Abstract
A band model of atmospheric transmission is used to investigate some of the multichannel sea surface temperature (MCSST) algorithms that are currently in use. The model is used with a set of atmospheric data to derive algorithms for the AVHRR and VAS radiometers that are on the operational NOAA and GOES satellites respectively. For the AVHRR instrument good agreement is found between the model algorithms and those derived in a recent study by McClain. The model results also show that in most cases it is necessary to take account of the satellite view angle. Poorer agreement is found between the theoretical (model) algorithms for the VAS split-window channels and those obtained by regression analyses of satellite data against a multitude of buoy measurements. Intensive ground-truth data are used to evaluate the algorithms. A comparison between measured deficits (sea surface temperatures minus satellite brightness temperatures) and model deficits suggests that the water vapor absorption coefficients used in the model may be too small. This suggestion is supported by ground-based radiometer measurements of clear sky emission.