by Judith Curry
How and why did the scientific consensus about sea level rise due to the disintegration of the West Antarctic Ice Sheet (WAIS), expressed in the third Intergovernmental Panel on Climate Change (IPCC) assessment, disintegrate on the road to the fourth?
The Rapid Disintegration of Projections: The West Antarctic Ice Sheet and the Intergovernmental Panel on Climate Change
Jessica O’Reilly, Naomi Oreskes, Michael Oppenheimer
Abstract. How and why did the scientific consensus about sea level rise due to the disintegration of the West Antarctic Ice Sheet (WAIS), expressed in the third Intergovernmental Panel on Climate Change (IPCC) assessment, disintegrate on the road to the fourth? Using ethnographic interviews and analysis of IPCC documents, we trace the abrupt disintegration of the WAIS consensus. First, we provide a brief historical overview of scientific assessments of the WAIS. Second, we provide a detailed case study of the decision not to provide a WAIS prediction in the Fourth Assessment Report. Third, we discuss the implications of this outcome for the general issue of scientists and policymakers working in assessment organizations to make projections. IPCC authors were less certain about potential WAIS futures than in previous assessment reports in part because of new information, but also because of the outcome of cultural processes within the IPCC, including how people were selected for and worked together within their writing groups. It became too difficult for IPCC assessors to project the range of possible futures for WAIS due to shifts in scientific knowledge as well as in the institutions that facilitated the interpretations of this knowledge.
Social Studies of Science published online 26 June 2012. [link] to abstract
Some background information from the Introduction:
In September 2009, dozens of scientists who study the West Antarctic Ice Sheet (WAIS) met for their annual workshop at the Pack Forest Conference Center, close to Mt Rainier National Park in the state of Washington. Among the residential cabins and dining hall and in a stand of second growth Douglas firs sat a large log-hewn structure, formed like a gigantic one-room schoolhouse. Inside, fleece-garbed and predominately male Antarctic scientists sat at rows of tables facing the front of the room, listening to research updates from their colleagues.
One presentation hit a sore spot among WAIS scientists. It addressed the stability of the WAIS and whether it would grow (due to additional snowfall) or shrink (due to melting and iceberg formation) in a warming world, thus subtracting from or adding to future sea level rise. The most recent assessment from the Intergovernmental Panel on Climate Change (IPCC) omitted any specific prediction of whether WAIS would lose ice, and if so, how slowly or rapidly it would do so. (A rapid loss is sometimes labeled a ‘collapse’ in colloquial terms, although this is a bit of a misnomer since the fastest conceivable disintegration would stretch over hundreds of years.) In the third assessment report (IPCC, 2001; abbreviated as TAR), chapter authors cited high uncertainty, but they also provided long-term projections for the highest potential rate of ice loss (called an ‘upper bound’) during such a ‘collapse’.1
Based on the then-recent literature, they concluded that WAIS would be stable in the short term. However, in the Summary for Policy Makers (SPM) of the fourth and most recent IPCC assessment report (IPCC, 2007; abbreviated as AR4), the authors stated that although relatively rapid loss of ice was already observed from parts of WAIS (due to melting and iceberg formation as the ice flowed off the ice sheet into the ocean), they could not provide an estimate of long-term behavior and the resulting contribution to sea level rise. In addition, the authors decided to exclude from their tabulated, numerical estimates of 21st century sea level rise the possibility of any further changes in flow rate (called ice dynamics) from either Greenland or Antarctica . In short, in the TAR, WAIS was deemed stable in the short term (through 2100) and a highly uncertain numerical estimate was provided for rapid disintegration in the long term. While the AR4 was being written, new observations undermined the previous consensus that WAIS would not contribute significantly to short-term 21st century sea level rise. The great difficulty with coming up with a credible numerical estimate for the short term led to no WAIS prediction – in the short or long term – in AR4. How and why did the scientific consensus expressed in the third IPCC assessment disintegrate on the road to the fourth?
Due to the recent observations and the resulting challenge to the models, AR4 authors decided that there was not enough agreement in the broader expert community to provide an assessment of the likelihood of rapid disintegration. They wrote in the SPM: ‘dynamical processes related to ice flow not included in current models but suggested by recent observations could increase the vulnerability of the ice sheets to warming, increasing future sea level rise. Understanding of these processes is limited and there is no consensus on their magnitude’ (IPCC, 2007: 17). Rather than allowing the IPCC to estimate the potential rate of disintegration over the long term, as with the third assessment, the new observational information gave the authors pause on providing a numerical prediction.
For the near term (21st century), AR4 does not explicitly present a numerical WAIS contribution (TAR did not do so either). Instead it assumed that the dynamical contribution to sea level rise from both the Greenland and Antarctic ice sheets would continue into the future at the same rate indicated by recent observations. The table (below) provides multiple sea level rise estimates based on several greenhouse gas emission scenarios, but notes that the numbers provided are a ‘model-based range excluding future rapid dynamical changes in ice flow’ (IPCC, 2007: 13). Though the later textual caveat suggests that the contribution of the ice sheet to a rise in sea levels could be significantly higher, this statement was overshadowed by the numbers, in the minds of many readers.
