Back to CDIAC Home Page
A Plan for Intermodel Comparison of Atmospheric CO2 Projections with Uncertainty Analysis
(June 1990)
Projecting future concentrations of atmospheric carbon dioxide (CO2) is one of the principle objectives of the U.S. Department of Energy's Carbon Dioxide Research Program (CDRP). These projections are needed to assess the likelihood of significant global and regional change as a consequence of the continued use of fossil fuels and to determine whether alternative scenarios of future energy use can significantly alter this likelihood.
The projection of future atmospheric CO2 concentration requires (1) estimates of the anthropogenic release of CO2 to the atmosphere from the combustion of fossil fuels and changes in land use and (2) global carbon cycle models describing the atmospheric retention of that CO2. However, all model projections of future CO2 concentrations involve some degree of uncertainty. Much of this uncertainty can be attributed to (1) uncertainties in future energy and land-use emissions, (2) uncertainties about the global carbon cycle reflected in the structural and conceptual differences between models, and (3) measurement error and uncertainty in the parameters and variables within a particular model. Fortunately, methods for quantifying model uncertainty exist so that projects can be made with confidence limits that reflect the associated uncertainties.
This document is a plan for an intermodel comparison of atmospheric CO2 projections that includes uncertainty analysis of the global carbon cycle models used to make those projections. The plan includes a procedure for the documentation, support, and archiving of global carbon cycle models within the Carbon Dioxide Information Analysis Center (CDIAC) at Oak Ridge National Laboratory (ORNL).
Uncertainty analysis is the examination of uncertainties in predictions from simulation models. The analysis identifies the dependence of model predictions on inputs, initial conditions, and parameters. Uncertainty analysis of global carbon cycle models can identify carbon cycle components and processes with the greatest sensitivities and uncertainties. This information can then be used to determine which uncertainties have the greatest influence on future atmospheric CO2 concentrations and where further research and data collection could be most effectively applied to reduce uncertainties. An intermodel comparison of atmospheric CO2 projections with uncertainty analysis can help define research that will improve model performance.
Back to CDIAC Home Page