<![CDATA[Rapid urbanization in Mamuju City, Indonesia, has altered land cover patterns and contributed to the development of Urban Heat Island (UHI) conditions. This study aims to analyze the spatial and temporal characteristics of UHI in Mamuju City during the dry season (June–September) from 2015 to 2024, and to examine the relationships between land surface temperature, land cover change, and atmospheric temperature. The study integrates ERA5 reanalysis air temperature data with Landsat 8/9 Level-2 imagery using the Google Earth Engine platform. Land cover classification was conducted using the Random Forest algorithm with 200 decision trees, incorporating spectral bands, spectral indices (NDVI and NDBI), and topographic variables. Land Surface Temperature (LST) was derived from Landsat thermal data. UHI intensity was calculated as the difference between mean LST of built-up and bare land areas (urban zone) and the mean LST of vegetated, agricultural, and water areas (rural zone). Pearson correlation analysis was applied on a per-year basis to assess relationships between LST, NDVI, NDBI, and ERA5 air temperature. The results show that UHI intensity remained in the strong category (ΔT = 5.1–7.6°C) throughout the study period. Mean LST increased from 30.6°C in 2015 to 32.1°C in 2024, accompanied by a 69.2% increase in built-up area and a 7.9% decrease in dense vegetation. Land cover classification achieved an overall accuracy of 72.0% with a Kappa coefficient of 0.65. LST showed a positive correlation with NDBI (r = 0.662–0.721) and a negative correlation with NDVI (r = −0.658 to −0.693). ERA5 air temperature showed a weak and statistically insignificant correlation with LST (r = 0.137, p = 0.707). These results indicate that observed surface temperature patterns in Mamuju City are closely associated with land cover composition and change, while coarse-resolution reanalysis data are not suitable for representing intra-urban thermal variability at the city scale.]]>
