Sea surface temperature (SST) analyses are used for climate studies for seasonal through centennial periods. The accuracy of the analyses changes with time because of changes in the observations. Beginning with the first in situ observations in the 1850s, the number and the accuracy of the observations have generally improved with time. In late 1981, accurate SST retrievals become available from the Advanced Very High Resolution Radiometer (AVHRR) instrument which has been carried on many NOAA polar orbiting satellites. These retrievals improved the data coverage over in situ observations alone. The satellite retrievals allowed better resolution of small-scale features such as Gulf Stream Eddies. In addition, especially in the Southern Hemisphere, SSTs could now be observed on a regular basis in many locations. However, even with careful quality control, biases occurred in the satellite SST retrievals.
These biases were due to changes in the atmosphere (e.g., injection of
aerosols from volcanic eruptions) as well as changes in the AVHRR
instruments themselves.
For this study, five SST analyses are compared with each other and with in situ and satellite data for the period 1982 through present. The analyses include products which use both in situ and satellite data and in situ data alone. The data include both operational and delayed-mode observations. The results show that analyses using satellite retrievals without careful bias correction using in situ data should not be used for climate studies. However, the results also demonstrate the strong advantage of using corrected satellite retrievals in analyses. The differences among the analyses are discussed, and a best product is recommended which depends on the size of the region analyzed and its location.