Comparative Analysis of Land Surface Temperature (LST) Retrieved from Landsat Level 1 and Level 2 Data: A Case Study in Pathumthani Province, Thailand

Abstract

Land Surface Temperature (LST) is a critical parameter for environmental monitoring, urban heat island (UHI) studies, climate research, and sustainable land management. This study presents a comprehensive comparative analysis of LST retrieved from Landsat Level 1 (L1) and Level 2 (L2) data in Pathumthani Province, Thailand, a rapidly urbanizing region. The research aims to evaluate the consistency, reliability, and practical implications of LST values derived from both data processing levels, using multi-temporal imagery from Landsat 5, 7, 8, and 9 acquired between 2004 and 2025. LST was derived from L1 data through a radiative transfer approach combined with NDVI-based emissivity estimation, while L2 data provided pre-processed surface temperature products generated using standardized atmospheric correction algorithms. The results reveal significant discrepancies between the two datasets, with L2-derived LST consistently exhibiting higher values than L1-derived LST by as much as 36.24°C in urban areas. Notably, Landsat 7 L2 data produced extreme LST values (e.g., 82.34°C), which are unrealistic for urban surface temperatures and raise concerns about potential overcorrection in the L2 processing chain. In contrast, L1 data yielded more plausible LST estimates for urban environments but systematically underestimated temperatures over water bodies due to limitations in emissivity assumptions for aquatic surfaces. The study highlights the critical influence of atmospheric correction methods, emissivity modeling, and sensor-specific algorithms on LST accuracy. While L2 products offer convenience as science-ready data, their tendency toward temperature overestimation in urban areas suggests caution in their use for UHI studies and thermal analysis. Conversely, L1 data, when processed with appropriate emissivity corrections, provide more reliable results for built-up landscapes but require additional validation for water bodies. These findings have important implications for researchers, urban planners, and policymakers, emphasizing the need for context-specific data selection in thermal remote sensing applications. The study concludes with recommendations for optimizing LST retrieval methods in heterogeneous environments and underscores the value of integrating multi-source validation data to enhance the accuracy of satellite-derived temperature products in support of climate resilience and sustainable urban development initiatives.