Description: To avoid an explicit simulation of the overflows across the Greenland-Scotland ridge, many models of the large-scale ocean circulation seek to include the net effect of the inflowing dense water masses by restoring temperature and salinity near the ridge to observed conditions. In this paper the authors examine the effect of different datasets for the northern restoring condition in two versions, eddy resolving and non-eddy resolving, of the model of the North and equatorial Atlantic that has been developed in recent years as a Community Modeling Effort for WOCE. It is shown that the use of smoothed climatological fields of temperature and salinity south of the Denmark Strait leads to strong deficiencies in the simulation of the deep flow field in the basin. A switch to actual hydrographic data from the Denmark Strait ignites a rapid dynamic response throughout the North Atlantic, affecting the transport and vertical structure of the deep western boundary current and, by virtue of the JEBAR efffect, the transport of the horizontal gyres. Meridional overturning and northward heat transport too weak in the cases with climatological boundary conditions, increase to more realistic levels in the subtropical North Atlantic. The initial response to switches in the high-latitude thermohaline forcing is mediated by fast waves along the westurn boundary, leading to changes in the deep western boundary current in low latitudes after about two years in the non-eddy-resolving cast. The initial timescale depends on the horizontal grid spacing of the model; in the high-resolution case, the first signal reaches the equator in a few months. The adjustment to a new, dynamic quasi equilibrium involves Kelvin waves along the equator and Rossby wave in the interior and is attained in less than two decades throughout the North Atlantic. It is suggested that these fast dynamic adjustment processes could play an important role in possible fluctuations of the thermohaline circulation, or transitions between different equilibrium states of the coupled ocean–atmosphere system, and may have determined the timescale of the observed climatic transitions before and during the last deglaciation.

Description: A high-resolution general circulation model of the North Atlantic, first developed at the National Center for Atmospheric Research and then run at the Institut für Meereskunde in Kiel for two different wind climatologies and reduced vertical friction, is evaluated in the upper 500 m for the western tropical Atlantic, 5°S to 15°N. Although the general features of the vigorous seasonal circulation changes documented in previous studies and in the earlier high-resolution model of Philander and Pacanowski (1986a) are reproduced, there are some interesting differences. Lack of eastward penetration of the Equatorial Undercurrent (EUC) and a thermocline that is too diffuse are model deficiencies due to the constant vertical eddy diffusion coefficient. In the lower friction case the undercurrent partially surfaces in the west, causing an eastward surface current on the equator, which is not apparent in the earlier model studies. Further, the zonal currents, in the low-friction version, have high-velocity bands, resulting, e.g., in two separate current cores in the North Equatorial Countercurrent (NECC) region; and an eastward surface core just south of the equator, connected to the EUC. Particularly interesting are equatorward undercurrents along the western boundary, one of which has already been confirmed in recent measurements off French Guyana. In winter it connects with the EUC in the model, in summer with the NECC. A northward undercurrent in the model exists off Brazil, between 5° and 10°S, but that is already close to the southern boundary of the model domain. The annual mean throughflow from the southern hemisphere into the Caribbean along the western boundary is small in the model, and in particular, there is no enhanced throughflow in winter, when the cross-equatorial North Brazil Current transport is not taken up by the NECC.

Description: Global mean and eddy fields from a four-year experiment with a 1/6° × 1/5° horizontal resolution implementation of the CME North Atlantic model are presented. The time-averaged wind-driven and thermohaline circulation in the model is compared to the results of a 1/3° × 2/5° model run in very similar configuration. In general, the higher resolution results are found to confirm that the resolution of previous CME experiments is sufficient to describe many features of the large-scale circulation and water mass distribution quite well. While the increased resolution does not lead to large changes in the mean flow patterns, the variability in the model is enhanced significantly. On the other hand, however, not all aspects of the circulation have improved with resolution. The Azores Current Frontal Zone with its variability in the eastern basin is still represented very poorly. Particular attention is also directed toward the unrealistic stationary anticyclones north of Cape Hatteras and in the Gulf of Mexico.

Description: During the last decade it has become obvious that the ocean circulation shows vigorous variability on a wide range of time and space scales and that the concept of a "sluggish" and slowly varying circulation is rather elusive. Increasing emphasis has to be put, therefore, on observing the rapidly changing ocean state on time scales ranging from weeks to decades and beyond, and on understanding the ocean's response to changing atmospheric forcing conditions. As outlined in various strategy and implementation documents (e.g., the implementation plans of WOCE, AMS, CLIVAR, and GODAE) a combination of the global ocean data sets with a state-of-the-art numerical circulation model is required to interpret the various diverse data sets and to produce the best possible estimates of the time-varying ocean circulation. The mechanism of ocean state estimates is a powerful tool for such a "synthesis" of observations, obtained on very complex space-time pattern, into one dynamically consistent picture of the global time-evolving ocean circulation. This process has much in common with ongoing analysis and reanalysis activities in the atmospheric community. But because the ocean is, and will remain for the foreseeable future, substantially under-sampled, the burden put on the modeling and estimations components is substantially larger than in the atmosphere. Moreover, the smaller dynamical eddy scales which need to be properly parameterized or resolved in ocean model simulations, put stringent requirements on computational resources for ongoing and participated climate research.