% (2015), and 97.29% (2025), outperforming SVM’s 95%, 93.06%, and 93.74% respectively. The research revealed that feature changes occurred during such transition periods. The results revealed agricultural land decreased markedly from 12,384.65 hectares (71.45%) to 7,951.21 hectares (45.88%), and urban land increased built-up areas from 780 hectares (4.50%) in the year 2000 to 2,741.52 hectares (15.82%) in the year 2025, an increase of approximately 1,961.55 hectares, or 11.32%. Furthermore, urban land rose rapidly due to declining vegetative cover, and the built-up areas increased at a rate of 78.46 ha per year. Overall, this research offers useful information regarding the urban land cover change that can assist decision makers, natural resource managers, and policymakers in making the right decisions. This concluded that urban expansion in the research area occurred, and subsequently, urbanization occurred at a high rate within the past years, leading to a decline in agricultural lands.