The wind-driven dynamics of the interior of the ocean basins are largely a linear response to the wind, and are generally well represented in current models, although there is still some observational debate over the reality of the classical Sverdrup balance (Wunsch and Roemmich, 1985). The main errors can usually be traced back to errors in the driving winds from the atmospheric model. The same can be said of surface Ekman transport, which makes an important contribution to poleward heat transport in the tropics (Danabasoglu, 1998). However, the western boundary currents and inertial recirculations which close the wind-driven gyres are generally poorly resolved by current models, and this may lead to an underestimate of the heat transport by this component of the system at higher latitudes (Fanning and Weaver, 1997a; Bryan and Smith 1998).
Section 8.4.2 and Chapter 7, Section 7.6 discuss the fundamental importance of poleward heat transport in modelling the climate system. Ocean heat transport is greatly improved in some more recent models, compared with the models in use at the time of the SAR (see, e.g., Table 8.2; Chapter 7, Section 7.6). Increased horizontal resolution (Fanning and Weaver, 1997a; Gordon et al., 2000) and improved parametrization of sub-grid scale mixing (Danabasoglu and McWilliams, 1996; Visbeck et al., 1997; Gordon et al., 2000) have been important factors in this. The fresh water transports of coupled models have not been widely evaluated (see Section 8.4.2 and Chapter 7, Section 7.6). Bryan (1998) shows how fresh water imbalances can lead to long-term drifts in deep ocean properties.
The thermohaline circulation (THC) plays an important role in poleward heat transport, especially in the North Atlantic. Table 8.2 shows the strength of North Atlantic THC for various models. In contrast to the SAR, some non-flux adjusted models are now able to produce a THC with a realistic strength of around 20 Sv, which is stable for many centuries. A common systematic error at the time of the SAR was a model thermocline that was too deep and diffusive, resulting in deficient heat transport because the temperature contrast between cold, southward and warm, northward flows was too weak. The models with realistic North Atlantic heat transports generally maintain a realistic temperature contrast (Table 8.2). Some models also show improved realism in the spatial structure of the THC, with separate deep water sources in the Nordic Seas and in the Labrador Sea (Wood et al., 1999).
Interior diapycnal mixing plays a critical role in the thermohaline circulation. Recent process studies (part of WOCE) have confirmed that such mixing is highly localised in the deep ocean (Polzin et al., 1997; Munk and Wunsch, 1998). This mixing is very crudely represented in climate models, and it is not known whether this deficiency has a significant effect on the model thermohaline circulations (Marotzke, 1997).
Although overall heat transports are now better represented in some models, the partition of the heat transport between different components of the circulation may not agree so well with observational estimates (Gordon et al., 2000). How important such discrepancies may be in modelling the transient climate change response is not well understood.
The deep western boundary current, which carries much of the deep branch of the North Atlantic THC, contains a number of strong recirculating gyres (e.g., Hogg, 1983). These recirculations may act as a buffer, delaying the response of the THC to climate anomalies. The representation of these recirculations in climate models, and their importance in transient climate response, have not been evaluated.
Many current models produce rather poor estimates of the volume transport of the Antarctic circumpolar current (ACC) (Table 8.2). The reason for this is not fully understood. Thermohaline as well as wind-driven processes are believed to be important (Cox 1989; Bryan 1998). The problem is shared by some eddy-permitting ocean models, so insufficient horizontal resolution does not seem to be the only factor. The path of the ACC is largely controlled by topography, and errors in the path can lead to significant local sea surface temperature errors. The Atlantic and Indian sectors of the southern ocean appear to be particularly susceptible (e.g., Figure 8.1b; Gent et al., 1998; Gordon et al., 2000). However, it is not clear how, or whether, the transport and SST errors impact on the atmospheric climate or on the climate change response of the models.
At high latitudes, deep convection and subsequent spreading of dense water form the deep water masses that fill most of the volume of the ocean. At mid-latitudes, the processes of mode water formation and thermocline ventilation are the means by which surface changes are propagated into the thermocline (Chapter 7, Section 7.3.1). These processes play an important role in determining the effective rate of heat uptake by the ocean in response to climate change (see Chapter 9, Section 22.214.171.124), and in the response of the THC (see Chapter 9, Section 126.96.36.199). Water mass formation processes can be evaluated directly from model fields (Guilyardi, 1997; Doney et al., 1998), or indirectly using model simulations of the ocean uptake of anthropogenic tracers such as CFCs and carbon 14 (Robitaille and Weaver, 1995; Dixon et al., 1996; England and Rahmstorf, 1999; Goosse et al., 1999; England and Maier-Reimer, 2001). A conclusion from many of these studies is that, while the models clearly show some skill in this area, ventilation processes are sensitive to the details of the ocean mixing parametrization used. Wiebe and Weaver (1999) show that the efficiency of ocean heat uptake is also sensitive to these parametrizations.
Other reports in this collection