Only strategies 1 and 2 are applicable to the problem of abrupt climate change, because of the nonlinear nature of the phenomenon. Coupled climate models of reduced complexity have been obtained with either strategy. Following the first, by zonally averaging the equations of motion in the ocean (e.g., Marotzke et al., 1988; Wright and Stocker, 1991), a very efficient ocean model component is obtained, which can be coupled to an energy balance model of the atmosphere (Stocker et al., 1992) or a statistical-dynamical model of the atmosphere (Petoukhov et al., 2000). The choice implies that the focus of investigation is restricted to the latitude-depth structure of the flow and to a priori chosen time scales that are accessible with these models. The second strategy was followed when three-dimensional ocean-circulation models were combined with a latitude-longitude energy balance model (Fanning and Weaver, 1997) or with an atmospheric-circulation model of reduced complexity (Opsteegh et al., 1998). Such combinations can be integrated for many thousands of years, thanks to the relative simplicity of the atmosphere.
Overall, models of reduced complexity are highly useful tools in paleoclimate research, and in particular for investigations of abrupt climate change, provided that they are used wisely. Clearly, they cannot replace general circulation models (GCM), because the reduced-complexity models consider only a limited set of constraints that are important in the climate system. The weaknesses of the reduced-complexity models are the incompleteness of dynamics and their often reduced resolution. Their strength is their computational efficiency, which permits extensive sensitivity studies or ensemble or even Monte Carlo simulations. Single simulations with reduced-complexity models are not useful to advance the science even if they happen to agree well with paleoclimatic data. However, these models are key tools in the process of quantitative hypothesis-building and -testing, not only in paleoclimatology but also in climate dynamics in general, because they allow the investigation of some feedback or process in its purest, isolated form. Often, the understanding gained from the reduced model is used to interpret the results from complex models. In addition, results from simple models or conceptual considerations have helped to define the strategy pursued in the use of complex models. In the following section, the committee adopts this approach to investigate the most fundamental questions concerning the stability of the THC and its role in abrupt climate change.