Developments in detection and attribution methods and their implications for impact attribution

Wednesday, 17 December 2014: 1:40 PM
Myles Robert Allen, University of Oxford, ECI/School of Geography and the Environment, Oxford, United Kingdom
I will review recent developments in the detection and attribution of anthropogenic influence on both climate and human and natural systems affected by climate in the light of the recent Working Group 1 and 2 reports of the Intergovernmental Panel on Climate Change (IPCC). On traditional questions such as how much of the global mean warming observed over the past 60 years can be attributed to human influence on climate, I will show how a very simple analysis, readily automated, effectively reproduces the results of much more complex studies and might therefore form the basis for a routine update that might be performed whenever the observational record is extended or a new set of climate simulations are performed. While it would not obviate the need for further exploration of detection methods, publishing such an algorithm and updating results every year, much as the observed temperature record is updated, would provide a much-needed anchor point for attribution assessments. At the other end of the spectrum of detection and attribution challenges, an important distinction is emerging in the climate impacts literature between known sensitivities of systems potentially affected by climate change and direct observations of impacts of human influence. I will argue that, in terms of the IPCC¹s Good Practice Guidance Paper on detection and attribution, this distinction maps precisely onto the distinction between single-step and multi-step attribution, implying that a much broader range of impacts might be attributable to human influence on climate, in the multi-step attribution sense, than the impact community has agreed on so far. In the interests of clarity of future assessments, a debate is required now across the climate research community whether the single-step/multi-step framing is acceptable, in the hope of arriving at a consensus approach in time for future assessments.