Comparison Between GOES-East and -West for Land Surface Temperature Retrieval From a Dual-Window Algorithm

In this letter, land surface temperature (LST) is derived from Geostationary Operational Environmental Satellite (GOES)-East (GOES-E) and GOES-West (GOES-W) using a revised dual-window LST algorithm developed by Sun and Pinker in 2004. LST derived from GOES is also evaluated against ground observations. The results show that the LSTs from GOES-E are warmer than those from GOES-W in the morning but lower in the afternoon, while there is no big difference around noon and during night. It is found that the original brightness temperatures from GOES-E and GOES-W show similar patterns to LSTs over time. The discrepancy in LSTs is most probably due to the fact that the Earth surface is warmer in GOES-E than in GOES-W in the morning but cooler in the afternoon. Some other factors may include the difference in satellite viewing geometry, image navigation and registration, calibration, and spectral response functions. It is expected that these effects should be small as those demonstrated in nighttime difference. When evaluated against the ground observations, over the overlap region, the LST bias error is positive from GOES-E but negative from GOES-W in the morning, leading to a positive LST difference, while the bias from GOES-W is close to zero but negative from GOES-E in the afternoon, resulting in a negative LST difference. Nevertheless, the LST root-mean-square errors from GOES-E and GOES-W are very close and reach the maximum around noontime. To synergistically use LST from different satellite sensors, we suggest using nighttime data; further study is needed to best use daytime LSTs.