<![CDATA[This study aims to enhance the spatial resolution of land surface temperature (LST) mapping in Malang City, Indonesia, and to analyze spatiotemporal urban thermal dynamics associated with rapid urbanization. Landsat 8 OLI/TIRS thermal data were integrated with Sentinel-2 multispectral imagery using a Random Forest (RF)–based downscaling approach to refine LST resolution from 30 m to 10 m. Spectral indices, including NDVI, NDBI, NDWI, and NDMI, were employed as predictor variables, and model performance was evaluated using R² and RMSE metrics, supported by in-situ temperature measurements for validation. The results demonstrate strong downscaling performance, with R² values of 0.8374 (2019), 0.8468 (2022), and 0.7675 (2024), while field validation yielded a correlation coefficient of 0.722 and an RMSE of 4.63°C. Spatial and temporal analyses reveal a significant increase in mean LST from 24.67°C in 2019 to 27.21°C in 2024, indicating accelerated urban warming, particularly during 2022–2024. This warming is closely associated with land-use transformation, increased impervious surfaces, and regional climatic influences. In conclusion, the RF-based downscaling approach effectively captures fine-scale urban thermal heterogeneity and provides reliable high-resolution LST information, supporting urban heat mitigation planning and climate adaptation strategies in rapidly growing tropical cities.]]>
Copyrights © 2025
