David Keith

Harvard University, Atmospheric Research Project

David Keith and Granger Morgan (of the Carnegie-Mellon cross- disciplinary group) interviewed 16 climate experts in order to:

• better quantify existing uncertainty in climate prediction
• research resource allocation questions and cost of information
• seek support for other integrated assessment efforts
• improve techniques for eliciting expert judgments in technical domains
• improve communication of uncertainty between climate prediction community and impact assessment community - especially what things each wants and needs to know from the other.

Face-to-face interviews consisted of an introduction, general discussion, uncertainty in prediction of policy-relevant variables, disaggregated sources of uncertainty in global change in temperature, research resource allocation, and surprise. On the most basic questions, the experts agreed about the answers as well as the range of uncertainty. On more difficult questions, such as how much will the gradient between the poles and the equator change or how will precipitation change, there is much less agreement.

Key issues in the general discussion section included mega-models (don't trust results but the process of constructing these things is useful and important; there's a failure to link disciplines); down scaling; modeling versus data collection; mission-oriented versus curiosity-driven research; linking focused studies to global models.

The budget allocation exercise, which asked each expert to make funding choices for the US Global Change Research Program, found general consensus that too much money goes into NASA's Earth Observing System (EOS); many thought we could have multiple, smaller, more flexible platforms that would be more effective than the large, expensive EOS we now fund. Another interesting outcome was that the experts believe there is a chance (~20%) that after a 15- year research program funded exactly as they wanted, there would be an increase in uncertainty about climate change.

There appeared to be evidence of a possible bandwagon effect: there was closer agreement on the temperature sensitivity issue than on other questions, perhaps because there has been more discussion of that question and people have been influenced by the opinions of others.

How to determine who is a credible expert is still a key question, and who you choose will always bias the results to some extent. This group used three criteria in choosing their experts: prominence, location (due to funding realities), and a matrix of subject areas to cover. When trying to assess questions of surprise and uncertainty, techniques such as this survey of experts can be a good way of trying to assign probabilities to events.