The key problem to be addressed is the response of the climate system to changes in forcing. In many cases there is a fairly direct and linear response and many of the simulated changes fall into that class (Chapter 9). These concern the large-scale general circulations of the atmosphere and ocean, and they are in principle represented in current comprehensive coupled climate models. Such possible large-scale dynamical feedbacks are also influenced by small-scale processes within the climate system. The various feedbacks in the climate system may amplify (positive feedbacks), or diminish (negative feedbacks) the original response. We often approximate the response of a particular variable to a small forcing by a scaling number. A prominent example of such a quantification is the equilibrium global mean temperature increase per Wm^-2 change in the global mean atmospheric radiative forcing (see Chapter 9, Section 9.2.1).

While many aspects of the response of the climate system to greenhouse gas forcing appear to be linear, regime or mode transitions cannot be quantified by a simple number because responses do not scale with the amplitude of the forcing: small perturbations can induce large changes in certain variables of the climate system. This implies the existence of thresholds in the climate system which can be crossed for a sufficiently large perturbation. While such behaviour has long been known and studied in the context of simple models of the climate system, such thresholds are now also found in the comprehensive coupled climate models currently available. Within this framework, the possibility for irreversible changes in the climate system exists. This insight, backed by the palaeo-climatic record (see Chapter 2, Section 2.4), is a new challenge for global change science because now thresholds have to be identified and their values need to be estimated using the entire hierarchy of climate models.

To estimate the response properly, we must represent faithfully the physical processes in models. Not only must the whole system model perform reasonably well in comparison with observation (both spatial and temporal), but so too must the component models and the processes that are involved in the models. It is possible to tune a model so that some variable appears consistent with that observed, but we must also ask whether it comes out that way for the right reason and with the right variability. By examining how well individual processes are known and can be modelled, we can comment on the capabilities and usefulness of the models, and whether they are likely to be able to properly represent possible non-linear responses of the climate system.

Examples of some processes in the climate system and its components that have been dealt with in the Second Assessment Report (IPCC, 1996) (hereafter SAR) and still are important topics of progress:

• The hydrological cycle. Progress has been made in all aspects of the hydrological cycle in the atmosphere, involving evaporation, atmospheric moisture, clouds, convection, and precipitation. Because many facets of these phenomena are sub-grid scale, they must be parametrized. While models reasonably simulate gross aspects of the observed behaviour on several time-scales, it is easy to point out shortcomings, although it is unclear as to how much these affect the sensitivity of simulated climate to changes in climate forcing.
  
• Water vapour feedback. An increase in the temperature of the atmosphere increases its water-holding capacity; however, since most of the atmosphere is undersaturated, this does not automatically mean that water vapour, itself, must increase. Within the boundary layer (roughly the lowest 1 to 2 km of the atmosphere), relative humidity tends to remain fixed, and water vapour does increase with increasing temperature. In the free troposphere above the boundary layer, where the water vapour greenhouse effect is most important, the behaviour of water vapour cannot be inferred from simple thermodynamic arguments. Free tropospheric water vapour is governed by a variety of dynamical and microphysical influences which are represented with varying degrees of fidelity in general circulation models. Since water vapour is a powerful greenhouse gas, increasing water vapour in the free troposphere would lead to a further enhancement of the greenhouse effect and act as a positive feedback; within current models, this is the most important reason for large responses to increased anthropogenic greenhouse gases.
  
• Cloud radiation feedback. Clouds can both absorb and reflect solar radiation (thereby cooling the surface) and absorb and emit long-wave radiation (thereby warming the surface). The compensation between those effects depends on cloud height, thickness and cloud radiative properties. The radiative properties of clouds depend on the evolution of atmospheric water vapour, water drops, ice particles and atmospheric aerosols. These cloud processes are most important for determining radiative, and hence temperature, changes in models. Although their representation is greatly improved in models, the added complexity may explain why considerable uncertainty remains; this represents a significant source of potential error in climate simulations. The range in estimated climate sensitivity of 1.5 to 4.5°C for a CO₂ doubling is largely dictated by the interaction of model water vapour feedbacks with the variations in cloud behaviour among existing models.
• Sub-grid scale processes in ocean models. Although improved parametrizations in coarse resolution models are available to represent mixing processes, they do not, in their present form, induce sufficient variability on the broad range of time-scales exhibited by eddy-resolving ocean models. Thus present generations of coarse resolution ocean models may not be able to decide questions about those types of natural variability that are associated with sub-grid scale processes, and increased resolution is highly desirable.
  

IPCC Homepage