Evaluation of Satellite Land Surface Temperatures Using Ground Measurements from Surface Radiation Budget Network

ABSTRACT Evaluation of satellite land surface temperature (LST) is one of the most difficult tasks in LST retrieval algorithm development, because of spatial and temporal variability of land surface temperature and surface emissivity variations. A large number of high quality “match-up” satellite and ground LST data is needed for the evaluation process. In developing a LST algorithm for the GOES-R Advanced Baseline Imager, we produced a set of “match-up” dataset from SURFace RADiation (SURFRAD) budget network ground m easurements and GOES-8 and -10 satellite measurements. The dataset covers one-ye ar GOES Imager data over six SURFRAD sites in the United States. A stringent cloud filteri ng procedure was applied to minimi ze cloud contamination in the match-up dataset. Each of the SURFRAD sites contains enough match-up data pairs for ensuring significance of statistical analyses of the LST algorithm. The evaluation was performed by directly and indirectly comparing the SURFRAD and satellite LSTs of each site. The direct comparison was illustrated using scatter plots and histogram plots of the ground and the satellite LSTs, while the indirect comparison was performed using a matrix analysis model developed by Flynn (2006)[1]. We demonstrated that LST measurements from the SURFRAD instrument can be used in our evaluation of the GOES-R LST algorithm development and the precision of the GOES-R LST algorithm can be fairly well estimated. Keywords: Land surface temperature, remote sensing, GOES-R