2014 Hurricane Forecast - A Few, Maybe, Sometime

Time again for that annual guessing game "How many Hurricanes and Tropical Storms will there be, how many will hit the U.S. and, how much damages will they cause".  Fortunately we have NOAA on the job to be the contestant, and their answer to those questions for the 2014 hurricane season is:

NOAA’s 2014 Atlantic Hurricane Season Outlook indicates that a near-normal or below-normal hurricane season is likely this year. The outlook calls for a 50% chance of a below-normal season, a 40% chance of a near-normal season, and only a 10% chance of an above-normal season. See NOAA definitions of above-, near-, and below-normal seasons. The Atlantic hurricane region includes the North Atlantic Ocean, Caribbean Sea, and Gulf of Mexico.
. . .
Based on the current and expected conditions, combined with model forecasts, we estimate a 70% probability for each of the following ranges of activity during 2014:

• 8-13 Named Storms
• 3-6 Hurricanes
• 1-2 Major Hurricanes
• Accumulated Cyclone Energy (ACE) range of 40%-100% of the median.

. . .
These expected ranges are centered below the official NHC 1981-2010 seasonal averages of 12 named storms, 6 hurricanes, and 3 major hurricanes.

In short, a broad range, centered on values a little lower than average, due to an expected El Niño.  Not a terribly risky forecast, but still, we could have a terrible year, and then NOAA would have egg on their collective faces.

How has NOAA fared historically on this prediction.  Fortunately, someone else has waded through the data to look at that question: From 2011:

NOAA’s seasonal hurricane forecasts: Do they have any predictive value?

For each of the years, NOAA’s prediction range is about equal to one standard deviation, which implies a 1/3 chance of being right provided it is near the mean. Thus, if they were going by chance, then they should have had 13 (39/3) correct; instead, they were right 16 times. Unfortunately, because of the small data set, we can only say that they are better than chance at about the 80% level. In addition, they under-predicted 10 times and over-predicted 13 times, so there is a very slight tendency to over-predict. With more data, that trend would probably disappear.

Just a little better than rolling the dice.  It's also telling that they tend to overpredict.  Overpredicting is better for a forecasting institution in that people are more likely to be upset by an underprediction that goes horribly wrong than an overprediction that fails and results in a weak hurricane season.

I'd like to see a re-analysis of this with 2012 and 2013 data. 

But lets hope they're right and we have a weak hurricane season.