A severe Atlantic hurricane season is on tap for 2007, according to the May 31
seasonal forecast issued by Dr. Bill Gray and Phil Klotzbach of Colorado State University (CSU) today. The Gray/Klotzbach team is calling for 17 named storms, 9 hurricanes, and 5 intense hurricanes--unchanged from their April forecast. An average season has 10-11 named storms, 6 hurricanes, and 2 intense hurricanes. The forecast calls for a much above normal chance of a major hurricane hitting the U.S., both along the East Coast (50% chance, 31% chance is normal) and the Gulf Coast (49% chance, 30% chance is average). The Caribbean is also forecast to have an above normal risk of a major hurricane.
The forecasters cite the expected lack of an El Ni�o event, the continuation of above average sea surface temperatures in the Atlantic, and slower trade winds (which result in reduced evaporative cooling of the ocean), as the justification for their forecast of a much above average hurricane season.
Figure 1. Top: Tropical Atlantic Ocean temperatures in the Main Development Region for hurricanes (green box) were 0.6 C above average during March and April 2007. This anomalous warmth is expected to persist though hurricane season. Bottom: The 0.6 C above average temperatures are consistent with the exceptionally warm temperatures seen since 2003. Image credit: NOAA.
How good are the CSU forecasts?The CSU forecast team has been making seasonal hurricane forecasts since 1984. If one grades their late May forecasts based on predictions of a below average, average, or above average season, they have done pretty well over the past eight seasons. Seven of their past eight forecasts have been correct. Their only failure occurred last year, when they called for a very active season, and it was a normal year with 10 named storms, 5 hurricanes, and 2 intense hurricanes. A more rigorous way of determining forecast skill is to compute the mathematical
correlation coefficient. A correlation coefficient of 1.0 is a perfect forecast, and 0.0 is a no-skill forecast. The
late May CSU forecasts have a respectable correlation coefficient of 0.57 for predicting the number of named storms (1984-2005). This decreases a bit to 0.46 and 0.42 for number of hurricanes and intense hurricanes, respectively. These are respectable correlation coefficients, and the late May CSU forecasts are worth paying attention to. This is in contrast to the December and April CSU forecasts, which have had a correlation coefficient near zero (and thus no skill).
Last year, the CSU team made their first steering current forecast. They predicted that a ridge of high pressure over the Eastern U.S. would steer more storms than average towards the Gulf Coast. However, the opposite happened--a trough of low pressure set up over the Eastern U.S.--and the 2006 steering current forecast was a bust. They've given up on trying to predict what this year's steering currents might be, citing the need to perform more research on this issue. In theory, such a forecast is possible. Gray and Klotzbach published a
2004 paper showing a statistical relationship between variations in Atlantic sea surface temperature and whether hurricanes are more likely to hit the U.S. Gulf Coast or East Coast. These SST variations influence the steering patterns, and help determine whether a persistent trough of low pressure will settle over the East Coast and recurve hurricanes out to sea--as happened in 2006--or whether a ridge of high pressure will settle in, pushing more storms towards the Gulf Coast--as happened in 2004 and 2005. The problem with all of these statistical Atlantic seasonal forecasts is that the atmosphere/ocean system is always changing in new ways that have not occurred in the past. Thus, a statistical scheme that works for forecasting past activity is much worse at predicting the coming year's activity. There is hope that the
global dynamical computer models used to forecast the weather will soon be able to surpass the statistical methods used by the CSU team. Indeed, recent papers have shown the the European model (ECMWF) and GFDL model both make seasonal hurricane forecasts that rival the CSU forecasts in skill. No word yet on when these new computer model seasonal forecasts will be available to the public though--more research is needed to develop them.
The bottom line: expect a very active Atlantic hurricane seasonThe CSU forecast matches up well with the TSR, Inc. forecast (16 named storms, 9 hurricanes, and 4 intense hurricanes) and the NOAA forecast (13-17 named storms, 7-10 hurricanes, and 3-5 major hurricanes). Also, keep in mind that the active hurricane period that began in 1995 has never seen two consecutive years with average or below-average hurricane activity. Given these factors, I am confident that the coming season will be a very active one. The two most recent years that had patterns of El Ni�o/La Ni�a events and SSTs similar to what are expected this year were
1995 and
2003. Note that 1995 was the third busiest hurricane season on record, with 19 named storms. However, the great majority of these storms recurved out to sea, since a trough of low pressure settled over the Eastern U.S. I have a similar hope for this season--if the steering currents are your friend, even a top-five hurricane season can have an ordinary number of landfalls. Let's hope the steering currents are our friend this year!
