The accuracy of satellite sea surface temperature observations depends on the ability of the satellite sensors to view the sea with little error introduced by the atmosphere. By operating in the atmospheric windows described above, the effects of the gas molecules in the atmosphere have been minimized. However, clouds, water vapor, and aerosols are additional sources of error in the atmosphere that must be considered.

Surface infrared energy cannot penetrate dense, widespread clouds. Therefore, in the presence of a thick cloud cover, the satellite sensor will measure the infrared radiation emitted from the tops of the clouds, and not the sea surface beneath the clouds. Thick, dense, widespread clouds are typically several degrees colder than the sea surface. This makes it easy to detect them in satellite infrared imagery of sea surface temperature, and to correct the imagery.

Figure 2.04 is a typical AVHRR SST image for a single day (July 1, 1994). Note that most of the oceans are cloud covered on an individual day (approximately 80%), and that the sea surface is not visible to the satellite in these cloud covered regions.

Thin clouds such as cirrus and very low stratus clouds are of major concern. They will lower the apparent sea surface temperature measured by the satellite, but not so much that the temperatures are obviously wrong. Because cirrus clouds are very cold, only a few small clouds can contribute large errors to sea surface temperature measurements. Because they are thin, cirrus clouds are nearly invisible and difficult to detect. Correcting sea surface temperature measurements from satellites for the presence of cirrus and low stratus clouds remains a challenging problem for oceanographers.

Water vapor and aerosols decrease the signal from the surface by absorption. The problem with both water vapor and aerosols is that their concentrations in the atmosphere are highly variable in both space and time.

Table 2 shows approximately how much lower the sea surface temperature measured at the satellite will be from the actual sea surface temperature (in °K) because of these various atmospheric sources of error.

Table 2. Atmospheric Sources of Error
from Stewart, Methods of Satellite Oceanography
3.7 µm wavelength
10.5 µm wavelength
undetected clouds
0-10 K
undetected clouds
0-10 K
aerosols 0.3-5 K water vapor 1-8 K
water vapor 0.3-1 K aerosols 0.1-2 K
other gases 0.1 K reflected sky 0.2-0.7 K
other gases 0.1 K