Integrating Remote Sensing and Machine Learning to Analyze Urban Growth and Its Environmental Effects: A 30-Year Assessment in Başakşehir, Turkey

Abstract This study investigates the impacts of urbanization in Başakşehir, Istanbul, through the analysis of critical environmental indicators: Land Surface Temperature, Normalized Difference Vegetation Index, Proportion of Vegetation, Normalized Urban Heat Island and Urban Thermal Field Variance Index. Using a hybrid supervised machine learning approach integrating Convolutional Neural Networks and Random Forest for Land Use/Land Cover classification, the research achieved an accuracy rate of 93.33%. The findings highlight the complex relationships among urban expansion, ecological health, and environmental changes, advocating sustainable urban planning strategies to address the challenges posed by rapid urbanization. Nonparametric tests, particularly the Mann–Kendall trend test and Sen’s slope estimator, assessed temporal trends in meteorological data, and statistically significant results were obtained for maximum and minimum temperatures (p < 0.001). These results highlight urbanization as a major driver of local climate change, including the Urban Heat Island (UHI) effect. The analysis also reveals vegetation degradation and recovery trends, highlighting the need for urban planning that includes green areas to reduce the UHI effect and enhance ecological resilience. This research provides valuable insights for policymakers by advocating effective conservation strategies that balance urban development with environmental sustainability.