Imager (TMI) in December 1997 made SST from PMW retrievals possible
(Gentemann et al. 2010). Global coverage of PMW-derived SST was achieved in
2002 with the launch of the Advanced Microwave Scanning Radiometer (AMSR)
instrument. The advantage gained with PMW is that radiation at these longer
wavelengths is largely unaffected by clouds and it is generally easier to correct for
atmospheric effects. The PMW signal is, however, affected by sea-surface
roughness and precipitation. Fortunately the surface-roughness and atmosphere
signatures are quite distinct from the SST signature, and their influence can be
readily removed (Gentemann et al. 2010). SST retrieval by PMW is prevented only
in regions with sunglint, rain and near land, providing almost complete global
coverage every 2 days, although at a coarser spatial resolution than TIR-derived
SST measurements (25 km compared to 1 km).
12.1.2 Measurement Accuracies
The radiance measured by TIR and PMW radiometers originates in the surface
‘skin’ (*0.1 mm) and ‘sub-skin’ (*1 mm) layers of the ocean respectively, and
not in the body of the water below, as represented by the ‘bulk’ temperature, which
is measured by in situ thermometers. Near-surface temperature gradients can result
from various processes, such as the absorption of solar radiation, heat exchange
with the atmosphere, and subsurface turbulent mixing (Minnett 2010). The surface
skin-layer of the ocean is generally cooler by a few tenths of a degree than the
underlying bulk temperature just below the surface because the heat flux is nearly
always from the ocean to the atmosphere (Hanafin 2002; Hanafin and Minnett
2001). This relationship is fairly well defined and remains the same on average for
temperatures during the night at depths of a few meters, as well as during the
daytime for wind speed conditions [6 m/s, but is variable under low winds
(Donlon et al. 2002; Minnett 2010). The magnitude of the vertical temperature
gradient is influenced by the diurnal cycle, cloud cover, and wind speed, which
influences the turbulent mixing (Price et al. 1986; Fairall et al. 1996; Gentemann
and Minnett 2008). Under low wind speed conditions, the heat generated in the
upper ocean by insolation is not well mixed through the surface layer and results in
thermal stratification, with temperature differences between the uppermost layer
and the water below. Hence, the difference between the skin or sub-skin temperature and that measured by an in situ thermometer under low winds is variable
and highly dependent on the depth of the bulk measurement (Kearns et al. 2000;
Minnett 2003; Ward 2006).
SST measurements derived from TIR sensors are corrupted by the presence of
clouds, with the temperature of cloud-contaminated pixels generally colder than
the actual SST (Donlon 2010). For applications that require absolute accuracy of
the retrieved SST values, any pixel that is slightly cloud contaminated should be
excluded. For applications that make use of the relative SST values of adjacent
pixels, such as the location of an oceanographic feature, cloud contamination is
12 Thermal Applications
315
(Gentemann et al. 2010). Global coverage of PMW-derived SST was achieved in
2002 with the launch of the Advanced Microwave Scanning Radiometer (AMSR)
instrument. The advantage gained with PMW is that radiation at these longer
wavelengths is largely unaffected by clouds and it is generally easier to correct for
atmospheric effects. The PMW signal is, however, affected by sea-surface
roughness and precipitation. Fortunately the surface-roughness and atmosphere
signatures are quite distinct from the SST signature, and their influence can be
readily removed (Gentemann et al. 2010). SST retrieval by PMW is prevented only
in regions with sunglint, rain and near land, providing almost complete global
coverage every 2 days, although at a coarser spatial resolution than TIR-derived
SST measurements (25 km compared to 1 km).
12.1.2 Measurement Accuracies
The radiance measured by TIR and PMW radiometers originates in the surface
‘skin’ (*0.1 mm) and ‘sub-skin’ (*1 mm) layers of the ocean respectively, and
not in the body of the water below, as represented by the ‘bulk’ temperature, which
is measured by in situ thermometers. Near-surface temperature gradients can result
from various processes, such as the absorption of solar radiation, heat exchange
with the atmosphere, and subsurface turbulent mixing (Minnett 2010). The surface
skin-layer of the ocean is generally cooler by a few tenths of a degree than the
underlying bulk temperature just below the surface because the heat flux is nearly
always from the ocean to the atmosphere (Hanafin 2002; Hanafin and Minnett
2001). This relationship is fairly well defined and remains the same on average for
temperatures during the night at depths of a few meters, as well as during the
daytime for wind speed conditions [6 m/s, but is variable under low winds
(Donlon et al. 2002; Minnett 2010). The magnitude of the vertical temperature
gradient is influenced by the diurnal cycle, cloud cover, and wind speed, which
influences the turbulent mixing (Price et al. 1986; Fairall et al. 1996; Gentemann
and Minnett 2008). Under low wind speed conditions, the heat generated in the
upper ocean by insolation is not well mixed through the surface layer and results in
thermal stratification, with temperature differences between the uppermost layer
and the water below. Hence, the difference between the skin or sub-skin temperature and that measured by an in situ thermometer under low winds is variable
and highly dependent on the depth of the bulk measurement (Kearns et al. 2000;
Minnett 2003; Ward 2006).
SST measurements derived from TIR sensors are corrupted by the presence of
clouds, with the temperature of cloud-contaminated pixels generally colder than
the actual SST (Donlon 2010). For applications that require absolute accuracy of
the retrieved SST values, any pixel that is slightly cloud contaminated should be
excluded. For applications that make use of the relative SST values of adjacent
pixels, such as the location of an oceanographic feature, cloud contamination is
12 Thermal Applications
315