Development of a methodology to utilize land surface temperature satellite data for climate studies

Land Surface Temperature (LST) measured by satellites has the potential to provide useful information about the surface energy and soil hydrology, and can be used to monitor the climate and climate change globally at high spatial resolution. However, unlike Sea Surface Temperature, LST products are not widely used in climate change studies through direct analysis or data assimilation in climate models. This point is easily explained by multiple factors. The LST is retrieved from thermal infrared or passive microwave remote sensing measurements that need atmospheric and/or surface emissivity corrections usually associated with large uncertainties. Moreover, due to the spatial heterogeneity of the land surface, satellite pixels can contain information related to several land types, making LST data retrieval and interpretation difficult.

The objective of the present effort is to develop a methodology that monitors the quality of LST products derived from satellites and contributes to the definition of the next generation of LST retrieval algorithms. Ultimately, this may result in the tailoring of LST products with the high level of required data characteristics essential to support climate studies. The validation approach is able to explore scaling issues over terrestrial surfaces spanning a large range of climate regimes and land cover types, including forests and mixed vegetated areas, and establish the accuracy of LST products for the science-user community. Until now, LST validation was only performed over homogeneous surfaces such as lakes and deserts. The strength of the new approach presented here consists in the merging of information collected at different spatial resolutions to fully explain the satellite products, i.e. measurements from ground sites, airborne platforms, and satellites. For example, the spatial representativeness of LST data is about 10x10 m2 if measurements are made at ground level, 50x50 m2 if made by an aircraft, and 100x100 m2 to 1x1 km2 with satellites at high or moderate resolutions, respectively.

Since the LST is determined by the fluxes of energy and water between the soil, the vegetation and the atmosphere, the present methodology is based on a land surface model that explicitly represents and aggregates the physical processes from ground truth to satellite products. Here, data obtained from airborne flight campaigns provide information to quantify the spatial variability of LST within a satellite pixel and consequently the representativeness of LST data recorded at ground stations. Initial data, to be used in the development of the methodology, are obtained over two micrometeorological tower sites in Tennessee, USA, part of the AmeriFlux network and the US Climate Reference Network (CRN), from MODIS satellites products and a highly instrumented Piper Navajo airborne science research aircraft. Partially funded by NOAA, this research supports the calibration and validation efforts of the NPOESS Preparatory Project (NPP) and Joint Polar Satellite Systems (JPSS) programs.