The impact of anthropogenic perturbation on the climate system can be projected by calculating all the key processes operating in the climate system through a mathematical formulation which, due to its complexity, can only be implemented in a computer program, referred to as a climate model. If all our current understanding of the climate system were explicitly included, the model would be too complex to run on any existing computer; hence, for practical purposes, simplifications are made so that the system has reduced complexity and computing requirements. Since different levels of simplifications are possible, a hierarchy of models tends to develop (see Chapter 1 and Harvey et al., 1997).
The need to balance scientific understanding against computational efficiency and model realism often guides the choice of the particular class of models used. In addition, it is usually necessary to balance the relative level of detail in the representation, and the level of parametrization, within each component of the climate system.
The most complex climate models, termed coupled atmosphere-ocean general circulation models (and abbreviated as AOGCM in this report), involve coupling comprehensive three-dimensional atmospheric general circulation models (AGCMs), with ocean general circulation models (OGCMs), with sea-ice models, and with models of land-surface processes, all of which are extensively reviewed in the SAR (Chapters 4 and 5). For AOGCMs, information about the state of the atmosphere and the ocean adjacent to, or at the sea surface, is used to compute exchanges of heat, moisture and momentum between the two components. Computational limitations mean that the majority of sub-grid scale processes are parametrized (see Randall and Wielicki, 1997 and Chapter 7). Occasionally atmospheric models with simple mixed-layer ocean models (much discussed and utilised in the SAR) are still used.
Simplifications can be made so that the climate model has reduced complexity (e.g., a reduction in dimensionality to two or even zero). Simple models allow one to explore the potential sensitivity of the climate to a particular process over a wide range of parameters. For example, Wigley (1998) used a modified version of the Wigley and Raper (1987, 1992) upwelling diffusion-energy climate model (see Kattenberg et al., 1996; Raper et al., 1996) to evaluate Kyoto Protocol implications for increases in global mean temperatures and sea level. While such a simple climate model relies on climate sensitivity and ocean heat uptake parameters based on coupled atmosphere-ocean models and ice-melt parameters based upon more complex ice sheet and glacier models, it nevertheless allows for a first-order analysis of various post-Kyoto emission reductions. Simple climate models are also used within larger integrated assessment models to analyse the costs of emission reduction (Peck and Teisberg, 1996; Manne and Richels, 1999) and impacts of climate change (Nordhaus, 1994; Tol, 1999).
Other reports in this collection