Very high-resolution land surface temperature retrieval from ISRO’s EOS-08 EOIR observations: radiative transfer modelling, validation, and urban thermal characterization

Very high-resolution thermal infrared (TIR) observations enable detailed characterization of surface thermal heterogeneity across complex landscapes. In this study, land surface temperature (LST) is retrieved from the Electro Optical Infrared (EOIR) payload onboard ISRO’s EOS-08 satellite at a spatial resolution of approximately 10 m using a generalized single-channel (GSC) algorithm. Algorithm coefficients are derived through extensive radiative transfer (RT) simulations using the MODTRAN model under diverse atmospheric conditions, land surface types, and viewing geometries to ensure global applicability as well as physical consistency. Land surface emissivity (LSE) corresponding to the EOIR longwave infrared band (7.7–11 µm) is estimated using the normalized difference vegetation index threshold method based on Sentinel-2 surface reflectance data. The retrieval framework is evaluated through both theoretical validation and a radiance-based validation approach. In the latter, retrieved LST values are reintroduced into the RT model to simulate at-sensor radiance, which is then converted into brightness temperature (BT) and compared with EOIR-observed BT. The GSC algorithm was applied to 290 sites across the globe covering different climatic zones. Across various sites, the comparison achieved a mean coefficient of determination of approximately 0.82, a mean root mean square error of ~1.37 K, and a negligible mean bias error (~0.039 K), demonstrating strong radiometric consistency and minimal systematic error. The capability of the retrieved LST is further demonstrated over multiple metropolitan cities, including Ahmedabad, Delhi, Dubai, and medium urban Biba city, primarily under night-time conditions. The retrieved LST products resolve fine-scale thermal variability associated with urban cores, transportation infrastructure, industrial areas, vegetated regions, and water bodies. The results confirm that EOS-08 EOIR observations, combined with physically based modelling and radiance-domain validation, provide a robust solution for very high-resolution LST retrieval and detailed urban thermal characterization.