The following was written by Richard Smith, Director of SAMSI.
What is it like to work on a committee of the National Research Council (NRC), especially if it’s on a topic that has the potential to make headlines? I had the chance to find out recently when I was asked to serve on the committee that produced the report Climate and Social Stress: Implications for Security Analysis (http://www.nap.edu/openbook.php?record_id=14682&page=1). Although the committee members and NRC staff were careful not to overplay the conclusions, the implications – that climate change could become a trigger for civil unrest and even war – were sure to attract attention.
It wasn’t the first time I had served on an NRC committee – some years ago, I served on a committee charged with reviewing one of the reports of the US government’s Climate Change Science Program. But that was very much an under-the-radar report, reviewing somebody else’s document rather than coming up with original conclusions of our own. This one was likely to be different.
The committee itself was a diverse group, including a former chairman of the National Intelligence Council and the former national security adviser to Vice President Al Gore, but was mostly composed of academics, including a number of social and political scientists, three climatologists, a decision science expert from RAND, and one statistician (me). The climatologists included two people I already knew well – Dave Easterling from the National Climatic Data Center (NCDC) in Asheville, and Otis Brown, Director of NC State’s Cooperative Institute for Climate and Satellites (also based at NCDC). Holding everything together was the study director, Paul Stern of the NRC. Paul did an amazing job of coordinating the entire activity, keeping everything on time and, I suspect, writing a considerable part of the actual text of the report.
There were four meetings of the committee, three at the National Academies main building in Washington and the fourth on the campus at Stanford (to even up, at least a little, the travel requirements on the east and west coasters). Each of the meetings was divided into private and public portions. The public portions included hearing from the study’s sponsor about the interest of the U.S. intelligence community in supporting this committee, and talks from outside experts (including Dan Cooley, who spent last Fall as a SAMSI visitor and led our spatial extremes working group).
So let me try to summarize some of the issues related to climate extremes, as they came up in this committee. There has been a lot of discussion in the literature about the extent to which the increasing frequency of extreme weather events may be said to have been a consequence of human-induced climate change. Superstorm Sandy is the most recent example of this kind of event. However, the experts are still not unanimous about the extent to which extreme events may be attributed to the human influence, and our committee decided not to try to address that issue. Instead, we focused on the changing probability of extreme events, and trying to understand how they are changing regardless of the underlying cause. Another feature of this committee’s work was to focus on fairly short-term events – projections to the end of the century may be of interest to the Intergovernmental Panel on Climate Change, but the feeling here was to concentrate on the next ten years. So the question became: how will probabilities of extreme events change over the next ten years? In the end we agreed that there was not enough published literature on this question to allow us to draw definitive conclusions, beyond a general agreement that such events are becoming more frequent. So we ended up with the rather imprecise but hard to argue conclusion: “Expect Surprises”.
Statistics of single extreme events, however, were not the only part of our focus. The report also discussed clusters of extreme events. The idea was that dependences between events in different places might lead to cascades of events, with possibly much greater consequences than just single extreme events. It seemed that bivariate (or spatial) extreme value theory might be a useful tool for this, which is why the committee invited Dan Cooley to speak.
Well, it turned out it wasn’t quite so easy to apply bivariate extreme value theory to the kinds of events the committee was interested in. One focus of our attention was the simultaneous occurrences of high temperatures in Russia and flooding in Pakistan during July and August 2010. A 2012 paper analyzed data from both these events and concluded that there was a plausible physical link between the two. However, an analysis of historical data did not find previous instances of this particular pairing of extreme events, so it was difficult to draw conclusions about joint tail probabilities.
Nevertheless, bivariate extreme value theory did prove useful in a different and unexpected context. Dave Easterling drew my attention to a 2008 paper that pointed out a connection between drought occurrences in the south west USA and the coast of Argentina (likely connected with the meteorological phenomena known as La Niña). Dave asked me whether this correlation would also apply to extreme events – the answer turned out to be yes.
So what was the take-home message? There are various plausible mechanisms by which climate change could affect national, and global, security, but quantifying the risk is far from easy. It was definitely a worthwhile activity in which to take part, and suggested a number of potential topics for my, and maybe also SAMSI’s, future research.
As a final irony, the report was due to be released to the public in Washington on October 31 – the day after Sandy struck much of the eastern seaboard. The press conference was hastily rescheduled for the following week. Expect surprises, indeed.