and the prediction of future climate. Relatively realistic multidecadal simulations of the North Atlantic, the Southern Ocean, and the world ocean have recently been carried out. The results of these experiments are being analyzed by numerous groups to aid in understanding the ocean general circulation (e.g., Boning et al., 1991; Semtner and Chervin, 1992). Data collected by comprehensive field programs such as WOCE and TOGA can be interpreted better through the use of realistic models, and field data provide essential tests for the models. WOCE is sponsoring a community modeling effort whereby different models of global circulation are compared. Overall scientific progress is maximized by the interaction of models and observations.
Future progress in modeling will involve new techniques and significantly faster computers to conduct simulations with more realistic hydrodynamics, improved resolution of eddies, longer time integration, and more testing of methods of handling subgrid-scale variables.
Technological advances will probably enhance ocean modeling more than changes in methodology. Computers are expected to attain speeds in excess of one trillion floating-point operations per second (a teraflop) before the year 2000. This thousandfold improvement over computers of 1990 will allow major improvements in simulation capability, such that realistic global models might be achieved. Their maximal use will require the development of highly parallel algorithms. Because most ocean models are formulated in terms of local space-time processes, they should be easily implemented on massively parallel computers.
The computer and communications requirements for archiving, analyzing, and visualizing the output of eddy-resolving basin-to global-scale models are vast. Ongoing federal programs in high-performance computing should help to develop some of the necessary resources. Ocean modeling was highlighted as one application of high-performance computing in the interagency Federal Coordinating Council for Science, Engineering, and Technology supplement to the president's budget for fiscal year 199.3 (FCCSET, 1992). Also, large observational programs are critical because basin-to global-scale, long-term ocean data sets are required to initiate and validate ocean models.
Information on oceanographic research funding in the United States for the 11 fiscal years from 1982 to 1992 is compiled here.