2007-2011 hurricane season predictions reviewed
June 2, 2012
Before releasing a post summarizing the predictions for activity in the hurricane season to come, I like to review those of seasons past.
I do this to address a couple of concerns people have about these forecasts. One is that when the media reports upon them, they rarely, if ever review the performance of the forecasts in years past. The second is that the forecasts always (intentionally) over-state activity and as such exist solely to generate hype and anxiety for the forthcoming season. One lesser concern, even more conspiratorial in nature, is that the classification of tropical storms, etc is manipulated so that NOAA can “hit their numbers”.
As for second concern, 2005 remains the canonical refutation of that notion. The highest forecast for that season from the organizations tracked here was Colorado State’s, which called for 15 named storms, 8 hurricanes, and 4 major hurricanes. The actual numbers for the season were 28,15 , and 7. When you review the five hurricane seasons since, it is clear that there is not a sustained inclination to overstate activity.
A concern I won’t address in the post is the general utility of these forecasts to the average person. I expressed my thoughts in a 2007 commentary on hurricane season forecasts :”They are of little value to individuals. National Hurricane Center forecaster Richard Pasch said it best: ‘An active year is the year when you get hit.’”
With that out of the way, let us look back on the 2011 season and the pre-season predictions. The season featured 19 tropical storms, 7 hurricanes, 4 major hurricanes, and 127 units of Accumulated Cyclone Energy (ACE). The consensus numbers from the six organizations I track were 15, 8, 4, and 148.
Three of the forecasts originated from reasearchers affilitaed with universities (Colorado State, Florida State, and North Carolina State). NOAA’s forecast came from its Climate Prediction Center. Tropical Storm Risk is a British consortium, and the sixth forecast was produced by the United Kingdoms’s Meteorological Office.
2011 TS/H/MH *** |
2011 ACE | |
Observed | 19/7/4 | 127 |
5 season trailing average |
13/7/3 | 102 |
Consensus | 15/8/4 | 148 |
CSU | 16/9/5 | 160 |
FSU | 17/9/x | 163 |
NCSU | 15/8/4 | x |
NOAA | 15/8/5 | 140 |
TSR | 14/8/4 | 124 |
UKMET | 13/x/x | 151 |
Everyone under-forecast the number of tropical storms. Short-lived tropical storms such as Franklin, Gert, and Jose, brought the number of named storms above predicted levels without causing the other numbers to go over as well. All forecasts were over by one or two on the number of hurricanes, but were either on the number or were within one of the number of major hurricanes.
Most forecasts slightly to moderately overstated the amount of Accumulated Cyclone Energy. Tropical Storm Risk’s forecast was the exception as it nearly hit the number exactly.
As we’ll see below, the first two forecasts from Florida State were impressive, particularly the ACE predictions. 2011’s forecast, alas, wasn’t particularly note-worthy. Yes, it was the closest in predicting the number of tropical storms. However, that is the least important of the four numbers. ACE is the most representative number of aggregate tropical cyclone activity. The importance of the storm categories goes down with the damage potential, hence major hurricanes are most important followed by hurricanes, and then the number of tropical storms.
Here are the predictions for the past five seasons. Note that the “observed” numbers for 2007 removed Andrea and Olga due them being out of season (an adjustment I failed to make in past reviews and something that was brought to mind by Alberto and Beryl this year). Also, the UK Met Office has been releasing forecasts since 2007, but only since 2011 have they been June-November forecasts that are easily comparable to the others (past forecasts were July-November). For each year, I’ve link to NOAA’s season summaries, which provide good reviews of the atmospheric conditions that prevailed for the season and how they affected the NOAA forecasts.
2007 | 2008 | 2009 | 2010 | 2011 | |
Observed | 13/6/2 | 16/8/5 | 9/3/2 | 19/12/5 | 19/7/4 |
5 season trailing average |
17/8/4 | 17/8/4 | 17/9/5 | 16/7/4 | 13/7/3 |
Consensus | 15/9/4 | 14/8/4 | 11/6/2 | 17/10/5 | 15/8/4 |
CSU | 17/9/5 | 15/8/4 | 11/5/2 | 18/10/5 | 15/8/5 |
FSU | 8/4/x | 17/10/x | 17/9/x | ||
NCSU | 13/9/5 | 14/7/x | 13/7/x | 17/10/x | 15/8/4 |
NOAA | 15/9/4 | 14/8/4 | 12/6/2 | 19/11/5 | 15/8/5 |
TSR | 16/9/4 | 14/8/3 | 11/5/2 | 18/10/4 | 14/8/4 |
UKMET | 13/x/x |
In the past five years, the consensus storm numbers have been overforecasts twice (2007 and 2009) and in the ball park the other three seasons. Note that 2009 was an El Niño year. When the oscillation and its affects are not properly accounted for, over-forecasts result.
Observed | 2007 | 2008 | 2009 | 2010 | 2011 |
ACE | 72 | 145 | 51 | 166 | 127 |
5 season trailing avg | 158 | 160 | 154 | 119 | 102 |
Con. | 158 | 139 | 76 | 177 | 148 |
CSU | 170 | 150 | 85 | 185 | 160 |
FSU | 65 | 156 | 163 | ||
NOAA | 147 | 136 | 85 | 186 | 140 |
TSR | 156 | 131 | 69 | 182 | 124 |
UKMET | 151 |
2007’s ACE was brutally over-forecast by everyone. With the exeption of the major hurricanes Dean and Felix, storms that season did not have long lives and therefore did not contribute much to the total. In percentage terms, the consensus forecast for 2009 was substantially overblown as well. 2011 wasn’t as bad. The consensus predictions for 2008 and 2010 were excellent
My next post will be a round-up of the forecasts for the 2012 season.
***(Note that in their forecasts as published, some organizations release ranges of numbers, or in the case of Tropical Storm Risk, numbers that aren’t whole numbers. The numbers here are averages and roundings of ranges and TSR’s numbers are rounded as well. The consensus is calculated using the unrounded numbers. That’s why the consensus numbers of major hurricanes is four, when by appearances, it should be five; TSR’s published number was 3.7 and NOAA’s range was 4-9. Those got rounded up to four and five, respectively for the purposes of the list, but the unrounded numbers were used to calculate the average of the forecasts)