240
P.J. Minnett
provides sufficient information to enable the standard deviation of error in each
observation type to be estimated, thus allowing a more appropriate apportioning of
the uncertainties.
14.5.1 AVHRR
The statistics shown in Table 14.1 are for the AVHRR Pathfinder SST fields
(Kilpatrick et al., 2001) and show a near-zero mean error and standard deviation
of about half a degree. The coefficients in the atmospheric correction algorithms
used in the Pathfinder project are derived by robust regression between the AVHRR
brightness temperatures and in-situ measurements from drifting buoys, so the
retrieved SSTs are a “bulk SST” and they are related to the skin SST, through the
mean skin temperature differences occurring in the matchups between satellite and
in-situ data.
Table 14.1 Uncertainties in the Pathfinder AVHRR SST retrievals, from cloud-free comparisons
between all AVHRR’s in the period 1985–1998 and drifting buoys (from Kilpatrick et al., 2001)
Data
Mean
St dev
All
0.02 K
0.53 K
Using a radiometric SST from the M-AERI (Minnett et al., 2001), reduces the
error (Table 14.2), with the exception of polar regions (Kearns et al., 2000) where
the multichannel atmospheric correction algorithm has been shown to be prone to
larger uncertainties (Vincent et al., 2008).
Table 14.2 Uncertainties in the AVHRR SST retrievals vs. M-AERI (after Kearns et al., 2000)
Mean (K) St dev (K) N
Data
Mean (K) St dev (K) N
Excluding arctic cruise
All
0.14
0.36
299 0.07
0.31
219
Day
0.18
0.40
142 0.00
0.24
62
Night 0.10
0.33
157 0.10
0.33
157
14.5.2 MODIS
The statistics of the uncertainties in the MODIS (Esaias et al., 1998) SST retrievals
from the instruments on Terra and Aqua are shown in Tables 14.3, 14.4, 14.5 and
14.6, derived from comparisons with buoys and radiometers. The periods covered
are from the start of each of the missions to mid 2007, and the processing scheme
is V5 for both instruments. The statistics are shown for daytime and nighttime
retrievals using two channels in the 11–12 μm atmospheric transmission window,
P.J. Minnett
provides sufficient information to enable the standard deviation of error in each
observation type to be estimated, thus allowing a more appropriate apportioning of
the uncertainties.
14.5.1 AVHRR
The statistics shown in Table 14.1 are for the AVHRR Pathfinder SST fields
(Kilpatrick et al., 2001) and show a near-zero mean error and standard deviation
of about half a degree. The coefficients in the atmospheric correction algorithms
used in the Pathfinder project are derived by robust regression between the AVHRR
brightness temperatures and in-situ measurements from drifting buoys, so the
retrieved SSTs are a “bulk SST” and they are related to the skin SST, through the
mean skin temperature differences occurring in the matchups between satellite and
in-situ data.
Table 14.1 Uncertainties in the Pathfinder AVHRR SST retrievals, from cloud-free comparisons
between all AVHRR’s in the period 1985–1998 and drifting buoys (from Kilpatrick et al., 2001)
Data
Mean
St dev
All
0.02 K
0.53 K
Using a radiometric SST from the M-AERI (Minnett et al., 2001), reduces the
error (Table 14.2), with the exception of polar regions (Kearns et al., 2000) where
the multichannel atmospheric correction algorithm has been shown to be prone to
larger uncertainties (Vincent et al., 2008).
Table 14.2 Uncertainties in the AVHRR SST retrievals vs. M-AERI (after Kearns et al., 2000)
Mean (K) St dev (K) N
Data
Mean (K) St dev (K) N
Excluding arctic cruise
All
0.14
0.36
299 0.07
0.31
219
Day
0.18
0.40
142 0.00
0.24
62
Night 0.10
0.33
157 0.10
0.33
157
14.5.2 MODIS
The statistics of the uncertainties in the MODIS (Esaias et al., 1998) SST retrievals
from the instruments on Terra and Aqua are shown in Tables 14.3, 14.4, 14.5 and
14.6, derived from comparisons with buoys and radiometers. The periods covered
are from the start of each of the missions to mid 2007, and the processing scheme
is V5 for both instruments. The statistics are shown for daytime and nighttime
retrievals using two channels in the 11–12 μm atmospheric transmission window,