HR: 17:30h
AN: A52F-06    [PDF]
TI: 1/32$\deg$ Global Ocean Modeling and Prediction
AU: * Shriver, J F
EM: shriver@nrlssc.navy.mil
AF: Naval Research Laboratory, Code 7300 Bldg 1009, Stennis Space Center, MS 39529 United States
AU: Hurlburt, H E
AF: Naval Research Laboratory, Code 7300 Bldg 1009, Stennis Space Center, MS 39529 United States
AU: Smedstad, O M
AF: Planning Systems Inc., MSAAP Bldg 9121, Stennis Space Center, MS 39529 United States
AU: Wallcraft, A J
AF: Naval Research Laboratory, Code 7300 Bldg 1009, Stennis Space Center, MS 39529 United States
AU: Rhodes, R C
AF: Naval Research Laboratory, Code 7300 Bldg 1009, Stennis Space Center, MS 39529 United States
AB: The resources of a three-year Department of Defense (DoD) High Performance Computing (HPC) Challenge project was crucial in the development a data-assimilative 1/32$\deg$ global ocean nowcast/forecast system, which included the associated basic research and exploratory development. This system is scheduled for transition to the Naval Oceanographic Office (NAVO) in the first quarter of FY04. This 1/32$\deg$ global system will be an upgrade to the existing 1/16$\deg$ NRL Layered Ocean Model (NLOM) nowcast/forecast system which is operational at NAVO. A critical issue in forecast system design is determining the resolution required. Ocean models require finer resolution and more computer time than atmospheric models in part because the space scales for variability due to flow instabilities (oceanic mesoscale eddies vs. atmospheric highs and lows) are about 20-30 times smaller than found in the atmosphere. We need to resolve the oceanic eddy space scale very well because (1) upper ocean - topographic coupling via flow instabilities has a major impact on the pathways of many upper ocean currents (including mean pathways) and very fine resolution of the flow instabilities is required to get sufficient coupling, (2) very fine resolution is required to obtain (a) inertial jets and sharp oceanic fronts which span major ocean basins as observed and (b) the associated nonlinear recirculation gyres which affect the shape of large-scale ocean gyres, (3) it is necessary to resolve small islands and narrow passages which affect current pathways and current transports in many regions, (4) in data assimilative mode we do not want the ocean model to "fight" the data because the natural behavior of the ocean model is inconsistent with the observations, and (5) a very high horizontal resolution model is needed to help get an accurate mean sea surface height field to add to the deviations obtained from satellite altimetry (observations alone do not provide sufficient resolution to do this). The need for 1/32$\deg$ resolution ($\sim$3.5 km at mid-latitudes) has been demonstrated through extensive research, including essential contributions from earlier DoD HPC Challenge projects as well as non-challenge HPC usage. These projects also helped establish 1/16$\deg$ ($\sim$7 km at mid latitudes) as the minimum resolution needed for a fully eddy-resolving global ocean prediction system. Major improvement in model simulation skill was obtained with an increase to 1/32$\deg$ resolution, but only modest additional improvement with a further increase to 1/64$\deg$ resolution. Hence, our target resolution globally is 1/32$\deg$. The ocean model, after 30 years of climatological spinup, was run interannually spanning the period 1979-present. In addition to re-equilibration, the climatological simulation was used to fix problems that arose, optimize model parameters, assess the impact of the resolution increase on model realism and dynamics and to perform climatological model-data comparisons. These results were used to generate statistics and a model sea surface height (SSH) mean for data assimilative experiments. The model assimilates SSH from 3 satellite altimeters (ERS-2, GFO and JASON-1) and sea surface temperature from satellite IR imagery. Assimilation of satellite altimeter data is critical in allowing the model to map individual current meanders and eddies. Forecast experiments were then initialized from the data assimilative experiments. Results from this 1/32$\deg$ system were then compared with observed data to assess model realism and the value added of the resolution increase from 1/16$\deg$ to 1/32$\deg$.
DE: 4255 Numerical modeling
DE: 4263 Ocean prediction
DE: 4520 Eddies and mesoscale processes
SC: Atmospheric Sciences [A]
MN: 2003 Fall Meeting