by Judith Curry
I’ve spent the last several days at the AGU meeting in San Francisco. With 19,200 participants, there is an overwhelming amount things going on all that the same time. Here are some highlights of the meeting (some of which I witnessed first hand, others are based on second hand reports). And also some comments on some of the controversies that are being discussed in the blogosphere.
Featured video lectures are now available on this website. Of particular relevance to climate, see
- John Holdren: Scientists, Science Advice, and Science Policy in the Obama Administration
- Julia Slingo: Society’s Growing Vulnerability to Natural Hazards and Implications for Geophysics Research
- Michael Oppenheimer: Scientists, Expert Judgment, and Public Policy: What is Our Proper Role?
- Tim Palmer: A Very Grand Challenge for the Science of Climate Prediction — Towards a Community-Wide Prototype Probabilistic Earth-System Model
- Ellen Mosley-Thompson: Past and Contemporary Climate Change: Evidence From Earth’s Ice Cover
- Didier Sornette: Dragon-Kings, Black Swans and Prediction
The only talks on this list that I saw were Tim Palmer and Didier Sornette. Tim Palmer’s presentation was superb and very relevant to our discussions of climate model uncertainty. Sornette’s talk certainly had the coolest title of the conference. It was very interesting and I learned a lot, but if you listen to the presentation, there is diminishing returns after the first 10 minutes or so.
My favorite talk in the regular sessions was “Power Law and Scaling in the Energy of Tropical Cyclones” by A. Corral, A. Osso; J. LLebot. They have a recent Nature Physics publication with the same title that is behind paywall.
The AGU had a session on “Which of these books have you read?” which highlights popular books authored by AGU members, on topics related to global environmental change. I didn’t see any skeptical books on the list, not sure how to interpret that. I guess a number of people were invited to participate, not all accepted the invitation.
I didn’t attend the session, but received reports from several scientists, including several who actually walked out in the middle. Overall, the session was rather lackluster. Greg Craven made a presentation on his book “How it all ends” that has been highly controversial.
This was followed by a panel discussion with the bestselling authors and Michael Oppenheimer on “What should be the role of AGU members in the current public discussion of climate change?” I did not attend this session, either. From reports that I’ve heard, this session was aptly described by Steve Mosher at WUWT, including a good talk by Oppenheimer, and another bizarre performance by Craven.
A few words are in order about Greg Craven. Before this meeting, I had never heard of him. Apparently he is a high school teacher, I am not exactly sure why he is a member of the AGU. Apparently his book and youtube video are pretty good (I watched the youtube video). I can’t imagine how to explain his behavior at AGU. I have no idea how to explain how/why he was invited to participate in the AGU. If I had been session chair, I would have interrupted his performances at several junctures (but that is said in hindsight, I can’t imagine what was going on in the session chair’s mind during this performance).
In any event, the AGU has a new emphasis on communication, which I think is a good thing, but they are still finding their way . . . I suspect that they’ve learned not to invite Craven to speak at future meetings.
In a section on controversies, its difficult to leave out Mike Mann. He spoke in a session entitled “Institutional Support for Science and Scientists in an Age of Public Scrutiny.” Mann’s talk was entitled “Climate Scientists In The Public Arena: Who’s Got Our Backs?” It seems that what Mann has learned from his experiences over the past several years is how to write op-eds for the Washington Post.
Conversation with Chris Mooney
I would like to address the post made at WUWT on the selection of Mooney and Floyd DesChamps to the AGU Board of Directors. As per the AGU press release, these appointments were made to bring expertise in science policy and communication. “Their selection reflects AGU’s commitment to applying the results of scientific research to challenges faced by the global community, many of which are based in the geosciences.” The article at WUWT describing this press release started with “Pigs have been flying at AGU, apparently. All hope is lost for this organization. Get out while you can.”
