Ground validation of land surface temperature and surface emissivity from thermal infrared remote sensing data

Abstract

Land Surface Temperature (LST) and Land Surface Emissivity (LSE) are the direct driving force of long wave radiation and latent heat flux exchange at the surface-atmosphere interface, which are also two important parameters on surface energy budget balance and water balance at regional or global scales, its temporal and spatial variations on surface-atmosphere interface have a wide range of applications on weather forecasting, climate change, water cycle, geological exploration, agriculture, forestry monitoring and the urban thermal environment research, and many other scientific fields. Remote sensing provides an effective approach to obtain LST and LSE at global scale rather than point measurements for its rather wider spatial coverage and temporal revisit convenience. The validation of quantitative retrieval on LST and LSE products is conducive to find the defects of remote sensing data processing or drawbacks on retrieval algorithms, as well as clarify the accuracy and uncertainty of the operational products, which is of great convenience for the application and popularity of these products. In this paper, firstly different definitions of LST and LSE are reviewed, and then the scientific connotation of LST and LSE which can be retrieved from thermal infrared remote sensing data and measured from in-situ experiments is explained. The theoretical background, such as radiative transfer theory and methods of LST and LSE retrieval from remote sensing data are then summarized and outlined. After that, the framework of LST and LSE validation are summarized systematically, the validation metrics of LST and LSE derived from remotes sensing data including accuracy, precision, uncertainty, completeness and stability is constructed. Based on the validation framework, the methods of ground validation for LST and LSE are introduced (including directly validation and indirectly comparison), and followed by the methods of ground measurements of LST or related auxiliary data. The method of scale conversion method from point level to pixel level for heterogeneous and non-isothermal surfaces are emphasized and analyzed, and the main error sources of LST validation and LSE validation are discussed, respectively. A summarization of the main sites or networks for the validation of LST and LSE is conducted and the spatial distribution and the main characteristics (such as heterogeneity, land cover) of the typical LST and LSE validation sites or networks are briefly analyzed. The current LST and LSE products derived from satellite remote sensing data, which utilize the abovementioned validation sites or networks are summarized to report their validation accuracy or uncertainty, and related development on validation of LST and LSE are reviewed. Finally, some problems of validation of LST and LSE are also presented, and then the future outlook and trends of validation are outlined and justified.