Communicating risks in IPCC reports
While serving on the National Research Council’s Committee for America’s Climate Choices, 2009-2011, Socolow became aware of the difficulties experienced by the Intergovernmental Panel on Climate Change (IPCC) in communicating risks to the public. Helped by private exchanges with a few IPCC leaders, Socolow was able to reconstruct some of the internal disagreements that had prevented the IPCC from dealing in a forthright way with the possibility of the early arrival of one or more of the widely discussed very nasty disruptions to societal well-being associated with climate change. He developed his findings in an article in the October 2011 issue of Climatic Change, “High-consequence outcomes and internal disagreements: tell us more, please.” The entire issue, edited by fellow Policy & Integration Group member Michael Oppenheimer and Gary Yohe, is devoted to the communication of uncertainty in IPCC documents.
Socolow reveals that in the preparation of the most recent IPCC report on climate science (the report of Working Group 1 for the Fourth Assessment Report), the lead authors were unable to reach agreement about how to express the probability of responses of the climate system to climate change that are worse than those falling in a band called “likely.” Figure 20, which was not published, was prepared at a time when some authors wanted to assign this region a probability of “17%,” while others wanted to say only that responses in this zone “cannot be excluded,” without assigning any numerical probability. In the end, “cannot be excluded” was chosen, and the fact that two views were in contention was not reported. Socolow urges the authors of the next report (the Fifth Assessment Report), currently in preparation, to be more forthcoming about their differences. He also proposes specific rules that would encourage the use of quantitative analysis consistently.
Socolow and Oppenheimer have been able to recruit several formidable faculty members in the social sciences and humanities to explore the communication of uncertainty through a new university-wide program. Led by Robert Keohane, a political scientist who studies international institutions, a faculty group has embarked on a three-year project sponsored by the Princeton Institute for International and Regional Studies (PIIRS). The project brings together climate scientists, ethicists, and experts in several social sciences, as well as people who are specifically interested in communication across disciplines. Senior and junior visiting scholars have been recruited to be in residence during academic year 2012-13, when the project will be in high gear.
Policy-relevant information on sea level rise
Despite increasing awareness of the potential for rapid ice sheet change, sea level rise (SLR) projections in the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) decreased as their quantitative uncertainty range narrowed. Effective policy decisions demand a more complete and transparent accounting of sources of sea level change. However, weak observational constraints on the century-timescale ice sheet response, and the lack of a comprehensive model, have led to difficulties in quantifying the SLR contribution of Antarctica and Greenland. The Oppenheimer group is focused on 1) presenting ice sheet projections in a form that is amenable to climate-related decision-making and 2) improving the comprehensiveness and robustness of ice sheet models.
Improving the utility of sea level projections
Last year, the Oppenheimer Group presented a novel approach to projecting ice sheet-driven sea level change that integrates expert judgment, data constraints, and state-of-the-art coupled models. Because SLR projections using this method are probabilistic and explicitly account for the locations and rates of ice discharge, ice loss rates implied by global paleoclimate and/or semiempirical analyses may be examined in greater detail. This probabilistic framework also facilitates risk analyses and uncertainty reduction efforts.
An initial demonstration of the technique suggests that sea level rise (SLR) from Antarctica of greater than 40 cm by 2100 (implied by several recent analyses) is extremely unlikely (see Figure 21). The form of ice discharge scenarios, their associated uncertainty, and the correlation in ice loss between ice drainages control the tail area of sea level distributions (the information that is most relevant for climate policy and coastal management decisions). This initial analysis will be improved by the assimilation of regional and simplified dynamic models and data constraints. Researchers plan to incorporate this framework into integrated assessment models and decision analyses, with the ongoing goal of improving communication across the science-policy interface.
Ice sheet-ocean coupled modeling
The Oppenheimer Group employs a hierarchy of models to study physical processes at the margins of ice sheets. With collaborators at GFDL, Johns Hopkins, and MIT, they are developing a state-of-the-art model that can examine key physical processes in the ice-ocean system and be incorporated into larger-scale numerical models, elucidating global climate/ice sheet feedbacks. Simulations conducted with these models have already identified several key steps in the evolution of the coupled system, illustrating the time and spatial resolution required by largescale models.
Researchers have also studied the sensitivity of ice streams to ocean temperature, showing a response that is driven by favored locations of ice shelf thinning (Figure 22). Ocean- or iceinduced changes in ice thermodynamics, iceberg calving, and shear margin strength may significantly change these results, and the group continues to develop process-based parameterizations to assess their influence.
Given uncertainty in the timing and nature of historic forcing, the transient present-day state of the ice sheet, and parameter and model uncertainty, constraints on the past behavior of ice streams may be best obtained using an ensemble approach. We have recently performed a perturbed physics ensemble using a simplified coupled model. The Bayesian precalibration technique we are employing is applicable to large-scale, more complex models, and will improve uncertainty characterization in sea level projections.