I had an extensive conversation with Chris Mooney at the AGU meeting. For background, I first met Mooney Jan 2006 at the annual meeting of the American Meteorological Society. I had just read his first book The Republican War on Science, and was delighted to learn that he was working on a new book about the hurricane “wars.” The eventual book “Storm World” is a superb piece of science journalism, which I reviewed at amazon.com. I was also one of three people he asked to write a recommendation letter to support his application for Knight Science Journalism Fellow at MIT.
Mooney was aware of the WUWT piece, and said that it wasn’t receiving any attention at all by the AGU, as far as he could tell. Politically, Mooney is definitely a Democrat. He pointed out to me that Floyd DesChamps is actually a Republican, who was McCain’s staffer in the preparation of the McCain-Lieberman bill. A “green” Republican, but a Republican nevertheless.
Mooney’s actual contribution to the AGU meeting included two very well received presentations (skeptic Jim OBrien mentioned to me that he wanted to buy Mooney’s book Unscientific America based on his presentation), one of which is described here.
I like Chris Mooney, and have learned a lot from him about communicating science. Do I agree with everything he says? Well, I don’t read everything he says, and I am sure there is much out there that I would probably disagree with. But I think he is a good addition to the AGU Board. Yes, he is an english major, but he is a science journalist, and spent a year at MIT (last year) taking science courses related to climate, energy, biotech. IMO climate science gets a low grade in both communication and in interfacing with policy. The AGU is trying to address this issue in a sensible and productive way.
FYI, Mooney blogs at The Intersection.
My talk entitled “Climate surprises, catastrophes, and fat tails” is posted here. Note, this topic will be covered in two forthcoming posts (before the end of the calendar year). Some background info is provided on the Scenarios thread.
The subtitle of my talk is: How the decision-analytic framework is influencing the scientific interpretation and assessment of climate change uncertainty.
Many of you have seen this figure before, its from the IPCC 4th Assessment report. The Figure provides the probability density of the equilibrium climate sensitivity to doubled CO2. The curves in this figure are obtained from both model simulations and observations, including paleo.
Now when you look at this diagram, do you find yourself trying to identify some sort of “best estimate”, like 3C? Or a likely range, such as between 2 and 4.5C? Or are you intrigued by what might be going on at the tails of the distributions? In my talk today, I argue that how most people look at this diagram, including climate scientists, is conditioned by the decision analytic framework associated with the precautionary principle and optimal decision making. The consequences of this for both climate science and policy making are discussed.
The precautionary principle states that lack of full scientific certainty should not be used as a reason for postponing action to prevent dangerous climate change. So how much certainty do you actually need to trigger the precautionary principle? Well, that is fuzzy, but the issue seems to be identifying what constitutes dangerous climate change and eliminating the likelihood that the magnitude of the sensitivity will be small. Apparently the level of certainty in the IPCC assessments has been deemed sufficient, since the UN Framework Convention Treaty has established a goal of stabilization of greenhouse gases.
So now that the precautionary principle has triggered an international treaty, what should the CO2 stabilization target be? The optimal decision making model works in the following way: more research reduces uncertainty, which builds a scientific and political consensus that results in meaningful action. When uncertainty is well characterized and the model structure is well known, classical decision analysis can suggest statistically optimal strategies for decision makers. Climate models are used to optimize the stabilization targets. Do we really have enough confidence in the climate models to use them to set a stabilization target?
Optimal decision making strategies are arguably a mismatch for the climate change problem owing to the deep uncertainty surrounding climate sensitivity, among other sources of uncertainty. Robert Lempert characterizes the decision making environment surrounding climate change as one of deep uncertainty, owing to long time horizons, substantial uncertainty in our understanding of the climate system, and the potential for surprises. Does the deep uncertainty characterization mean that we should not act on the threat of climate change?
