GOES Land Surface Temperature - Algorithm
The Algorithm Theoretical Basis Document (ATBD) describes in detail the procedures for developing and using a land surface temperature (LST) algorithm designed for the GOES imager. It includes a description of the requirements and specifications of the LST products and some specific information about the GOES imager that is relevant to the derivation of the LST products. The main part of the ATBD is a description of the science of the proposed GOES imager LST algorithm. The process of algorithm selection is documented. This includes a review of satellite LST research, a selection of candidate algorithms and a description of a large simulated GOES imager data set that was used to derive algorithm coefficients and test the candidate algorithms. The simulated radiances were calculated using sensor spectral response functions (SRF) that are expected from the actual GOES imager instrument. A description of the expected implementation of the LST algorithm is provided. Ancillary data sets needed for the LST calculation are listed.
Nine split window algorithms were built, which were adapted from the literature, and two algorithms proposed for the GOES –M (12)-Q series for evaluation as the GOES-imager LST algorithm. All algorithms used explicit spectral emissivity and satellite view angle. Algorithm regression coefficients were derived from the simulated data set.
The selected algorithm was applied to GOES imager data. The retrieved LSTs were compared against independent ground truth data and the results were analyzed. The algorithm was found to meet specs with the test data sets. Perfectly cloud free data with cloud cover fraction less than 5% is assumed in all testing of the ATBD research. A process for routine evaluation of the operational GOES-imager LST is described. This includes automatic matchups against ground truth and methodology for product evaluation. Comparisons of the dual-window algorithm with the split window algorithm, LSTs derived from the GOES-East and GOES-West, and polar orbiting system (MODIS) are conducted. It is found the accuracy from the dual-window algorithm by combining middle infrared (MIR) with infrared window channel (11.0 µm) is worse than that from split-window algorithm, indicating the lack of split window in the GOES- M (12) –Q series may degrade the performance of GOES imager LST products. Finally practical matters such as computer resources, instrument performance and its effects on the product are considered.
Please see the Algorithm Theoretical Basis Document for details.