2012 Atlantic Hurricane Season Prediction Roundup

Having reviewed the past performance of these forecasts in my previous post, it is now time to look at the forecasts for the current hurricane season.

The official start of the season comes with two named storms, Alberto and Beryl already having come and gone.  As most forecasts are explicitly June 1 – November 30, those storms do not count against these forecasts as listed here (I’ve adjusted the number of  named storms forecast by Colorado State  downward by two, per their note) .
Is  this pre-season activity  significant when considering activity for the rest of the season?  The Colorado State forecast notes

Pre-1 June activity has very little bearing on the rest of the hurricane season. The only two seasons on record with two named storms prior to 1 June were 1887 and 1908. While 1887 was a very active season, 1908 had average levels of activity. The last season with a U.S. landfall prior to 1 June was 1976, which was a relatively quiet season.

Also note that by this time in the 2007 season, there already were two names in the books.  While that season finished with a respectable number of named storms (15), aggregate activity as measured by ACE (72 units)  was below the long term average.

Sorry Beryl, you don’t count

Here is a summary of the forecasts from the six organizations I track. It lists the number of named storms, hurricanes, and major hurricanes forecast along with the predicted amount of Accumulated Cyclone Energy.  The consensus forecast (calculated as a simple average of unrounded numbers from all forecasts) is for twelve named storms, six hurricanes, and 97 units of ACE. This forecast is slightly below the average of the past five seasons and roughly in line with the 1981-2010 average.

	TS/H/MH	ACE
Observed
5 season trailing average	15/7/4	111
Consensus	12/6/2	97
CSU	13/5/2	80
FSU	13/7/x	122
NCSU	9/6/2	x
NOAA	12/6/2	95
TSR	13/6/3	98
UKMET	13/x/x	90

I’ll know briefly review each forecast in alphabetical order. Note that a few forecasts contain much more information than that’s what displayed here. I’ll mention it as necessary so that those interested may delve into the forecasts on their own; I won’t be covering everything in detail, however.

The grandfather of these seasonal forecasts is that issued by researchers affiliated with Colorado State University; Dr. William Gray issued his first forecast in 1984 and since 2005 Phillip Klotzbach has been listed as the primary author with Dr. Gray second.  The forecast is statistically driven, though, for the current scheme, one input comes from a computer model.  The methodology has been alternatively tweaked and replaced in whole over the years; the current scheme is new to 2012.

The Colorado State 2012 hurricane season forecast calls for 11 named storms to form after June 1 with 5 becoming hurricanes and 2 of those being major hurricanes and a combined 80 units of ACE.  The forecast discusses mixed signals as to whether a full blown El Niño event will develop or neutral conditions will prevail along with near-normal to cooler than normal Sea Surface Temperatures  (SSTs) in the Atlantic as being factors in their forecast. Regarding the latter issue, the forecast states “Overall, the Atlantic is experiencing more marginal conditions this year than in many previous years during the active era that has been experienced since 1995.” It is worth noting that their forecast for Accumulated Cyclone Energy would be the lowest forecast since 1994 *** and is the lowest of the six forecasts tracked here.

The forecast also contains landfall forecast probabilities along with a lengthy discussion of whether rising CO2 levels in the atmosphere have contributed to the increase in major hurrricanes in the Atlantic since 1995.

Since 2009, a team at Florida State University lead by Tim LaRow has been issuing forecasts. The forecasts are generated using computer model simulations that were originally described in the paper Atlantic Basin Hurricane Simulations.  Due to apparent limitations in model resolution, the model does not generate hurricanes that are at or above the major hurricane threshold, hence the lack of a forecast number of major hurricanes.

Not Tim LaRow

The FSU 2012  hurricane season forecast calls for 13 tropical storms and 7 hurricanes.  The forecast for ACE is the highest: 122 units.  It is worth remembering that FSU’s ACE forecast was also the highest and therefore ended up being the most wrong. This year, however, the FSU forecast is alone at the high end; last year the CSU forecast was nearly identical to that of FSU.

Since 2005, professor Lian Xie of North Carolina State University has been the lead for releasing forecasts from his institution. His statistical approach garnered significant attention in 2006 when it correctly forecast a quiet season when others suggested more activity.

The NC State 2012 hurricane season forecast  gives a range of numbers, which  average to 9  named storms, 6 hurricanes, and 2 major hurricanes.  The forecast also predicts storm numbers for the Gulf of Mexico and gives some landfall probabilities, though it is not nearly as extensive as the predictions from Colorado State.

