- The extensive requirement of land surface temperature (LST) for envir-onmental studies and management activities of the Earth's resources has made the remote sensing of LST an important academic topic during the last two decades. Many studies have been devoted to establishing the methodology for the retrieval of LST from channels 4 and 5 of Advanced Very High Resolution Radiometer (AVHRR) data. Various split-window algorithms have been reviewed and compared in the literature to understand their di erences. Di erent algorithms di er in both their forms and the calculation of their coe cients. The most popular form of split-window algorithm is T s = T 4 + A(T 4 Õ T5)+ B , where T s is land surface temperature, T4 and T 5 are brightness temperatures of AVHRR channels 4 and 5, A and B are coe cients in relation to atmospheric e ects, viewing angle and ground emissivity. For the actual determination of the coe -cients, no matter the complexity of their calculation formulae in various algo-rithms, only two ways are practically applicable, due to the unavailability of many required data on atmospheric conditions and ground emissivities in situ satellite pass. Ground data measurements can be used to calibrate the brightness temper-ature obtained by remote sensing into the actual LST through regression analysis on a sample representing the studied region. The other way is standard atmo-spheric pro® le simulation using computer software such as LOWTRAN 7. Ground emissivity has a considerable e ect on the accuracy of retrieving LST from remote sensing data. Generally, it is rational to assume an emissivity of 0.96 for most ground surfaces. However, the di erence of ground emissivity between channels 4 and 5 also has a signi® cant impact on the accuracy of LST retrieval. By combining the data of AVHRR channels 3, 4 and 5, the di erence can be directly calculated from remote sensing data. Therefore, much more study is required on how to accurately determine the coe cients of split-window algorithms in the application of remote sensing to examine LST change and distribution in the real world.