A system for producing high-resolution real-time analyses and three-to-ten-day forecasts of the ocean's temperature, salinity and current structure is being developed. It assimilates in situ profile data, surface temperature and surface height data and is driven by surface fluxes from the UK Met Office's numerical weather prediction system. Model configurations with 10-km and 30-km horizontal grid spacing are nested within larger-area, lower-resolution configurations. Three-year trial integrations of the system have been performed.
We will study the statistics of the differences between the observations and the model and make appropriate improvements to the assimilation methods.
Numerical models of the dynamical evolution of the temperature, salinity and current structures of
the ocean covering entire ocean basins can now be run with 40 or more vertical levels and gridpoints
separated by 10 kilometres or less in the horizontal [Smith et al., 2000]. Such models are able to
represent the evolution of the meanders of the Gulf Stream and Kuroshio fronts and the other
energetic "mesoscale" motions in the ocean whose natural horizontal length-scales are of the
order of 50 to 150 kilometres.
Major improvements have been made in the 1990s and will be made in the next five years to
measurements, available within a day or two of real time, of the ocean and its interaction
with the atmosphere. Only within the last five years have adequate orbital corrections to
the mean surface height data from altimeters been available within two days of measurement.
The in situ network of measurements of temperature and salinity as a function of depth will
be significantly improved in the next five years by the ARGO project [Roemmich et al., 1999],
which will deploy and then maintain 3,000 autonomous profiling floats across the global ocean.
Each float will transmit in real time, once every 10 to 15 days, a profile of temperature and
salinity to 2,000-meter depth. Wide swath surface wind stresses from advanced scatterometer
instruments [Milliff et al., 1999] and high-resolution surface temperature data from both
geostationary and polar-orbiting satellites will also be available. Satellite information
on other surface flux and sea-ice fields is also becoming more accurate and extensive
Our main objectives are to combine these models and measurements to produce high-resolution
real-time analyses and three-to-ten-day forecasts of ocean temperature, salinity and current
fields, and to build a community of users who find our products useful. These are the main
objectives of the Global Ocean Data Assimilation Experiment [GODAE, 2000] in which we will
participate. Examples of products which could be of value include: surface currents for
search and rescue, near-bottom currents for oil drilling, the locations and strength of
upwelling in fronts for fisheries, and sound speed structure for naval operations. Our
models also provide boundary data for models forecasting the waters of the continental
shelves. Coupled models of the atmosphere and ocean for seasonal forecasting also benefit
from using ocean measurements to set their initial conditions, but usually have coarser
The system we are developing for making analyses and forecasts is built around an ocean and
sea-ice model which is also developed for climate simulation [Gordon et al., 2000]. We
usually drive it using six-hourly average surface fluxes of momentum, heat and moisture
from the Met Office's numerical weather prediction (NWP) system. Thermal profile and surface
temperature measurements are assimilated into the model as described in Bell et al. .
Sea-ice concentration data from the Canadian Met Center are also assimilated into the
sea-ice model. A global configuration of this system with a one-degree grid spacing has
produced five-day forecasts daily since 1997.
Our recent work has focused on three configurations of the model: the global model with
a one-degree grid spacing; a model covering the Atlantic and Arctic with a 30-kilometer
grid; and one covering the Gulf of Mexico and Caribbean with a 10-kilometer grid. The
areas covered by the higher-resolution models will be increased in future. The
higher-resolution models are nested in the lower-resolution ones using the Flow
Relaxation Scheme [Davies, 1976]. Altimeter data are assimilated (Hines, 2001) using
along-track data and a method based on Cooper and Haines . A set of three-year
integrations (1997-1999) of these models has been performed and is being assessed. Some
integrations assimilated all available data, and others only subsets of the measurements.
Figure 1 illustrates the improvement to the surface temperature field of the model with
10-kilometer resolution in the Gulf of Mexico resulting from assimilation of altimeter data.
We will assess our analyses and forecasts using measurements before they influence the
system. We will also use independent sources of information, such as surface drifter
data (which are not assimilated), to assess the results. To improve the information
extracted from the measurements we will calculate statistics on the variances and
correlations of differences between the measurements and the analyses, compare them
with the statistics used in the data assimilation schemes and amend the statistics and
methods used to assimilate the measurements into the model fields appropriately. We
will also seek to gain access to better data sources. In particular, the surface temperature
data we presently assimilate has very coarse (2.5°) horizontal resolution. Global satellite
data at 50-kilometer resolution and data for the Atlantic at five-kilometer resolution will
Near the Equator, most ocean model systems presently have significant temperature biases.
Assimilation of thermal profile data significantly reduces the biases but drives unrealistic
vertical overturning circulations. Bell et al.  presents a technique for analyzing the
model system's bias and reducing the unrealistic circulations. The use of this technique
with altimeter data will be explored both along the equator and in the western boundary
currents of the coarser-resolution models (where there are also major biases).
Demonstrations of the practical value of altimeter data are vital to facilitate the
transfer of the funding of satellite altimeters from research organisations like NASA
and CNES to operational agencies such as EUMETSAT and NOAA. Our work is a contribution
to this effort and to GODAE.
Bell M.J., R.M. Forbes, A. Hines, 2000: Assessment of the FOAM global data assimilation system for real-time operational ocean forecasting. J. Mar. Sys, 25, 1-22.
Bell M.J., M.J. Martin, N.K. Nichols, 2001: Assimilation of data into an ocean model with systematic errors near the Equator. Ocean Applications Tech. Note 27, Met Office, Bracknell.
Cooper M., K. Haines, 1996: Altimetric assimilation with water property conservation. J. Geophys. Res., 101, C1, 1059-1077.
Davies H.C., 1976: A lateral boundary formulation for multi-level prediction models. Quart. J. Roy. Meteor. Soc., 102, 405-418.EUMETSAT, 2000: SAF Training Workshop, Ocean and Sea Ice. Perros-Guirec, December 1999, EUMETSAT, Darmstadt, Germany.GODAE 2000: The Global Ocean Data Assimilation Experiment Strategic Plan. GODAE Report No. 6. GODAE International Project Office, Melbourne, Australia.
Gordon C., C. Cooper, C.A. Senior, H. Banks, J.M. Gregory, T.C. Johns, J.F.B. Mitchell, R.A. Wood, 2000: The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Climate Dyn., 16, 147-168.
Hines A., 2001: Implementation and tuning of an altimeter data assimilation scheme for high-resolution FOAM models. Ocean Applications Tech. Note No. 26, Met Office, Bracknell.
Milliff R.F., M.H. Freilich, W.T. Liu, R. Atlas, W.G. Large, 1999: Global ocean surface vector wind observations from space. OceanObs99: International Conference on the Ocean Observing System for Climate. Volume I, Session 1C. October 1999. St. Raphael, CNES.
Roemmich D., O. Boebel, Y. Desaubies, H. Freeland, B. King, P.Y. Le Traon, R. Molinari, W.B. Owens, S. Riser, U. Send, K. Takeuchi, S. Wiffels, 1999: ARGO: The global array of profiling floats. OceanObs99: International Conference on the Ocean Observing System for Climate. Volume I, Session 2A. October 1999. St. Raphael, CNES.
Smith R.D., M.E. Maltrud, F.O. Bryan, M.W. Hecht, 2000: Numerical simulation of the North Atlantic Ocean at 1/10°. J. Phys. Oceanogr., 30, 1532-1561.