14 October 2010

Congrats to NOAA, UKMET and Klotzbach/Gray!!

Regular readers of this blog know that the issue of prediction is a core focus of my research.  This year I have been following the prediction of an active hurricane season made by the various seasonal forecasting groups.

As of today, we can say that these predictions have demonstrated skill, meaning that they have outperformed a prediction based on a naive baseline of climatology (technically, the UKMET office has a few units of ACE to go, but I am assuming that will occur).  This can be seen in the graph above, graciously produced by Ryan Maue of Florida State.

Now there are, and should be, debates about the nature of the "naive baseline" -- for instance, given that the North Atlantic has been very active since 1995, is use of long-term climatology as a naive baseline for calculating skill setting the bar too low?  If we instead evaluated the seasonal forecasts based on a trailing 10-year climatology, I'd guess that they would not yet show skill.

In any case, methodological questions aside, congrats to the seasonal forecasters.  Their basin-wide forecasts of activity have proven accurate, and that is a notable accomplishment.  Of course, successfully predicting basin-wide activity offers little value in efforts to anticipate damage.  More on that soon.