256
H.M. Beggs
Table 15.3 Definitions of SST product types (classified by depth) used in this chapter, following
the GHRSST convention (Donlon et al., 2007)
SST product type Nominal depth
SST sensors supplying product type
SST skin
~10–20 μm
In-situ radiometers, (A)ATSR, MODIS, SEVIRI,
GOES, MTSAT-1R
SST subskin
~1 mm
AMSR-E, TMI
SST depth
Specified depth (~1 m for
buoys)
Moorings, drifters, ships, Argo, XBT, CTD,
AVHRR (calibrated using moorings/drifters to
buoy depths)
SST fnd
1 mm to ~10 m (depending
on solar, ocean,
meteorological
conditions)
Moorings, drifters, ships, Argo, XBT, CTD
(depending on solar, ocean and meteorological
conditions); all satellite IR and MW sensors
using diurnal variation models to convert from
skin or subskin to foundation SST
SST blend
10 μm to ~ 10 m
All
Combination of data from infrared and microwave sensors from polar-orbiting
satellites and infrared sensors on geostationary satellites, often with in-situ measurements from drifting buoys and ships, now results in high-quality SST level
4 analyses with high spatial resolution, ranging from 0.25 ◦ (e.g. Operational
MGDSST – Guan and Kawamura, 2004) down to 0.01 ◦ (experimental G1SST –
Chao et al., 2009).
Table 15.4 presents a list of several of the operational, global, daily SST analyses
that use IR SST along with the types of SST inputs used. Some analyses, such as
OSTIA (Stark et al., 2007) and ODYSSEA (Autret and Piollé, 2007), use the highly
accurate, dual-view, AATSR IR SST data in conjunction with in-situ SST to correct
for biases in the other satellite SST data streams. Others use a combination of drifting buoy and ship SST observations to correct for regional and seasonal biases (e.g.
MGDSST, CMC, AVHRR_OI, AVHRR_AMSR_OI). However, issues remain with
the calibration of the IR and MW SST data streams, particularly at high latitudes
north of 50 ◦ N and south of 50 ◦ S (Reynolds et al., 2010) where available in-situ SST
observations are still relatively sparse.
Two operational analyses in Table 15.4 (OSTIA and GAMSSA – Zhong
and Beggs, 2008) convert the radiometric SSTskin from IR satellite sensors and
SSTsubskin from microwave sensors (AMSR-E and TMI) to an estimate of the foundation SST (see Section 15.2) by applying a simple empirical algorithm (Donlon
et al., 2002) that filters out day-time input observations for surface winds <6 m/s.
Comparisons between OSTIA and GAMSSA SSTfnd and independent buoy SST
observations from the following day (0.4 and 0.5 ◦ C, respectively), indicate that filtering the input observations in this manner reduces the standard deviation errors
compared with those obtained from a global SSTblend analysis such as NCDC
AVHRR_AMSR_OI (0.6 ◦ C, Beggs et al., 2009b).
H.M. Beggs
Table 15.3 Definitions of SST product types (classified by depth) used in this chapter, following
the GHRSST convention (Donlon et al., 2007)
SST product type Nominal depth
SST sensors supplying product type
SST skin
~10–20 μm
In-situ radiometers, (A)ATSR, MODIS, SEVIRI,
GOES, MTSAT-1R
SST subskin
~1 mm
AMSR-E, TMI
SST depth
Specified depth (~1 m for
buoys)
Moorings, drifters, ships, Argo, XBT, CTD,
AVHRR (calibrated using moorings/drifters to
buoy depths)
SST fnd
1 mm to ~10 m (depending
on solar, ocean,
meteorological
conditions)
Moorings, drifters, ships, Argo, XBT, CTD
(depending on solar, ocean and meteorological
conditions); all satellite IR and MW sensors
using diurnal variation models to convert from
skin or subskin to foundation SST
SST blend
10 μm to ~ 10 m
All
Combination of data from infrared and microwave sensors from polar-orbiting
satellites and infrared sensors on geostationary satellites, often with in-situ measurements from drifting buoys and ships, now results in high-quality SST level
4 analyses with high spatial resolution, ranging from 0.25 ◦ (e.g. Operational
MGDSST – Guan and Kawamura, 2004) down to 0.01 ◦ (experimental G1SST –
Chao et al., 2009).
Table 15.4 presents a list of several of the operational, global, daily SST analyses
that use IR SST along with the types of SST inputs used. Some analyses, such as
OSTIA (Stark et al., 2007) and ODYSSEA (Autret and Piollé, 2007), use the highly
accurate, dual-view, AATSR IR SST data in conjunction with in-situ SST to correct
for biases in the other satellite SST data streams. Others use a combination of drifting buoy and ship SST observations to correct for regional and seasonal biases (e.g.
MGDSST, CMC, AVHRR_OI, AVHRR_AMSR_OI). However, issues remain with
the calibration of the IR and MW SST data streams, particularly at high latitudes
north of 50 ◦ N and south of 50 ◦ S (Reynolds et al., 2010) where available in-situ SST
observations are still relatively sparse.
Two operational analyses in Table 15.4 (OSTIA and GAMSSA – Zhong
and Beggs, 2008) convert the radiometric SSTskin from IR satellite sensors and
SSTsubskin from microwave sensors (AMSR-E and TMI) to an estimate of the foundation SST (see Section 15.2) by applying a simple empirical algorithm (Donlon
et al., 2002) that filters out day-time input observations for surface winds <6 m/s.
Comparisons between OSTIA and GAMSSA SSTfnd and independent buoy SST
observations from the following day (0.4 and 0.5 ◦ C, respectively), indicate that filtering the input observations in this manner reduces the standard deviation errors
compared with those obtained from a global SSTblend analysis such as NCDC
AVHRR_AMSR_OI (0.6 ◦ C, Beggs et al., 2009b).