WunderBlog Archive » Category 6™

Category 6 has moved! See the latest from Dr. Jeff Masters and Bob Henson here.

NOAA, TSR, UKMET, PSU, WSI, and WU Community Predict Active Atlantic Hurricane Season

By: Dr. Jeff Masters, 9:11 PM GMT on May 24, 2013

NOAA forecasts an above-normal and possibly very active Atlantic hurricane season in 2013, in their May 23 outlook. They give a 70% chance of an above-normal season, a 25% chance of an near-normal season, and 5% chance of a below-normal season. They predict a 70% chance that there will be 13 - 20 named storms, 7 - 11 hurricanes, and 3 - 6 major hurricanes, with an Accumulated Cyclone Energy (ACE) 120% - 205% of the median. If we take the midpoint of these numbers, NOAA is calling for 16.5 named storms, 9 hurricanes, 4.5 major hurricanes, and an ACE index 162% of normal. This is well above the 1981 - 2010 average of 12 named storms, 6 hurricanes, and 3 major hurricanes. Hurricane seasons during the active hurricane period 1995 - 2012 have averaged 15 named storms, 8 hurricanes, and 4 major hurricanes, with an ACE index 151% of the median. Only five seasons since the active hurricane period that began in 1995 have not been above normal--including four El Niño years (1997, 2002, 2006, and 2009), and the neutral 2007 season.

Figure 1. Hurricane Michael as seen by NASA's Aqua satellite at 12:20 pm EDT Thursday September 6, 2012. At the time, Michael was a major Category 3 hurricane with 115 mph winds. Hurricane Sandy was the only other major Atlantic hurricane of 2012. Image credit: NASA.

The forecasters cited the following main factors that will influence the coming season:

1) Above-average sea surface temperatures (SSTs) are expected in the hurricane Main Development Region (MDR), from the Caribbean to the coast of Africa between between 10°N and 20°N. SSTs in the MDR during April were 0.4°C above average, and were 0.33°C above the oceans in the remainder of the global tropics. Long-range seasonal computer model forecasts predict a continuation of above-average SSTs in the MDR during much of hurricane season.

2) We are in an active period of hurricane activity that began in 1995, thanks to a natural decades-long cycle in hurricane activity called the Atlantic Multi-decadal Oscillation (AMO).

3) No El Niño event is expected this year. El Niño events tend to suppress Atlantic hurricane activity. Neutral conditions have been present since last summer, and are predicted to remain neutral through hurricane season by most of the El Niño computer forecast models.

NOAA said, "This combination of climate factors historically produces above-normal Atlantic hurricane seasons. The 2013 hurricane season could see activity comparable to some of the very active seasons since 1995." NOAA is increasingly using output from ultra-long range runs of the computer forecast models we rely on to make day-to-day weather forecasts, for their seasonal hurricane forecasts. These models include the NOAA Climate Forecast System (CFS), NOAA Geophysical Fluid Dynamics Lab (GFDL) model CM2.1, the European Centre for Medium Range Weather Forecasting (ECMWF) model, the United Kingdom Meteorology (UKMET) office model, and the EUROpean Seasonal to Inter-annual Prediction (EUROSIP) ensemble.

Figure 2. Graphic from the 2013 NOAA Atlantic hurricane season forecast highlighting the reasons for this year's anticipated active character.

How accurate are NOAA's seasonal hurricane forecasts?
A talk presented by NHC's Eric Blake at the 2010 29th Annual AMS Conference on Hurricanes and Tropical Meteorology studied the accuracy of NOAA's late May seasonal Atlantic hurricane forecasts, using the mid-point of the range given for the number of named storms, hurricanes, intense hurricanes, and ACE index. Over the past twelve years, a forecast made using climatology was in error, on average, by 3.6 named storms, 2.5 hurricanes, and 1.7 intense hurricanes. NOAA's May forecast was not significantly better than climatology for these quantities, with average errors of 3.5 named storms, 2.3 hurricanes, and 1.4 intense hurricanes. Only NOAA's May ACE forecast was significantly better than climatology, averaging 58 ACE units off, compared to the 74 for climatology. Using another way to measure skill, the Mean Squared Error, May NOAA forecasts for named storms, hurricanes, and intense hurricanes had a skill of between 5% and 21% over a climatology forecast. Not surprisingly, NOAA's August forecasts were much better than the May forecasts, and did significantly better than a climatology forecast.

