Variational Data Assimilation for the Global Ocean - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 297

currents and fronts, and generation and propagation of coastally trapped waves.
Hurlburt et al. ( 2008a ) gives a good discussion of the requirements for an ocean
model to be eddy-resolving. High resolution global ocean forecast models present
challenges for the assimilation component of the forecasting system given the huge
model state vector and the ever increasing number of satellite and in situ ocean
observations available for the assimilation. Accordingly, the global analysis has to
be both computationally efficient and accurate to account for the oceanographic
features resolved by the high resolution model. At the same time the analysis must
use all of the available observations and create and maintain dynamically adjusted
corrections to the model forecast.
The purpose of this chapter is to provide an overview of a new variational ocean
data assimilation system that has been developed as an upgrade to an existing
multivariate optimum interpolation (MVOI) system ( Cummings 2005 ). Compared
to the MVOI the 3DVAR algorithm has several advantages. First, the 3DVAR
performs a global solution that does not require data selection. In the MVOI,
observations are organized into overlapping analysis volumes and the solution can
depend on how the volumes are defined. This is not the case in the 3DVAR, as the
global solve allows all observations to influence all grid points, a requirement for
an optimum analysis. Second, through the use of observation operators, 3DVAR
can incorporate observed variables that are different from the model prognostic
variables. Examples of this in the ocean are integral quantities, such as acoustic
travel time and altimeter measures of sea surface height, and direct assimilation
of satellite radiances of sea surface temperature (SST) through radiative transfer
modeling. Finally, 3DVAR permits more powerful and realistic formulations of
the background error covariances, which control how information is spread from
the observations to the model grid points and model levels. The error covariances
also ensure that observations of one model variable produce dynamically consistent
corrections in the other model variables.
The 3DVAR referred to in this paper is the Navy Coupled Ocean Data Assim-
ilation (NCODA) system, version 3. NCODA 3DVAR is in operational use at the
U.S. Navy oceanographic production centers: Fleet Numerical Meteorology and
Oceanography Center (FNMOC) in Monterey, CA, and the Naval Oceanographic
Office (NAVOCEANO) at the Stennis Space Center, MS. NCODA is truly a
unified and flexible oceanographic analysis system. It is designed to meet all Navy
ocean data analysis and assimilation requirements using the same code. In two-
dimensional mode, NCODA provides SST and sea ice concentration analyses for
lower boundary conditions of the Navy global and regional atmospheric forecast
models. In three-dimensional mode, it is executed in a sequential incremental update
cycle with the Navy ocean forecast models: the Hybrid Coordinate Ocean Model
(HYCOM) on the global scale, and the Navy Coastal Ocean Model (NCOM) on
the regional scale. Here, NCODA provides updated initial conditions of ocean
temperature, salinity, and currents for the next run of the ocean forecast model.
The analysis background fields, or first guess, are generated from a short-term
ocean model forecast, and the 3DVAR computes dynamically consistent corrections
to the first-guess fields using all of the observations that have become available