The Intersection

Throughout the spring and into summer, I’ve pointed out repeatedly that there has been a consensus, among seasonal hurricane forecasters, that there will be an active season this year in the Atlantic. 17 named storms was Bill Gray’s number; NOAA said 13-17. Methodologically, both Gray and NOAA use a “statistical” forecasting technique: In other words, they get these numbers based upon correlations between various climatological factors (and particularly El Nino) and the number of storms that appeared in past seasons.

But there is another, newer type of forecasting: Dynamical seasonal prediction, which uses multiple climate model runs to determine the number of tropical storms that will appear (hurricane-like storms in the models are taken as a guide to real life storm activity even though they are only rough analogues at best). This type of work has been pioneered in Europe. And last year, the dynamical models clearly beat the statistical models (which were all predicting an active season). Instead, El Nino showed up and more or less shut down Atlantic hurricane activity after September.

i-a754075f43e69a1e578e1fd50db2eb10-MetOffice.gif Last summer at around this time, I visited the UK Met Office and the European Centre for Medium Range Weather Forecasts and talked to scientists there–but they were not willing to make their dynamical forecast for 2006 public. This year, however, it has just been released: They are expecting a less active season, 7-13 storms between July and November. Apparently this is not based upon the notion that El Nino will come back again, but rather, on the prediction that Atlantic sea surface temperatures will cool down some.

And so, you might say that the bets are down. There’s a clear gap between the two types of forecasts–statistical and dynamical–going into the busier part of this season. Only time will tell which is right….

P.S.: As I mentioned earlier, in a new paper also using a statistical technique, Thomas A. Sabbatelli and Michael E. Mann of Penn State University predicted 15±4 total Atlantic storms for 2007. This is in press at the Journal of Geophysical Research – Atmospheres. The Sabbatelli/Mann methodology is also based in significant part upon SSTs and the state of the El Nino-Southern Oscillation…so I have to wonder, why are the European models apparently picking up an SST cooling that no one else seemed to be expecting?

P.P.S.: Jeff Masters also blogs the new forecast and has a much more thorough discussion of the dynamical seasonal hurricane forecasting technique. Storm World also contains a bit on this. Masters thinks the wisest outlook on the now disparate forecasts is to take a consensus of them all, which would leave us expecting “13 more named storms this year, for a total of 15″……


  1. #1 Bill
    June 20, 2007

    Meanwhile, today’s Miami Herald has an interesting story about supposed suppression of NOAA….

  2. #2 Sheril Kirshenbaum
    June 20, 2007

    The Met Office graphic brings to mind a briefing I attended last year on hurricanes as a National Security issue in the Northern Atlantic. Projections for NYC are devastating and struck a chord since that’s my hometown. If the Hudson and East Rivers do indeed overflow, they would need to evacuate 3 million people – more than six times the population of pre-Katrina New Orleans. Flooding of the subways, a large portion of Long Island, and beyond would result in over $100 billion in economic losses. MSNBC had a story on this last week.

  3. #3 Chris Mooney
    June 20, 2007

    It gets potentially worse under global warming scenarios for sea level rise (forget any other changes to hurricanes). According to NASA GISS:

    “Adding as little as 1.5 feet of sea level rise by the 2050s to the surge for a category 3 hurricane on a worst-case track would cause extensive flooding in many parts of the city. Areas potentially under water include the Rockaways, Coney Island, much of southern Brooklyn and Queens, portions of Long Island City, Astoria, Flushing Meadows-Corona Park, Queens, lower Manhattan, and eastern Staten Island from Great Kills Harbor north to the Verrazano Bridge.”

  4. #4 llewelly
    June 20, 2007

    As an interesting coincidence(?), there’s been recent warming off the coast of Ecuador – the opposite of what would be expected if conditions were continuing to trend toward the forecast La Nina.

  5. #5 John Fleck
    June 20, 2007

    Chris –

    Have they done “hindcasting”, running the models based on past pre-season data to see how well they perform relative to the pre-season forecasts done by the statistical forecasters?

  6. #6 Chris Mooney
    June 21, 2007

    Hi John,
    I haven’t read the latest paper as Jeff Masters obviously has, but yes, it seems clear they have done hindcasting. You should contact the authors of the work to get more details.

New comments have been temporarily disabled. Please check back soon.