Figure 3. Forecast skill of the TSR, NOAA (National Oceanic and Atmospheric Administration) and CSU (Colorado State University) for the number of hurricanes in the Atlantic during 2003-2012, as a function of lead time. Forecast precision is assessed using the Mean Square Skill Score (MSSS) which is the percentage improvement in mean square error over a climatology forecast (six hurricanes in a given year.) Positive skill indicates that the model performs better than climatology, while a negative skill indicates that it performs worse than climatology. Two different climatologies are used: a fixed 50-year (1950-1999) climatology, and a running prior 10-year climate norm. NOAA does not release seasonal outlooks before late May, and CSU stopped providing quantitative extended-range December hurricane outlooks in 2011. Skill climbs as the hurricane season approaches, with modest skill levels by early June, and good skill levels by early August. Image credit: Tropical Storm Risk, Inc (TSR).

TSR predicts an active hurricane season: 15.3 named storms
The May 24 forecast for the 2013 Atlantic hurricane season made by British private forecasting firm Tropical Storm Risk, Inc. (TSR) calls for an active season with 15.3 named storms, 7.5 hurricanes, 3.4 intense hurricanes, and an Accumulated Cyclone Energy (ACE) of 130. The long-term averages for the past 63 years are 11 named storms, 6 hurricanes, 3 intense hurricanes, and an ACE of 103. TSR rates their skill level as modest for these late May forecasts--11% - 25% higher than a "no-skill" forecast made using climatology. TSR predicts a 63% chance that U.S. land falling activity will be above average, a 21% chance it will be near average, and a 16% chance it will be below average. They project that 4.4 named storms will hit the U.S., with 2 of these being hurricanes. The averages from the 1950-2012 climatology are 3.1 named storms and 1.4 hurricanes. They rate their skill at making these late May forecasts for U.S. landfalls just 8% - 12% higher than a "no-skill" forecast made using climatology. In the Lesser Antilles Islands of the Caribbean, TSR projects 1.5 named storms, 0.6 of these being hurricanes. Climatology is 1.1 named storms and 0.5 hurricanes.

TSR’s two predictors for their statistical model are the forecast July - September trade wind speed over the Caribbean and tropical North Atlantic, and the forecast August - September 2013 sea surface temperatures in the tropical North Atlantic. Their model is calling for warmer than average SSTs and slower than average trade winds during these periods, and both of these factors should act to increase hurricane and tropical storm activity.

UKMET office predicts a slightly above normal Atlantic hurricane season: 14 named storms
The UKMET office forecast for the 2013 Atlantic hurricane season, issued May 13, calls for slightly above normal activity, with 14 named storms, 9 hurricanes, and an ACE index of 130. In contrast to the statistical models relied upon by CSU, TSR, and NOAA, the UKMET forecast is done strictly using two dynamical global seasonal prediction systems: the Met Office GloSea5 system and ECMWF system 4. In 2012, the Met Office forecast was for 10 tropical storms and an ACE index of 90. The actual numbers were 19 named storms and an ACE index of 123.

WSI predicts an active hurricane season: 16 named storms
The April 8 forecast from the private weather firm WSI (part of The Weather Company, along with The Weather Channel, Weather Central, and The Weather Underground), is calling for an active season with 16 named storms, 9 hurricanes, and 5 intense hurricanes.

Penn State predicts an active hurricane season: 16 named storms
The May 11 forecast made using a statistical model by Penn State's Michael Mann and alumnus Michael Kozar is calling for an active Atlantic hurricane season with 16 named storms, plus or minus 4 storms. Their prediction was made using statistics of how past hurricane seasons have behaved in response to sea surface temperatures (SSTs), the El Niño/La Niña oscillation, the North Atlantic Oscillation (NAO), and other factors. The statistic model assumes that in 2013 the May 0.87°C above average temperatures in the MDR will persist throughout hurricane season, the El Niño phase will be neutral, and the North Atlantic Oscillation (NAO) will be near average.