Conclusions: Collapsing projections
The disintegration of WAIS projections in the AR4, compared with previous IPCC assessment reports, was the result of multiple forms of novelty colliding with one another: new data, new writing teams, higher visibility for the IPCC, and new structure for the report made an epistemic and institutional jumble for IPCC authors to try to deal with. While the authors agreed that the new data were the primary impetus for the collapse of consensus in this case, the ways in which people and chapters were organized accentuated the high degree of uncertainty surrounding WAIS.
IPCC reports have a specific sort of documentary life, as well as a specific and developing relationship with policymakers. IPCC assessment reports are written with the understanding that there will be subsequent IPCC reports. In this sense, IPCC reports are provisional. In addition, the ways in which chapters are organized, writing teams are formed and managed, and even how the various deadlines are set are as integral to the IPCC process as are the scientists and their publications.
As described in this paper, IPCC authors also sometimes chose to enable some projections while inhibiting others. Other writing teams had been able to make projections of some sort even when consensus was limited. For example, in the case of climate sensitivity, the authors showed the full, wide, divergent spectrum of plausible outcomes and associated probabilities. Similarly, when assessing the combined ice sheet contribution to long-term sea level rise, Working Group II presented a range of potential outcomes and characterized its confidence in this assessment. In contrast, the approach taken by Working Group I was to step back from presenting a wide range of possibilities. It would be necessary to study many more such cases and also to undertake a deep analysis of the way these assessments are taken up by the policy community to decide which approach is more efficacious.
The decision by the AR4 authors not to make a WAIS prediction emerged from the complex arrangements needed to write a scientific assessment, including the translational tuning necessary for relating data to models and for situating uncertain science into a policy-relevant document. Foucault (1977: 189) suggests that individuals within institutions become emplaced into a ‘network of writing’, becoming entrenched in practices and habits that create policies and other official documents. While the AR4 authors formed such a network, this configuration did not account for institutional misteps – states, bureaucracies, and chapter organization in scientific reports. In the WAIS example, the ‘network of writing’ struggled when the practices and habits of the institution changed, as well as when a radical shift occurred in the subject matter being written about.
The people interviewed about this case never mentioned pressure to come to a consensus, though they suggested various strategies for trying to come to an agreement. Perhaps the authors were correct in not trying to enforce consensus about the rapid disintegration of WAIS when it was clear that there was none.
Alternately, the authors could have considered publishing the range of possible numbers and scenarios, highlighting their lack of consensus and the high uncertainty surrounding the rapid disintegration of WAIS.
How are models, assessments, and projections translated from science into policy relevant documents? We argue that the institutions, such as the IPCC, are acting as the translators and mediators. By examining WAIS projections in IPCC reports – particularly the omission of a prediction in the most recent report – translation from science to policy is enabled as well as hindered by the IPCC institution, an important actor in this drama.
In AR4, expert authors unmade WAIS projections, citing new data, a lack of consensus, and sheer ignorance. The authors also had to grapple with significant organizational challenges.
The IPCC exists to enable production of climate change assessments in a credible manner, and it generally succeeds in doing so. However, the assessment reports shape and are shaped by processes meant to streamline and organize the writing process. Shackley (1997: 79) noted that concern over IPCC chapter writing groups and the IPCC peer review process mimics an ‘enclave’ with a tendency to ‘close ranks’ when faced with criticism.
Elzinga (1996) stated that the IPCC’s apparently closed assessment process can lead to epistemic and paradigmatic skewing. These processes have been subject to critique and review, most notably in the InterAcademy Council review published in 2010, following some errors detected in AR4 (Committee to Review the Intergovernmental Panel on Climate Change, 2010). The IPCC is an institution in the service of science and policy, and it borrows an admixture of organization styles from both.
Within the IPCC, authors, delegates, and IPCC employees use management strategies from both international diplomacy and scientific practice to create each assessment report, using hybrid management to deal with the complexities of research and policymaking in international governance. Hybridizing the powerful regimes of international policy and science, though, requires significant translation between the subjective and contingent understandings of each. Such hybridization also falls under the mantle of ‘science diplomacy’, where scientific information is applied to policy goals.
JC comments: This is a very interesting case study of IPCC deliberations. There are no ‘shenanigans’ on the part of the scientists here, in fact Jonathan Gregory stands out as trying to do a consistent job of accurately presenting sea level rise information across the different chapters in the AR4. The problems as I see it are overconfidence in the TAR, and a failure in AR4 to explicitly confront the changes from TAR to AR4 and clearly identify the reasons for this and the increased understanding of the scope of the uncertainties and ignorance. The end result was a confusing message to scientists and policy makers alike regarding the topic of the potential role of WAIS in future sea level rise.
The problems that are arose are endemic to the IPCC assessment process itself. My preference is to include the full range of plausible scenarios (including back of the envelope, semi-empirical and model-based), including a qualitative likelihood assessment if possible. Because each IPCC assessment report is a provisional document with the expectation of a future assessment, it is important to explicitly note changes from the previous assessment report, including growing uncertainties.
It is clear that there is no ‘consensus’ at all on this topic, and the choice to simply leave out a possible WAIS contribution to sea level rise (possibly positive or negative) doesn’t seem to be useful to policy makers.
Refer to my draft paper No consensus on consensus for my further analysis on how the IPCC might proceed in this type of situation (which is more endemic to the subject IMO than one is led to believe by the IPCC ‘confidence’ statements).