more about hurricanes

CNN has an article about the same old debate that I've blogged about before: does global warming affect tropical cyclones? [LINK][old post]. Kerry Emanuel has a new study coming out in Nature that shows that hurricanes in the Atlantic and Pacific have gotten bigger and stronger over the past few decades, and apparently he's convinced it is global warming and not some natural cycle in hurricane intensity. Chris Landsea is quoted in the CNN piece sounding skeptical.

Chris C Mooney has already blogged about this story, even posting the abstract from the Nature paper [LINK]. He mentions the CNN story, but praises it. After actually seeing the abstract, I'd say the CNN story is talking up the controversy. Yes, Emanuel is linking the change to global warming, but he is also clearly mentioning the natural variability.

The study is actually a fairly straight forward exercise. Emanuel got some data from the US National Hurricane Center and the Japanese analog, as well as some SST data and reanalysis. He uses a more or less intuitive measure of the power dissipated by cyclones, and then simplifies it somewhat. That gives him an index of hurricane power through time, and he uses regions of high cyclogenesis to construct the time series. Once he's got the index, it is easy to just compare it with historic records of surface temperature and wind, etc. There are a few fishy details. For one, he smoothes everything with a 1-2-1 filter, and does it twice. He says that is to reduce the interannual variability, but it might blur things and make correlations higher. His statistics are based on linear regression coefficients, which is fine and good, but with the smoothing I'm a little concerned. I'd have liked to see him remove the interannual variability with a more sophisticated method. These are details though. The concerns raised by Landsea in the CNN article are not obvious, but might be available in the supplementary methods section on the Nature website, but I haven't had a chance to read it.

So my conclusion for now is to tentatively believe these results. The records are a little noisy, even with smoothing, but the trends are probably real and significant. The index used makes sense as a measure of hurricane destructiveness, and Emanuel is pretty careful about attributing the trend to anthropogenic causes. This should be a concern. Tropical cyclones are a very important part of the climate because they tranport so much energy out of the tropics, and between the ocean and atmosphere. If we are screwing with them, there are definite possibilities for feedbacks that might amplify climate change in ways that haven't been considered very well.

No comments: