Satellite-Derived Sea Surface Temperatures: Evaluation ofGOES-8andGOES-9Multispectral Imager Retrieval Accuracy

Abstract
Sea surface temperature (SST) retrieval accuracy from the multispectral imager on the new generation of GOES satellites is analyzed. Equations for two and three infrared channels are empirically derived using cloud-free satellite radiances matched to buoy SST measurements obtained in 1995 and 1996. Both GOES-8 and GOES-9 demonstrate the capability to retrieve sea surface temperature at better than 1-K root-mean-square difference (rmsd) with negligible bias relative to buoy SST measurements. GOES-8 rmsd errors are found to be 0.79 K (day) and 0.81 K (night). GOES-9 rmsd errors are 0.65 K (day) and 0.59 K (night). The GOES-9 results are relatively comparable to those currently achieved operationally from the NOAA polar-orbiting satellite Advanced Very High Resolution Radiometer sensor. Investigation revealed that GOES imager multiple detector scan striping impacted SST accuracy, requiring sample array averaging for best results. Abstract Sea surface temperature (SST) retrieval accuracy from the multispectral imager on the new generation of GOES satellites is analyzed. Equations for two and three infrared channels are empirically derived using cloud-free satellite radiances matched to buoy SST measurements obtained in 1995 and 1996. Both GOES-8 and GOES-9 demonstrate the capability to retrieve sea surface temperature at better than 1-K root-mean-square difference (rmsd) with negligible bias relative to buoy SST measurements. GOES-8 rmsd errors are found to be 0.79 K (day) and 0.81 K (night). GOES-9 rmsd errors are 0.65 K (day) and 0.59 K (night). The GOES-9 results are relatively comparable to those currently achieved operationally from the NOAA polar-orbiting satellite Advanced Very High Resolution Radiometer sensor. Investigation revealed that GOES imager multiple detector scan striping impacted SST accuracy, requiring sample array averaging for best results.