A Machine Learning Assessment of Land Use and Land Cover Change on Urban Thermal Dynamics in Makassar

Rapid urbanization in Makassar City has intensified Land-Use and Land-Cover (LULC) transformation and altered urban thermal conditions. This study aims to assess the changes in LULC and their relationship with urban thermal dynamics in Makassar between 2019 and 2024. Sentinel-2 Level-2A imagery was used for LULC classification using the Support Vector Machine (SVM) algorithm. At the same time, Landsat-8 TIRS data were processed to derive the Land Surface Temperature (LST) and the Urban Thermal Field Variance Index (UTFVI). Four LULC classes were mapped: water, vegetation, built-up area, and bare land. The SVM classification achieved high reliability, with overall accuracies of 0.901 and 0.913 and kappa coefficients of 0.86 and 0.88 for 2019 and 2024, respectively. The results showed a 12% increase in built-up area and a 12% decrease in vegetation cover, accompanied by a 2.1°C increase in average LST and a 15% expansion of the UTFVI heat stress zone. Kruskal-Wallis tests confirmed significant differences in UTFVI between LULC classes in 2019 and 2024, with large effect sizes. These findings indicate that LULC transformation is a major driver of urban thermal stress in Makassar, highlighting the need for climate-adaptive spatial planning.