Development of a Hybrid SW-TES Method for Simultaneous Land Surface Temperature and Emissivity Retrieval From Landsat 8/9 Thermal Infrared Data

Land surface temperature (LST) is a widely recognized geophysical parameter for studying the interactions within the land-atmosphere system. Thermal infrared (TIR) remote sensing provides an unprecedented capability for retrieving LST accurately across scales from local landscapes to global coverage. After decades of development, numerous algorithms for recovering LST from spaceborne TIR measurements have been proposed. But either land surface emissivity (LSE) or atmospheric parameters were required in existing algorithms. Although the latest SW-TES algorithm was stated to be able to estimate LST and LSE simultaneously without relying on any prior information, the implementation of this method suggests data acquisition from a minimum of three TIR channels. However, the operational Landsat 8/9 satellites are limited to two 11.0 μm and 12.0 μm TIR channels, failing to separate LST and LSE simultaneously. To mitigate this issue, a two-channel-based SW-TES method was proposed here for direct LST and LSE retrieval from the Landsat 8/9 TIR measurements. Consequently, its performance was evaluated against in-situ measured LSTs from the American SURFRAD (SURFace RADiation budget) sites. Results revealed that the proposed method achieved an LST retrieval accuracy of about 1.88 K and 2.17 K regarding Landsat 8 and Landsat 9 TIR data, even surpassing their official LST products with the accuracy of about 2.54 K and 2.80 K. The LSE retrieval accuracy was evaluated via cross-validation with a discrepancy of <0.02. Therefore, conclusions can be drawn that the proposed approach is adequate to retrieve reliable LST from satellite measurements with just two TIR spectral bands. The findings are expected to be of great significance for generating the future LST products.