Varun Sivaram

Energy, Security, and Climate

CFR experts examine the science and foreign policy surrounding climate change, energy, and nuclear security.

Print Print Cite Cite
Style: MLA APA Chicago Close


How Likely Was Hurricane Sandy?

by Michael Levi
November 1, 2012


As the public debate over Hurricane Sandy turns in part from the immediate impacts of the storm to its possible links to climate change, a new theme is emerging: Sandy is not just part of a new normal – scientists actually predicted it well in advance.

One of the most prominent exhibits being used to back this case is a paper (PDF) published in Nature Climate Change earlier this year by Ning Lin, Kerry Emanuel, Michael Oppoenheimer, and Erik Vanmarcke. One blogger described it with the following headline: “Peer Reviewed Research Predicted NYC Subway Flooding by #Sandy”. CNN went with the following:

“[Scientists are] telling us we shouldn’t be surprised that this 900-mile-wide monster marched up the East Coast this week paralyzing cities and claiming scores of lives…. In a paper published by Nature in February, [Oppenheimer] and three colleagues concluded that the ‘storm of the century’ would become the storm of ‘every twenty years or less.’ New York Gov. Andrew Cuomo agrees. ‘After what happened, what has been happening in the last few years, I don’t think anyone can sit back anymore and say ‘Well, I’m shocked at that weather pattern,’’ Cuomo said Tuesday.”

The Lin et al paper is meticulous and absolutely fascinating. (I particularly recommend reading the introduction for a primer on how climate change does and doesn’t influence storm surges.) Everything that Oppenheimer told CNN is sound. But the paper does not say what the journalists and pundits claim or suggest that it does. A more careful reading of it points to more subtle – and intriguing – conclusions.

The Lin et al paper uses four different climate models to estimate the frequency of different storm surges at New York’s Battery both under present conditions and, assuming rising greenhouse gas emissions, in a hundred years. The results are summed up in this figure:

What do their models say about the present probability of a 4.16 meter flood surge, which is what we saw during Sandy? One of the four models says that it’s probably a less-than-one-in-ten-thousand-year event given current climatic conditions. A second model puts the odds closer to one-in-eight-thousand-years or so. (Take these numbers with a big grain of salt – I’m reading them off the above figure with a ruler.) The final two models put the odds at between one-in-two-thousand and one-in-three thousand. The models themselves, of course, have internal uncertainty, which the paper takes into account. But even the most foreboding of the analyses suggest that there is at most a one-in-ten chance that surges like the one we saw this Monday have become more-than-one-in-a-thousand-years-or-so events. The takeaway here is simple: none of the models said that the storm was anywhere remotely close to likely given current climate conditions. The message from those models is pretty much that Sandy was a very unlucky fluke.

The far more worrying part of the paper is what it projects for the future. I’m a focus-on-risk guy so I’m going to concentrate on what the paper says is unlikely but plausible. What the paper basically says is there’s a one in ten chance that storms like Sandy will become one-in-a-few-hundred year events. But the paper also says that flood surges north of about 1.5 meters – currently enough to create flooding in downtown New York – will go from one-in-nearly-a-hundred-year events to something that happens more than once every decade.

What I would really love to see (and I know that this isn’t simple to do) is a backcast from Lin et al that estimates what the likelihood of a Sandy-scale flood surge would have been a hundred years ago. That would help us answer the question on everyones’ minds: how much did climate change contribute to Sandy?

Post a Comment 2 Comments

  • Posted by Michael Wara


    I enjoyed your discussion of the Lin et al. paper but I think that the paper (to some extent) as well as your discussion underplays some important points. The key issue for storm surge in NYC is, in the paper’s terms, the annual frequency of various types of “NY Region storms.” The reality is that, as the paper acknowledges (page 2 paragraph 2), GCMs (1) cannot actually model hurricanes and (2) that to the extent that hurricane statistics can be generated from GCM model output, the numbers are all over the map, and (3) do not all show the same sign or magnitude of change under identical forcing. The underlying issue is that changes in surface ocean temps and vertical wind shear have opposite effects on tropical cyclone formation. The different models predict different magnitudes of change to both variables under climate change leading to widely varying predictions for cyclone formation.

    Take home: Using statistical models that are coupled to current generation GCMs to get at hurricane frequency is just not likely to be terribly accurate. One indicator of the challenge is that Lin et al. do not use model output for their estimate of current NY region storm intensity – they use data from the historical record. That’s because the models would likely show similar dispersion for current storm intensity as they do when forced with a climate change scenario.

  • Posted by David B. Benson

    Michael — The answer to your question is ‘some’. Enhanced sea surface temperatures late in the season is part of it. Much more problematic is attributing the blocking high over Greenland to climate change although I would be inclined to, a little.

    What is simply luck is the landfall location with timing at a full moon enhanced high tide. Blame that on the Halloween goblins.

Post a Comment

CFR seeks to foster civil and informed discussion of foreign policy issues. Opinions expressed on CFR blogs are solely those of the author or commenter, not of CFR, which takes no institutional positions. All comments must abide by CFR's guidelines and will be moderated prior to posting.

* Required