387
one, SAM-3, will consider advanced methods such as 4D variational
method.
The SAM suite uses the PALM coupler [Lagarde et al., 2001] which
performs explicit communications between different heterogeneous units
such as between model and assimilation operator units, and simplifies or
makes easier the transition from one SAM version to the other.
In September 2004, SAM1 is used on operational real time basis since
early days of year 2001, SAM2 is under development and will be integrated
in the operational chain, and SAM3 is studied on a research mode.
2.3.1
SAM-1
The SAM-1 is based on a Reduced Order Optimal Interpolation method
(De Mey and Benkiran, 2002). Two different assimilation versions of this
system are used in the MERCATOR prototypes: the SAM1v1 and SAM1v2
techniques.
-
Version 1 (SAM1v1) is based on the Cooper and Haines (1996)
lifting-lowering of isopycnal approach and allows assimilating SLA
data only. The SAM1v1 algorithm starts by calculating a global sea
level anomaly increment from innovations (i.e. model-data
differences) of along tracks sea level anomaly. It is the reduced
order optimal interpolation tool which calculates this increment įȘ.
Based on the statistics for the 3 last months of the ocean model
simulation, įȘ is partitioned into a baroclinic and a barotropic
contribution. The baroclinic part is used to build the temperature
increment and the salinity increment by vertically shifting the
isopycnals for each water column of the model. Geostrophic
velocities are diagnostically adjusted to the new mass field. The
barotropic contribution to sea level is used to build a barotropic
current stream function increment from which one can deduce a
barotropic velocity increment. More details can be found in Ferry et
al. (2005).
-
Version 2 (SAM1v2) is MERCATOR’s first multivariate
assimilation method that allows assimilating simultaneously
temperature and salinity profiles, SST and SLA data. This system
uses fully multivariate 1D Empirical Orthogonal Functions (EOFs
of T(z), S(z) and the barotropic stream function) to perform a ROOI
(Benkiran et al., 2005). This assimilation system works as follows:
first, the differences between the observed and forecast SLA, SST
data and in situ measurements are computed over a one week model
integration. These differences are then analyzed using a fully
multivariate Optimal Interpolation (OI). It is worth noting that the
SST product assimilated is the daily Reynolds SST at the analysis
time (note that it is already used to correct the flux through a
MERCATOR OCEAN
one, SAM-3, will consider advanced methods such as 4D variational
method.
The SAM suite uses the PALM coupler [Lagarde et al., 2001] which
performs explicit communications between different heterogeneous units
such as between model and assimilation operator units, and simplifies or
makes easier the transition from one SAM version to the other.
In September 2004, SAM1 is used on operational real time basis since
early days of year 2001, SAM2 is under development and will be integrated
in the operational chain, and SAM3 is studied on a research mode.
2.3.1
SAM-1
The SAM-1 is based on a Reduced Order Optimal Interpolation method
(De Mey and Benkiran, 2002). Two different assimilation versions of this
system are used in the MERCATOR prototypes: the SAM1v1 and SAM1v2
techniques.
-
Version 1 (SAM1v1) is based on the Cooper and Haines (1996)
lifting-lowering of isopycnal approach and allows assimilating SLA
data only. The SAM1v1 algorithm starts by calculating a global sea
level anomaly increment from innovations (i.e. model-data
differences) of along tracks sea level anomaly. It is the reduced
order optimal interpolation tool which calculates this increment įȘ.
Based on the statistics for the 3 last months of the ocean model
simulation, įȘ is partitioned into a baroclinic and a barotropic
contribution. The baroclinic part is used to build the temperature
increment and the salinity increment by vertically shifting the
isopycnals for each water column of the model. Geostrophic
velocities are diagnostically adjusted to the new mass field. The
barotropic contribution to sea level is used to build a barotropic
current stream function increment from which one can deduce a
barotropic velocity increment. More details can be found in Ferry et
al. (2005).
-
Version 2 (SAM1v2) is MERCATOR’s first multivariate
assimilation method that allows assimilating simultaneously
temperature and salinity profiles, SST and SLA data. This system
uses fully multivariate 1D Empirical Orthogonal Functions (EOFs
of T(z), S(z) and the barotropic stream function) to perform a ROOI
(Benkiran et al., 2005). This assimilation system works as follows:
first, the differences between the observed and forecast SLA, SST
data and in situ measurements are computed over a one week model
integration. These differences are then analyzed using a fully
multivariate Optimal Interpolation (OI). It is worth noting that the
SST product assimilated is the daily Reynolds SST at the analysis
time (note that it is already used to correct the flux through a
MERCATOR OCEAN