MAGICC is a simple climate model that plays a number of important roles in climate modeling and climate policy, including a contribution to all five IPCC assessment reports. In this seminar, Dr Roger Bodman will review approaches to calibrating the model, reducing uncertainty in its projections, and managing the risks of climate change when dealing with uncertainties.
Abstract
MAGICC is a simple climate model that has been involved with all of the IPCC’s five assessment reports. It has played, and continues to play, a number of important roles in climate modeling and climate policy, for example, as an emulator for fully coupled complex models; for producing probabilistic estimates of global-mean temperature change; and for estimating the carbon budget for meeting the 2°C target. MAGICC is also used to convert emission scenarios to concentration pathways for AOGCMs, as well as being the ‘temperature change’ calculator in a number of Integrated Assessment Models.
In this seminar, I will review some approaches to calibrating the model’s parameters using a Bayesian statistical data assimilation process based on the Monte Carlo Metropolis-Hastings (MCMH) algorithm. Differences to other studies are noted and some issues, such as the choice of prior distributions and observations, discussed. Results for climate sensitivity in particular will be presented and compared to other studies that have used simple climate models for this purpose.
Probabilistic projections based on the Representative Concentrations Pathways (RCPs) will also be presented, contrasting results using the MCMH approach with emission driven RCP scenarios to the RCP results reported in the IPCC Fifth Assessment Report. The spread in the MAGICC results is considered in terms of the sources of uncertainty. What are the main contributors and what is the potential for reducing this uncertainty in the short-term?
Given that there is little scope for reducing scientific/model uncertainty in the short-term, what is the appropriate risk management response? It seems that the best way to manage the risks of climate change are to take action to significantly reduce greenhouse gas emissions so that the risks associated with them rather then have to deal with the subsequent impacts. Is this a fair summary and what then are the implications for further research?
Biography
Roger Bodman joined Victoria University in early 2012 after completing a PhD in climate science at the University of Melbourne under the supervision of Profs David Karoly, Peter Rayner and Ian Enting. He originally trained as a manufacturing engineer and computer programmer, accumulating over 20 years of experience in industry. A varied range of interests led to a return to higher education, eventuating in a BA with a major in Sanskrit along with studies in the history and philosophy of science, politics and environmental sustainability. He went on to undertake an MA in the Department of Philosophy at the University of Melbourne. Work as a Research Assistant led to investigating nuclear energy and climate change, including contributions to the Garnaut Climate Change Review. This work then transitioned into a PhD investigating climate change and uncertainty using a simple climate model. Current research is focussed on methods of calibrating this climate model against historical observations using a Bayesian statistical method in order to obtain probabilistic projections for future global-mean temperature change.