NOAA not ruling out a repeat of the 2005 hurricane season (storm numbers-wise)
May 27, 2010
Today, NOAA released their May 2010 Atlantic Hurricane Season Outlook. It calls for an 85% chance of an above normal, 10% of near-normal, and 5% chance for a below normal season. They assign a 70% probability that the possibilities that the number of named storms will fall in the 14-23 range, hurricanes will be 8-14, and the number of major hurricanes will be 3-7.
In this post, I will try to demonstrate why the spreads are so large (and not necessarily blatant butt-covering) and in the course of doing so, show how the specter of the record-setting 2005 season looms large. I will not be opining on its accuracy or utility.
The first part of the forecast as I’ve described it is pretty straight-forward. By saying that there’s an 85% chance of an above average season, they are stating an 85% likelihood that at leas two of the three things will occur: 11 or more tropical storms, 7 or more hurricanes, and 2 or more major hurricanes. The formulation of this part of the forecast is unchanged since 2001.
The second part is a bit more involved and is new as of 2008. Rather than make an outright prediction on a specific number of storms (like the Colorado State does) or a range of numbers (as the forecast did prior to 2007), they give a range of numbers that represents 70% of the possibilities offered by the expected climatological conditions.
As one can note from my previous post this did not significantly affect the range of forecast numbers. The spreads for each category were generally 4 for named storms 3 for hurricanes and 2 for major hurricanes compared to this year’s spreads of 9, 4, and 4, respectively. There appear to be two reasons for this. One is a policy change, the other reflective of the nature of the season that the forecasters anticipate.
Dr. Master’s post on the NOAA forecast included a link to a paper presented at this year’s American Meteorological Society’s hurricane conference. In assessing the accuracy of NOAA’s hurricane season forecasts, the authors note that the forecasts offered by NOAA in May and August tend to be similar to each other. If there’s any value whatsoever in the climate forecasts over the course of the hurricane season, this should not be the case***. The May forecast should reflect greater uncertainty (by having a larger spread of numbers) than the one released in August. The paper states “Future NOAA outlooks will probably have larger May ranges to account for higher uncertainties at that time to meet the 70% target verification rate.” and that’s certainly the case in this outlook. It is not clear, however, that this idea was in fact implemented deliberately to acheive the larger spread than past forecasts. It is possible that the extreme nature of the forecast forced the spreads to be larger.
To borrow the language of The Black Swan, Mediocristan enjoys below average to average hurricane seasons, while Extremistan is ravaged by active seasons. That is, the range of activity in the quietest of seasons is much narrower than the range in the most active.
If I were to come out and predict with 100% certainty that this season would be one of the ten quietest out of the past fifty, I would be prognosticating the following: 0-7 tropical storms, 0-4 hurricanes, and 0-1 major hurricanes. Conversely, if I were to guarantee a ten most active season, I would be forecasting a season in the following ranges: 14-28+ tropical storms, 8-15+ hurricanes, and 4-7+ major hurricanes. Note, that if the 2005 season were thrown out, the ranges for tropical storms and hurricanes change to 14-19 and 8-12 respectively. That’s how off the charts 2005 was. Given that the ranges forecast by NOAA exceed those and have a larger spread, it’s clear that they consider a 2005-like season in the realm of possibilities.
Of course, it would be possible to have the numbers of 2005 combined with the (relative) lack of catastrophe of the otherwise note-worthy 1995 season. After all, “It is not a seasonal hurricane landfall forecast, and it does not imply levels of activity for any particular region.” That is why one could have no doubt to its accuracy but still question its utility to the average person.