The methods for the retrieval of land surface temperature in central Shijiazhuang using ASTER data

Land surface temperature (LST) is crucial for studying climate change, agricultural drought, and ecological evaluation. Satellite thermal infrared remote sensing, such as the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), provides high-resolution (90 m) LST data. However, inversion accuracy varies by region and method. This study focuses on Shijiazhuang, using ASTER data and three algorithms-reference channel method, split-window algorithm, and temperature and emissivity separation algorithm-to invert LST. Data were preprocessed using ENVI, ARCGIS, MODTRAN, and MATLAB. Authenticity tests and accuracy evaluations were conducted based on meteorological data, Landsat data, and MODIS products. Results show that the temperature and emissivity separation algorithm has the highest accuracy and is most suitable for the study area. Inversion results range from 274 to 310 K, with mean temperatures of 293.18 K (reference channel) and 293.15 K (temperature and emissivity separation). The split-window algorithm has a lower low-temperature value (274.27 K) and a higher high-temperature value (308.10 K). The accuracy difference between algorithms and average surface temperature is 3.87-4.08 °C. Spatial distribution and linear fitting of pixel values are consistent across algorithms. Correlation analysis with MODIS products shows R² values of 0.72, 0.72, and 0.72, with the temperature and emissivity separation algorithm having the highest correlation. Conclusions indicate that the temperature and emissivity separation algorithm is the optimal method for LST inversion in Shijiazhuang's urban area, providing a reliable approach for high-precision monitoring and data-scarce regions.