"2007 Atlantic hurricane season: below forecasts"

Excerpt from this Reference Frame post:

In 2007, the number of named storms, hurricanes, major hurricanes was 14, 6, 2, close to the average values of NOAA which are 11.0, 6.2, 2.7. Both major hurricanes managed to become category 5 hurricanes but the number of category 5 hurricanes is not among the quantities that are normally being forecast. All the three standard numbers 14, 6, 2 are higher than virtually all forecasts.

To see that in detail, let us see what various teams predicted at various moments (if no year is specified, we mean 2007):

* Reality, 11/30: 14, 6, 2
* CSU, 12/8/06: 14, 7, 3
* NOAA 5/22: 13-17, 7-10, 3-5
* CSU 5/31: 17, 9, 5
* UKMO 6/19: 9-15, N/A, N/A
* CSU 8/3: 15, 8, 4
* NOAA 8/9: 13-16, 7-9, 3-5
* CSU 9/4: 15, 7, 4
* CSU 10/2: 17, 7, 3

You see that every single prediction of the number of major hurricanes was overshot: instead of the consensus value around 4, we only saw 2. Every single prediction of the total number of hurricanes was overstated, too. Also, most predictions for the total number of named storms were overestimates.

I would like to claim that this science might be sophisticated and interesting but it is not yet sufficiently mature to make useful predictions. If a scientific team chose the 1951-2005 average to be their prediction, it would be shown more accurate than virtually every single team that tried to predict what would happen, both in 2006 and 2007.