It must be remembered that scientific predictions are based on complex models and simulations – and these in turn are based on many assumptions and contain many uncertainties. The following topics are the ones which require the most urgent work to reduce uncertainties:

 

• Estimation of future emissions
   
• Estimation of biogeochemical cycling (including sources and sinks) of greenhouse gases, aerosols, aerosol pre-cursors – and of the likely future concentration and radioactive properties of these aerosols.
   
• Representation of climate processes in the models. More accurate representation of the feedback role of clouds, oceans, sea ice and vegetation in models is required in order to improve projections or rates and regional patterns of climate change.
   
• More accurate and systematic collection of information on climate and its variables over the long-term (e.g. solar output, atmospheric energy balance components, hydrological cycles, ocean characteristics and ecosystem changes). This information is required for the purposes of model testing, assessment of temporal and regional variability, and for detection and attribution studies.
   

It should also be noted that climate systems are nonlinear in nature. When nonlinear systems are rapidly forced they are subject to unexpected behaviour. This makes the prediction of future large and rapid climate system changes difficult and it implies that future climate changes may also involve ‘surprises’. Investigation into the behaviour of nonlinear processes and subcomponents of the climatic system may help scientists to make progress in this area.