The Tropical Storm Risk, Inc. 2007 Atlantic hurricane season forecastThe British private forecasting firm
Tropical Storm Risk, Inc. (TSR), issues monthly 2007 Atlantic hurricane season forecasts. Their May 3 forecast has almost the same forecast as the CSU team--16 named storms, 9 hurricanes, and 4 intense hurricanes. I like how they put their skill level right next to their forecast numbers: 30% better than chance skill at forecasting the number of named storms, 34% skill for hurricanes, and 30% skill for intense hurricanes. TSR projects that five named storms will hit the U.S., with 2.3 of these being hurricanes. Their skill in predicting the number of named storms hitting the U.S. is only 8% above a no-skill forecast, but the skill rises to 30% for hurricanes hitting the U.S. In the Lesser Antilles Islands of the Caribbean, TSR projects 2.4 named storms, and 0.7 hurricanes. TSR cites two main factors for their forecast of an active season: above normal Sea Surface Temperatures (SSTs) are expected in August-September 2007 across the tropical Atlantic, as well as slower than normal trade winds. Trade winds are forecast to be 0.78 meters per second (about 1.6 mph) slower than average, which would create greater spin for developing storms, and allow the oceans to heat up due to reduced evaporative cooling. SSTs are forecast to be about 0.2 degrees C above normal. TSR gives an 84% chance that the 2007 Atlantic hurricane season will rank in the top third of active seasons observed since 1957. Their next forecast will be issued June 4.
The NOAA 2007 Atlantic hurricane season forecastNOAA is also predicting a very active 2007 hurricane season in the Atlantic
NOAA's season hurricane forecast issued May 22 predicts a very high (75% chance) of an above-normal hurricane season, a 20% chance of a near-normal season, and only a 5% chance of a below-normal season. They expect 13-17 named storms, 7-10 hurricanes, and 3-5 major hurricanes (a normal season has 10-11 named storm, 6 hurricanes, and 2 intense hurricanes). Most of these storms are expected during the usual August-October peak of hurricane season, but NOAA does not give any breakdown of which portions of the coast are more likely to be affected. They give two reasons for predicting an above-normal hurricane season:
1) A continuation of conditions since 1995 that have put us in an active hurricane period (in particular, the fact that seas surface temperatures in the Atlantic Main Development Region for hurricanes are currently about 0.6 C above normal, Figure 1).
2) The strong likelihood of either neutral or La Ni�a conditions in the tropical Pacific Ocean.
Tropical disturbance in the Western CaribbeanA large area of disturbed weather developed over the Western Caribbean last night. This disturbance is bringing winds of up to 55 mph over the ocean, according to the 7:07am EDT pass of the
QuikSCAT satellite. The
NOAA Buoy off the coast of the Yucatan Peninsula recorded winds this morning at 30mph, gusting to 35mph. There is no circulation evident on QuikSCAT or satellite loops, but the disturbance does have the potential to develop into a tropical depression by Saturday as it moves to the northeast over Western Cuba and South Florida. A Hurricane Hunter aircraft is scheduled to investigate the system at 2pm EDT on Friday.
Wind shear is a not-too-unfriendly 10-20 knots, and the shear is expected to remain at these levels through Saturday. Thereafter, most of the models are indicating that the disturbance will get caught up by a strong trough of low pressure with high shear that should stop further development and sweep the system northeastward out to sea. I doubt this system has enough time to get organized into a tropical depression before wind shear rips it up, but the disturbance should bring welcome heavy rains to South Florida over the weekend. Lake Okeechobee recorded its record lowest water level yesterday--8.97 feet (about 4 feet below normal). This was the lowest level since record keeping began in 1931, according to a
Miami Herald article this morning. The lake has been dropping about 1/2 inch per rainless day. I expect 1-3 inches of rain over the area this weekend, which should temporarily stabilize the lake water level.
NHC issued this statement at noon today:
Special tropical disturbance statement
1150 am EDT Thu May 31 2007
Showers and thunderstorms in the northwestern Caribbean Sea... southeastern Gulf of Mexico and adjacent land areas are associated with a broad area of low pressure centered about 75 miles southeast of Cozumel Mexico. Although this system has some potential for tropical development over the next day or so...the low is expected to move slowly northward into the southern Gulf of Mexico where environmental conditions would likely favor further development as a non-tropical low. Regardless of development...this system should bring heavy rains across western Cuba and southern Florida over the next couple of days. Please monitor products issued by your local Weather Service office for more details.
Figure 2. Visible satellite image of the Western Caribbean tropical disturbance.
June outlook and the Barometer Bob showI'll be posting my forecast for the first two weeks of June tomorrow (June 1). I plan to offer 2-week hurricane activity forecasts on the 1st and 16th of each month (except August 1, when I'll be on vacation). These forecasts will have the probability of hurricane formation for the coming two weeks, where the hurricanes will go if they form due to the prevailing steering currents, plus a look at how sea surface temperatures, wind shear, the trade winds, and dry air coming off of Africa are affecting hurricane formation in the Atlantic.
Tonight, I'll be a guest on the
Barometer Bob Show, if you want to hear a sneak preview of my outlook for the first two weeks of June and hear about the tropical disturbance in the Western Caribbean. You can listen at
barometerbobshow.com, or dial in via their toll-free number 1-866-931-8437 (1-866-WE1THER).
Jeff Masters