A novel correlation-hypothesis based single channel method for land surface temperature retrieval with reduced atmospheric dependency

Abstract

As one of the critical parameters in the land-atmosphere exchange processes, land surface temperature (LST) plays an essential role in various domains, such as climate change, urban heat island effect, disaster monitoring, and evaporation retrieval. Thermal infrared (TIR) remote sensing is one of the main approaches to obtaining LST on a large scale. For the sensors with only one TIR channel, the single-channel (SC) methods are commonly and effectively used to retrieve LST, as they are most suitable under such limitations. However, atmospheric correction is essential for the SC methods, which involves significant uncertainty and complexity. To reduce the atmospheric dependency of SC methods, this study proposes a Correlation-Hypothesis based Single-Channel (CH-SC) method to retrieve LST. In this method, the LST can be retrieved using the top-of-atmosphere (TOA) brightness temperature from a single TIR channel and the LSEs from two virtual adjacent channels, while atmospheric water vapor content (WVC) is used solely to assess atmospheric conditions. Consequently, the CH-SC method exhibits the least sensitive in atmospheric parameter errors compared to existing SC methods. This inherent robustness results in superior stability of retrieval accuracy, enhancing its practicality for real applications. Subsequently, the CH-SC method was applied to Landsat 7 data, alongside the mono-window (MW) method and generalized single-channel (GSC) method. The retrieved LSTs were compared with in-situ measurements for validation. As a result, the CH-SC method exhibited strong performance compared to in-situ measurements, with an RMSE of 2.83 K in SURFRAD sites and 4.02 K in BSRN sites, which was comparable to Landsat 7 LST product (3.15 K at SURFRAD and 4.26 K at BSRN) and outperforming other SC methods. Generally, compared to the existing methods, the proposed method exhibits minimal dependence on atmospheric information while ensuring superior accuracy and stability, even under high water vapor conditions. That holds significant application value, especially for the sensors with limited TIR channels to enable real-time on-orbit computation of LST.