Geoscience Reference
In-Depth Information
Table 14.1
A summary of the observation types, observing platforms, data sources, the nominal
combined measurement and representation errors, and the period covered. The combined error is
replaced by the standard deviation of the observations about the super observation value if larger
than the assumed nominal error
Observing
Observation type
platform
Source
Combined error
Period covered
SSH
Altimeter
Aviso, 1 day
average
0.04 m
1993-2010
0:6
ı
C
SST
AVHRR/
Pathfinder
NOAA Coast
Wat ch
1981-2011
SST
AMSR-E
NOAA Coast
Wat ch
0:7
ı
C
2002-2010
SST
GOES
NOAA Coast
Wat ch
1
ı
C
2001-2010
a
0:5
ı
C
SST
MODIS-
Terra
NASA JPL
2000-2011
0:5
ı
C for T
Hydrographic data
Various
UK
Meteorological
Office
1950-2011
0.1 for S
a
The GOES SST are seriously biased during the period 2001-2002, so they are not used in ROMS
4D-Var until 2003
Fig. 14.2
A time series of log
10
the total number of super observations available each month
from EN3 and each satellite observing platform within the ROMS CCS model domain during
each month of the year during the period spanned by WCRA31.
Blue
: In situ observations from
EN3;
Red
: SST from AVHRR/PathFinder;
Black
: SST from AMSR-E;
Green
: SST from GOES;
Magenta
: SSH from Aviso
14.4.4
WCRA 4D-Var Configuration
Both WCRA13 and WCRA31 take the form of sequences of overlapping 8 day
analyses each separated by 4 days as illustrated in Fig.
14.3
. During each analysis
cycle, all of the available observations during the cycle time interval are assimilated
into ROMS using the dual formulation of 4D-Var in the form of PSAS. Each
analysis cycle utilizes 1 outer-loop and 15 inner-loops which has been demonstrated
by
Broquet et al.
(
2009a
),
Moore et al.
(
2011b
)and
Gurol et al.
(
2013
) to yield
adequate convergence toward the cost function minimum in the CCS region. The
prior
estimate for the initial conditions for each assimilation cycle is the
posterior
Search WWH ::
Custom Search