The PSU team has been making Atlantic hurricane season forecasts since 2007, and these predictions have done pretty well, except for in 2012, when an expected El Niño did not materialize:

2007 prediction: 15 named storms, Actual: 15
2009 prediction: 12.5, named storms, Actual: 9
2010 prediction: 23 named storms, Actual: 19
2011 prediction: 16 named storms, Actual: 19
2012 prediction: 10.5 named storms, Actual: 19

The wunderground community predicts an active hurricane season: 17 named storms
Over 100 members of the wunderground community have submitted their seasonal hurricane forecasts, which are compiled on trHUrrIXC5MMX's blog. The April 28 version of this list called for an average of 17 named storms, 8 hurricanes, and 4 intense hurricanes in the Atlantic. This list will be updated by June 3, so get your forecasts in by then! As usual, I am abstaining from making a hurricane season forecast. I figure there's no sense making a forecast that will be wrong nearly half the time; I prefer to stick to higher-probability forecasts.

NOAA predicts a below-average Eastern Pacific hurricane season: 13.5 named storms
NOAA's pre-season prediction for the Eastern Pacific hurricane season, issued on May 23, calls for a below-average season, with 11 - 16 named storms, 5 - 8 hurricanes, 1 - 4 major hurricanes, and an ACE index 60% - 105% of the median. The mid-point of these ranges gives us a forecast for 13.5 named storms, 6.5 hurricanes, and 2.5 major hurricanes, with an ACE index 82% of average. The 1981 - 2010 averages for the Eastern Pacific hurricane season are 15 named storms, 8 hurricanes, and 4 major hurricanes. So far in 2013, there has already been one named storm. On average, the 2nd storm of the year doesn't form until June 25.

NOAA predicts a below-average Central Pacific hurricane season: 2 tropical cyclones
NOAA's pre-season prediction for the Central Pacific hurricane season, issued on May 22, calls for a below-average season, with 1 - 3 tropical cyclones. An average season has 4 - 5 tropical cyclones, which include tropical depressions, tropical storms, and hurricanes. Hawaii is the primary land area affected by Central Pacific tropical cyclones.

The week ahead: 91E, and a heavy rainfall threat to Mexico
We're already behind last year's pace for named storms in both the Atlantic (where Tropical Storm Alberto formed on May 19, and Tropical Storm Beryl on May 26), and in the Eastern Pacific, where Bud formed on May 21 (the earliest date since record keeping began in 1949 for formation of the season's second named storm.) The Madden Julian Oscillation (MJO), a pattern of increased thunderstorm activity near the Equator that moves around the globe in 30 - 60 days, is currently located in the Eastern Pacific. The MJO is relatively weak, but is helping boost the chances that Invest 91E in the Eastern Pacific will develop. On Friday, NHC was giving 91E a 20% of developing into a tropical cyclone by Sunday. The 12Z Friday runs of the GFS and ECMWF models were predicting that a weak circulation off the coast of Costa Rica, well east of the separate circulation currently called 91E, could develop into a tropical depression by Tuesday. This system is a threat to spread heavy rains to the coast of Mexico from Acapulco to Guatemala on Tuesday and Wednesday.

In the Atlantic, the models are depicting high wind shear through June 1 over the majority of the regions we typically see May tropical cyclone development--the Caribbean, Gulf of Mexico, and Bahamas. The GFS model is showing a decrease in wind shear over the Western Caribbean after June 1, which would argue for an increased chance of tropical storm development then (though wind shear forecasts more than 7 days in advance are highly unreliable.) The prospects for an early June named storm in the Atlantic are probably above average, though, given that the MJO may be active in the Atlantic during the first week of June.

Have a great holiday weekend, everyone!

Jeff Masters


The views of the author are his/her own and do not necessarily represent the position of The Weather Company or its parent, IBM.