102
W.T. Liu et al.
Fig. 6.6 Difference between sea surface temperature and air temperature (2 m) for (a) January
and (b) July
variations have larger scale than oceanic temperature variations, the buoyancygenerated turbulence should have strong spatial correlation with T s at small
scales.
The formulation of ψ u was largely based on experiment data on land, validated
only with small amount of measurements over ocean. Although there have been
many investigations to improve flux parameterization, (e.g. Fairall et al., 1996) in
the past few decades, there is no significant change in the formulation of ψ u . As
shown by Liu et al. (2007) and Liu and Xie (2008), and discernable in Fig. 6.4,
the dynamic range of U N over T s anomalies computed from uniform wind fields is
smaller than the QuikSCAT measurement. One of the plausible reasons is that the
ψ u in use now underestimates the stability effect on vertical wind speed changes,
and needs to be investigated. Another reason is a positive atmospheric feedback
through convection (see Liu et al., 2007; Liu and Xie, 2008).
Although both NCEP and ECMWF have assimilated QuikSCAT data since
2002, the data were assimilated as actual wind vector and not stress. The wavenumber spectra constructed from ECMWF 10 m winds collocated with one swath
of QuikSCAT data on August 25, 2005, over the Atlantic (Fig. 6.7) shows that
QuikSCAT U N has higher power than the ECMWF winds at high wave numbers.
W.T. Liu et al.
Fig. 6.6 Difference between sea surface temperature and air temperature (2 m) for (a) January
and (b) July
variations have larger scale than oceanic temperature variations, the buoyancygenerated turbulence should have strong spatial correlation with T s at small
scales.
The formulation of ψ u was largely based on experiment data on land, validated
only with small amount of measurements over ocean. Although there have been
many investigations to improve flux parameterization, (e.g. Fairall et al., 1996) in
the past few decades, there is no significant change in the formulation of ψ u . As
shown by Liu et al. (2007) and Liu and Xie (2008), and discernable in Fig. 6.4,
the dynamic range of U N over T s anomalies computed from uniform wind fields is
smaller than the QuikSCAT measurement. One of the plausible reasons is that the
ψ u in use now underestimates the stability effect on vertical wind speed changes,
and needs to be investigated. Another reason is a positive atmospheric feedback
through convection (see Liu et al., 2007; Liu and Xie, 2008).
Although both NCEP and ECMWF have assimilated QuikSCAT data since
2002, the data were assimilated as actual wind vector and not stress. The wavenumber spectra constructed from ECMWF 10 m winds collocated with one swath
of QuikSCAT data on August 25, 2005, over the Atlantic (Fig. 6.7) shows that
QuikSCAT U N has higher power than the ECMWF winds at high wave numbers.