Key points are not available for this paper at this time.
Abstract The accuracy with which sea surface temperature (s.s.t.) can be measured using satellite infrared radiometers is limited primarily by uncertainties in the correction for atmospheric effects upon the measured radiance. This paper reports an investigation of the accuracy with which this correction can be made over the north‐eastern and tropical Atlantic Ocean using data at 3.7, 11 and 12 μm wavelengths from the Advanced Very High Resolution Radiometer (AVHRR/2) on the NOAA‐7 satellite, and the results are used as a test of atmospheric transmittance models. Simulations of atmospheric transmittances based on line‐by‐line calculations, using published line listings and experimental data on the water vapour ‘continuum’ absorption, provide regression relationships which permit the s.s.t. to be calculated from the brightness temperatures measured in each channel. New algorithms for both the ‘split window’ (11 and 12 μm) and the ‘triple window’ (3.7, 11 and 12 μm) have been derived for a range of airmasses from 1 to 2 to enable the s.s.t. to be retrieved from a 2280 km wide swath centred on the sub‐satellite track. To test the validity of the simulations, the s.s.t. values derived from the satellite measurements were compared with near‐coincident in situ measurements from oceanographic research ships. The absence of significant bias (˜0.1 K) in the comparison constitutes an important experimental verification of the atmospheric transmittance model used in the simulations, and the r.m.s. difference between ship and satellite values indicates that daytime measurement of s.s.t. using the ‘split‐window’ can be made to an accuracy of about ±0.6 K, in the eastern North Atlantic. Some possible sources of this uncertainty are discussed.
Llewellyn‐Jones et al. (Sun,) studied this question.