Identifying key sources of uncertainty in climate change projections

What sources of uncertainty should be included in climate change projections and what gains can be made if specific sources of uncertainty are reduced through improved research? DIALOGUE, an integrated assessment model, has been used to answer these questions. Central in the approach of DIALOGUE is the concept of parallel modeling, i.e., for each step in the chain from emissions to climate change a number of equivalent models are implemented. The following conclusions are drawn: The key source of uncertainty in global temperature projections appears to be the uncertainty in radiative forcing models. Within this group of models uncertainty within aerosol forcing models is about equal to the total forcing of greenhouse gas models. In the latter group CO2 is: dominant. The least important source of uncertainty appears to be the gas cycle models. Within this group of models the role of carbon cycle models is dominant. Uncertainty in global temperature projections has not been treated consistently in the literature. First, uncertainty should be calculated as a product of all uncertainty sources. Second, a particular choice of a base year for global warming calculations influences the ranking of uncertainty. Because of this, a comparison of ranking results across different studies is hampered. We argue that 'pre-Industrial' is the best choice for studies on uncertainty. There is a linear relationship between maximum uncertainty in the year 2100 and cumulative emissions of CO2 over the period 1990-2100: higher emissions lead to more uncertainty.