Variational Data Assimilation for the Global Ocean - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 330

effects of aerosols; the presence of which tends to introduce a cold bias in infrared
estimates of SST. To do this prior information on the microphysical properties of
dust and its amount and vertical distribution is obtained from the Navy Aerosol
Analysis Prediction System (NAAPS; http://www.nrlmry.navy.mil/aerosol/ ) . The
contribution of NAAPS aerosol information to the TOA-BTs is determined using
CRTM, which contains aerosol Jacobians defined for 91 wavelengths and 6 aerosol
species. Equation ( 13.11 ) is then expanded to a matrix to further partition

differences between observed and simulated TOA BTs into an additional aerosol
source of variability. Third, the method can be applied to radiances from ice covered
seas to determine ice surface temperature (IST). Knowledge of IST is important
since it controls snow metamorphosis and melt, the rate of sea ice growth, and
modification of air-sea heat exchange. IST has been added as an analysis variable in
the 3DVAR and is analyzed simultaneously with SST to form a seamless depiction
of surface temperature from the open ocean to ice covered seas. This capability will
be used in the coupled HYCOM/CICE system ( Posey et al. 2010 ).

4 4

13.7.3 SSH Velocity Assimilation

An alternative to assimilating SSH information referenced to the along-track mean
is to assimilate the dynamically important along-track SSH slope. Altimeter SSH
slopes provide the cross-track component of the vertically averaged geostrophic
current. As noted in Sect. 13.4.3 , current methods for assimilating altimeter SSH
data via synthetic temperature and salinity profiles have known deficiencies. One
major difficulty is the need to specify a reference MDT matching that contained
in the altimeter data; a non-trivial problem. The mean height of the ocean includes
the Geoid (a fixed gravity equipotential surface) as well as the MDT, which is not
known accurately enough relative to the centimeter scales of variability contained
in the dynamic topography. The use of SSH slopes obviates the need for a MDT.
To derive geostrophic currents from SSH slopes appropriate for the ocean
mesoscale, noise in the along-track altimeter data must be suppressed. For this
purpose a quadratic LOESS smoother (LOcally wEighted Scatterplot Smoother:
Cleveland and Devlin 1988 ; Schlax and Chelton 1992 ) with varying cutoff wave
lengths is applied. The wave lengths are adjusted in accordance with the Rossby
radius of deformation to account for the varying eddy length scales. The advantage
of this method is that noise in the data, the SSH slope derivative, and the u v

vector velocity components are all computed in a single operation. Figure 13.14
shows the LOESS smoothing of the altimeter SSH data along two tracks; track 109
across the Gulf Stream (Fig 13.14 a) and track 106 across the Kuroshio (Fig. 13.14 b).
The quality of the LOESS filter is clearly seen when the altimeter data exhibit
considerable noise (distance points 1,000-3,000, track 109; distance points 1,200-
2,440, track 106), and when the altimeter data show strong signals from crossing the
Gulf Stream and Kuroshio fronts (distance points 3000-3800, track 109; distance
points 400-1,000, track 106). Figure 13.15 shows the Atlantic and Pacific basin

;