Evaluating the performance of spectral built-up indices in explaining spatial variability of land surface temperature

Understanding the relationship between urban built-up surfaces and land surface temperature (LST) is essential for assessing urban heat dynamics and supporting climate-resilient urban planning. Although numerous spectral built-up indices have been developed to characterize impervious surfaces, their ability to explain LST variability differs considerably, particularly in tropical urban environments where spectral confusion among built-up areas, bare soils, and dry vegetation remains a challenge. This study evaluates six established spectral built-up indices: Normalized Difference Built-up Index (NDBI), Urban Index (UI), Built-up Index (BUI), Enhanced Built-up and Bareness Index (EBBI), Built-up Area Extraction Index (BAEI), and Band Ratio for Built-up Area (BRBA), alongside the proposed Urban Material Index (UMI), designed to enhance detection of thermally significant impervious surfaces. A cloud-free Landsat 9 OLI/TIRS Level-2 image acquired on 8 December 2024 was used to derive LST using the radiative transfer method. Subsequently, 2,086 distributed sample points were extracted to evaluate relationships between LST and the built-up indices using correlation and regression analyses. Results show that UI, BUI, NDBI, and UMI exhibit strong positive relationships with LST, with coefficients of determination (R 2 ) ranging from 0.791 to 0.812 and relatively low RMSE values, indicating their effectiveness in representing thermally active impervious surfaces. Conversely, BAEI and BRBA show substantially weaker relationships, suggesting lower sensitivity in heterogeneous urban landscapes. The proposed UMI performed comparably to the strongest conventional indices, highlighting its potential for urban thermal studies in tropical environments. These findings support informed selection of built-up indices for urban heat assessment, land-use management, and climate adaptation planning.