Not at all, but deep uncertainty changes the environment for making robust decisions, relative to optimal decision making. Robustness is a strategy that seeks to reduce the range of possible scenarios over which the strategy performs poorly. Robustness is a property of both the degree of uncertainty and the richness of policy options. Robust strategies consider unlikely scenarios and the possibility of surprise.
This brings us to catastrophes and surprises. Low probability, high consequence events provide particular challenges to developing robust policies. A recent paper written by economist Martin Weitzman used an expected utility analysis to argue that the fat tail associated with unexpectedly but not impossibly high values of climate sensitivity dominate climate change economics.
So lets take another look at this climate sensitivity figure. The issue with triggering the precautionary principle is arguing that low values of sensitivity are unlikely. The issue with robust decision making is developing a sufficiently broad range of scenarios that include surprises, i.e. out there on the tail of the distribution. The “best estimate”, which is the current focus of much scientific research, is less meaningful for robust decision making strategies than it is for optimal decision making.
Trying to generate a pdf from these distributions is not simple, and in any event would not be straightforward to interpret.
Possibility theory and the possibility distribution is arguably a better match for the scenario driven robust decision making process. Possibility theory is an imprecise probability theory whereby the possibility distribution distinguishes what is plausible versus the normal course of things versus surprising versus impossible.
The possibility distribution can be developed in the context of scenario creation and assessment using modal logic, which frames possible versus not possible worlds. A recent paper by Betz, who is a philosopher of science, argues for a new principle for constructing scenarios of future climate.
He classifies the IPCC’s method for scenario creation as modal induction, whereby future scenarios are inferred from modeled emissions scenarios that force climate model simulations. Betz argues for creation of future climate scenarios by modal falsification, which permits creatively constructed scenarios as long as they can’t be falsified by being incompatible with background knowledge.
Why do I think this idea of modal falsification is significant? Our current focus on the “best estimate” has directed the scientific focus away from the possible or plausible worst case scenarios, and they are not considered if they are not inferred from the climate models. Such scenarios are vaguely mentioned to provide a sense of “dangerous” in the context of motivating action under the precautionary principle. But where you actually draw the line in terms of what is possible versus highly implausible over a target time range is very important in the context of robust decision making and the identification of possible catastrophes and surprises.
The cartoon illustrates this challenge in the context of a discussion among dinosaurs: “I’m telling you we have to be prepared! An asteroid or comet collision is inevitable, aid if we’re not ready we could be made extinct. Oh Pshaw! We’ve ruled this planet for 160 million years and we will always rule it! Next thing you’ll be telling us to watch out for the mammals! HA HA!”
Further, in constructing the possible worst case scenarios, we have focused on the greenhouse gases, with insufficient attention paid to natural climate variability and abrupt climate change, independent of or amplified by greenhouse warming.
I hope I’ve changed how you look at this diagram and how you consider the uncertainty associated with climate sensitivity. As scientists, we should be exploring and trying to understand this uncertainty, particularly at the tails. I am all in favor of reducing uncertainty through improved understanding and improving climate models, but the complexity of the system guarantees that uncertainty is likely to continue to dominate this problem for the foreseeable future. By prematurely forcing consensus on a “best estimate,” we do a disservice to the science as well as to policy makers.
In conclusion: The drive to reduce scientific uncertainty in support of precautionary and optimal decision making strategies regarding CO2 mitigation has arguably resulted in:
- unwarranted high confidence in assessments of climate change attribution, sensitivity and projections
- relative neglect of defining and understanding the plausible and possible worst case scenarios
- relative neglect of decadal and longer scale modes of natural climate variability
- and conflicting “certainties” that result in policy inaction
A way forward is the decision analytic framework of robust decision making under deep uncertainty, which emphasizes scenario discovery and uncertainty analysis and identifying a broad range of robust decision strategies.
Implications of such a strategy for climate research are an increased emphasis on:
- exploring and understanding the full range of uncertainty
- scenario discovery using a broader range of approaches
- natural climate variability, abrupt climate change, and regional climate variability