72
J. S. Godfrey
simulating observed ocean circulation, simply by driving ocean models with the new
wind stresses and comparing with available observations (e.g., Kessler et al., 2003;
Sengupta et al., 2004; Xie et al., 2001).
For the heat flux per se, the most useful satellite observations have so far come
from SSMI (from July 1987 onward; Wentz, 1997). QuikSCAT has good resolution
but is available only in recent years (from June 1999 onward). The wind speed of
SSMI is very similar to that of QuikSCAT. Yu et al. (2004a) describe SSMI, and
discuss these issues. The older, generally excellent estimates of SST from infrared
radiation (e.g., Reynolds et al., 2002) were also supplemented by microwave measurements of SST that can penetrate through cloud—crucial especially in the tropics,
where cloudiness coincides with strong winds that often dramatically reduce SSTs.
The ISCPP satellite product estimates the shortwave and infrared fluxes at the surface,
by a combination of radiation measurements from a variety of satellites—some geostationary, some roving—and careful radiative calculations through the atmospheric
column. Estimation of near-surface air humidity from satellites is more problematic,
but satellite products are available which can be used with care. The 10-m air humidity
derived from SSMI can be obtained from Chou et al. (2003).
Yu and her collaborators at Woods Hole have sought resolutions of the problems
with the satellite products; they have then put these resources together, to create a
new global ocean heat flux product. The satellite data proved not to be enough, due to
regional biases associated with different radiometer drifts on different spacecraft, and
especially due to inadequacies in air humidity. These authors adopted an objective
analysis that combines satellite observations with model outputs of surface meteorology (Yu et al., 2004a,b). When the annual and zonal mean of the resulting net heat flux
is compared with the SOC product in the Atlantic, as a function of latitude, good agreement is obtained—especially in the Northern Hemisphere, where the sampling density
of ship observations used to form the SOC climatology is highest (see Figure 5.1).
Recently, comparison has also been made with all available IMET moorings, TAO
and PIRATA buoys in the equatorial Pacific and Atlantic, respectively, and with
ship reports from several research cruises. The results are encouraging (Yu, personal
communication).
It of course remains to be seen how accurate this new product is, in the rest of
the world oceans; and experience certainly suggests caution in assuming some new
product is of universally good quality. Three tools are available to test this accuracy:
one is the use of IMET moorings in several locations, chosen to represent climate
conditions prevailing over large areas of ocean. A second is the use of Volunteer
Observing Ships, equipped with IMET flux-measuring gear; these complement the
good time resolution of the first tool, with good space resolution along ship tracks.
1
1 As an example of the need for high spatial resolution of surface fluxes, note that all merchant ship
climatologies of wind stress completely missed an important, narrow meridional maximum of wind stress
curl just north of the equator in the east Pacific, which shows up in satellite wind stress data. Inclusion
of this maximum greatly improves the simulation of equatorial Pacific currents in ocean models (Kessler
J. S. Godfrey
simulating observed ocean circulation, simply by driving ocean models with the new
wind stresses and comparing with available observations (e.g., Kessler et al., 2003;
Sengupta et al., 2004; Xie et al., 2001).
For the heat flux per se, the most useful satellite observations have so far come
from SSMI (from July 1987 onward; Wentz, 1997). QuikSCAT has good resolution
but is available only in recent years (from June 1999 onward). The wind speed of
SSMI is very similar to that of QuikSCAT. Yu et al. (2004a) describe SSMI, and
discuss these issues. The older, generally excellent estimates of SST from infrared
radiation (e.g., Reynolds et al., 2002) were also supplemented by microwave measurements of SST that can penetrate through cloud—crucial especially in the tropics,
where cloudiness coincides with strong winds that often dramatically reduce SSTs.
The ISCPP satellite product estimates the shortwave and infrared fluxes at the surface,
by a combination of radiation measurements from a variety of satellites—some geostationary, some roving—and careful radiative calculations through the atmospheric
column. Estimation of near-surface air humidity from satellites is more problematic,
but satellite products are available which can be used with care. The 10-m air humidity
derived from SSMI can be obtained from Chou et al. (2003).
Yu and her collaborators at Woods Hole have sought resolutions of the problems
with the satellite products; they have then put these resources together, to create a
new global ocean heat flux product. The satellite data proved not to be enough, due to
regional biases associated with different radiometer drifts on different spacecraft, and
especially due to inadequacies in air humidity. These authors adopted an objective
analysis that combines satellite observations with model outputs of surface meteorology (Yu et al., 2004a,b). When the annual and zonal mean of the resulting net heat flux
is compared with the SOC product in the Atlantic, as a function of latitude, good agreement is obtained—especially in the Northern Hemisphere, where the sampling density
of ship observations used to form the SOC climatology is highest (see Figure 5.1).
Recently, comparison has also been made with all available IMET moorings, TAO
and PIRATA buoys in the equatorial Pacific and Atlantic, respectively, and with
ship reports from several research cruises. The results are encouraging (Yu, personal
communication).
It of course remains to be seen how accurate this new product is, in the rest of
the world oceans; and experience certainly suggests caution in assuming some new
product is of universally good quality. Three tools are available to test this accuracy:
one is the use of IMET moorings in several locations, chosen to represent climate
conditions prevailing over large areas of ocean. A second is the use of Volunteer
Observing Ships, equipped with IMET flux-measuring gear; these complement the
good time resolution of the first tool, with good space resolution along ship tracks.
1
1 As an example of the need for high spatial resolution of surface fluxes, note that all merchant ship
climatologies of wind stress completely missed an important, narrow meridional maximum of wind stress
curl just north of the equator in the east Pacific, which shows up in satellite wind stress data. Inclusion
of this maximum greatly improves the simulation of equatorial Pacific currents in ocean models (Kessler