Abstract

A concerted effort has been made to simulate the global ocean circulation with resolved eddies, using a highly optimized model on the best available supercomputer. An earlier 20‐year spin‐up has been extended for 12.5 additional years: the first 2.5 with continued annual mean forcing and the final 10.0 with climatological monthly forcing. Model output archived at 3‐day intervals has been analyzed into mean fields, standard deviations, products, and covariances on monthly, annual, and multiyear time scales. The multiyear results are examined here in order to give insight into the general circulation of the world ocean. The three‐dimensional flow fields of the model are quite realistic, even though resolution of eddies in high latitudes is marginal with a 0.5°, 20‐level grid. The use of seasonal forcing improves the simulation, especially in the tropics and high northern latitudes. Mid‐latitude gyre circulations, western boundary currents, zonal equatorial flows, and the Antarctic Circumpolar Current (ACC) all show mean and eddy characteristics similar to those observed. There is also some indication of eddy intensification of the mean flow of the ACC and of separated boundary jets. A global thermohaline circulation of North Atlantic Deep Water is identified in deep western boundary currents connected by the ACC. This deep circulation rises mainly in the equatorial Pacific. Several zonal jets are an integral part of this circulation near the equator. The deep flow rises toward the surface in a series of switchbacks. Much of the thermohaline return flow then follows an eddy‐rich warm‐water route through the Indonesian archipelago and around the southern tip of Africa. However, some intermediate level portions of the thermohaline circulation return south into the ACC and follow a cold water route through the Drake Passage. The representation of a global “conveyor belt” circulation with narrow and relatively high‐speed currents along most of its path may be the most important result of this modeling study. Statistics of scalar fields such as transport stream function and surface height are exhibited, as are time series and frequency spectra of certain variables at selected points. These provide a baseline for comparison both with observations and with other model studies at higher resolution. Mean and eddy characteristics of the near‐surface temperature and salinity fields are discussed, and surface forcing fields are examined. In particular, combined thermal and hydrological forcing effects are found to drive a conveyor belt circulation between the tropical Pacific and the high‐latitude North Atlantic. The effect of weak restoring terms to observed temperature and salinity at great depth and in polar latitudes is found mainly to augment the model's convective processes, which are poorly resolved with 0.5° grid spacing. However, the deep restoring terms are insignificant in both the tropics and the mid‐latitudes. The geographical distributions of eddy heat and salt transport are discussed. The eddies transport heat and salt down the gradients and along the mean flow in many regions of strong currents. Net meridional transports of heat and salt by both the total currents and the eddies are computed for the Atlantic, the Indo‐Pacific, and the global ocean. The total currents provide for poleward heat transport (except near 40°S, where the contribution from ACC instabilities is rather weak) and, in particular, for that needed to sustain the conveyor belt transport. Meridional eddy transports are especially important for warming the Pacific upwelling branch of the thermohaline circulation and for transporting salt across the equator into the North Pacific. Planned improvements to the model include a free‐surface treatment of the barotropic mode and additions of the Arctic Basin and sea ice. A fully prognostic extension of the existing integration is intended, with subsequent transitioning of the model onto a 0.25° grid having very realistic geometry. The 0.25° version of the model will run effectively on newly available supercomputers.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call