Land surface temperature (LST) is an essential input for modeling the processes of energy exchange and balance of the earth's surface. Thermal infrared (TIR) remote sensing is considered to be the most efficient way to obtain accurate LST, both regionally and globally. Currently, many LST retrieval algorithms have been developed, including the up-to-date SW-TES (SW: split window; TES: temperature-emissivity separation) method, which is claimed to be able to accurately derive LST without the need for atmospheric information and land surface emissivity (LSE) based on the selected multiple TIR channel configuration. However, this hybrid method is actually not applicable to observations with large viewing angles and was only preliminarily evaluated in Australia. In this study, this method was extended for application to global TIR measurements with different viewing angles. Additionally, the performance of this extended SW-TES method was assessed globally for different seasons by using the MODIS LST product as a reference, and was also validated using in-situ LST measurements from the SURFRAD (SURFace RADiation budget network) sites. The results showed that the LST retrievals using the extended SW-TES method were comparable to the MODIS LST product, with discrepancies of <2.7 K and < 1.8 K for global daytime and nighttime observations, respectively. Validations based on the SURFRAD in-situ LST measurements indicated that the extended method could be used to retrieve LST accurately with a root-mean-square error (RMSE) of approximately 3.6 K during the daytime and 2.4 K during the nighttime. However, special attention should be paid when applying the extended method to daytime observations on grasslands and shrublands during hot seasons, considering the relatively large discrepancy when using this method compared with that obtained with the MODIS LST product (>4.0 K). Overall, in this study, the SW-TES method was extended, and the performance was comprehensively evaluated at the global scale, which may help in facilitating its potential applications.

地表温度 (LST) 是模拟地球表面能量交换和平衡过程的重要输入。热红外 (TIR) 遥感被认为是在区域和全球范围内获得准确 LST 的最有效方法。目前，已经开发了许多LST反演算法，包括最新的SW-TES（SW：分裂窗口；TES：温度-发射率分离）方法，据称该方法无需大气就能准确推导LST。基于选定的多 TIR 通道配置的信息和地表发射率 (LSE)。但是，这种混合方法实际上不适用于大视角的观测，仅在澳大利亚进行了初步评估。在这项研究中，该方法被扩展用于具有不同视角的全局 TIR 测量。此外，通过使用 MODIS LST 产品作为参考，这种扩展 SW-TES 方法的性能在不同季节在全球范围内进行了评估，并且还使用来自 SURFRAD（表面辐射预算网络）站点的原位 LST 测量进行了验证。结果表明，使用扩展 SW-TES 方法的 LST 反演与 MODIS LST 产品相当，全球白天和夜间观测的差异分别为 <2.7 K 和 < 1.8 K。基于 SURFRAD 原位 LST 测量的验证表明，扩展方法可用于准确检索 LST，均方根误差 (RMSE) 在白天约为 3.6 K，在夜间约为 2.4 K。然而，考虑到使用该方法与MODIS LST产品（>4.0 K）相比，使用该方法时的差异较大，因此在将扩展方法应用于炎热季节对草地和灌木地的白天观测时应特别注意。总体而言，本研究扩展了 SW-TES 方法，并在全球范围内对其性能进行了综合评估，这可能有助于促进其潜在应用。