Since 1999, the Climate Prediction Center of NOAA has been issuing outlooks, in collaboration with the National Hurricane Center and the Hurricane Research Division.  The basic factors for the forecast are fairly unchanged;  the state of El Niño and SSTs in the Atlantic are the two primary considerations.  A variety of computer models are utilized to predict those two factors.

NOAA’s 2012 Hurricane Season Outlook gives a range of numbers, which average to 12 named storms, 6 hurricanes, and 2 major hurricanes.  Like the CSU forecast, it discusses the spread in El Niño forecasts as a source of uncertainty. It also reviews the near-normal to below average SSTs in the Atlantic.

The Tropical Storm Risk consortium was borne out of a UK government sponsored project named TSUNAMI. The first forecast was released under the latter name for the 1999 hurricane season with Professor Mark Saunders as the lead author.   It utilizes a statistical approach considering two factors: Forecast SSTs in the Mean Development Region of the Atlantic Ocean (10°-20°N — 20°-60°W, the classic breeding ground for Atlantic tropical cyclones) and forecast low level trade winds over the Atlantic and Carribean.

Tropical Storm Risk’s 2012 Atlantic hurricane forecast predicts 13 named storms, 6 hurricanes, and three major hurricanes. Of the two values used in the forecast, the one for trade winds suggest a slight effect of suppressing tropical cyclone formation, while the SST values are neutral.  The forecast also makes predictions for the number of landfalling storms in the United States and Caribbean. TSR had an strong 2011 forecast (with a near perfect prediction of ACE) , though, as their own forecast notes, “forecasts for the 2006, 2007, and 2009 hurricane  seasons were less impressive”.

The United Kingdom’s Meteorological Office has been releasing outlooks since 2007, though it was only last year that they started issuing them as June-November forecasts .  The forecast is derived from two global forecast models, one of their own along with one from the European Center for Medium range Weather Forecasts.

The UK Met Office hurricane season forecast for 2012 predicts 10 named storms and 90 units of accumulated cyclone energy.

These are the forecasts that I keep track of either due to their long reputable existence or, in the case of the younger ones, notable performances. (I’ll admit I was absolutely parochial in keeping track of FSU’s model right from the start; fortunately it was a strong performer in its first two seasons).  While it may seem that these are more than enough forecasts, there are yet more out there.

Everyone has a forecast

Penn State University researcher Micheal Mann has been releasing forecasts since 2007. His 2012 prediction (which comes from a statistical model) is 11 named storms.

Weatherbell Analytics is a private company that hired forecaster Joe Bastardi after he left Accuweather.  Weatherbell Analytics’ 2012 hurricane season forecast is for 11 named storms, 5 hurricanes, 3 major hurricanes and 85 units of ACE. It expects fewer storms to develop far at sea and relatively more to from close in to the United States.

In the context of hurricanes, the Geophysical Fluid Dynamics Laboratory (GFDL) is best known for its specialized dynamic hurricane model.  The lab has developed what it is calling a Hybrid Hurricane Forecast System for hurricane season outlooks. Unfortunately, the most recent discussion is for the January forecast run. That forecast predicted six hurricanes for the 2012 season.

Finally, the Cuban Institute of Meteorology (ISMET) has been producing forecasts since 1996. I have been unable to find an primary source document for the forecast, but a column penned by the Institute’s director, Dr. Jose Rubiera, quotes from it.  Their forecast calls for ten named storms with five becoming hurricanes.

The column, originally published in Granma, the official newspaper of the Communist Party of Cuba, is actually pretty interesting. It is entitled “El Pronostico de la Temporada de Huracanes en Su Propio Lugar“, which translates to “The Hurricane Season Forecast in its Proper Place”.  It discusses the purpose of hurricane season forecasts and their limitations.  Rubiera outright states that the forecasts have scientific value,” but do not possess a practical value to the general public”.

I find the column valuable enough to spend time working on a translation ****, which will be the content of my next post.

*** It’s only been since 2006 that the CSU forecast explicity forecasted ACE; before then it gave only a number called Net Tropcial Cyclone (NTC) activity, which is a completely different calculation, but like ACE, does give a measure of aggregate activity.  This season’s forecast NTC, 90, is the same as the 2009 forecast,  but the forecast for ACE that season was 85.   You have to go back to 1994, to find a NTC lower than 90. Had there been an ACE forecast, it likely would have been at 80 or lower.

**** Yes I could just run it through Google Translate or some such tool and it would probably be workable or “good enough”. Feel free to give it a whirl, yourself.  I did though, get an outstanding education in the Spanish language in high school and have made use of that learning at various points since, but not recently. So I’m going to grab this surprising opportunity and run with it.