Jeff co-founded the Weather Underground in 1995 while working on his Ph.D. He flew with the NOAA Hurricane Hunters from 1986-1990.
By: Dr. Jeff Masters , 1:55 PM GMT on August 27, 2009
Tropical Storm Danny continues to look disorganized this morning. The low level circulation center is exposed to view, with the heaviest thunderstorms lying several hundred miles east of the center. The center has undergone several relocations over the past 12 hours, and may do so yet again this morning, in order to position itself nearer to the heaviest thunderstorm activity. Despite its disorganization, Danny continues to generate strong winds, with the Hurricane Hunters and QuikSCAT both reporting winds in the 55 - 60 mph range early this morning. There is plenty of dry air in Danny's vicinity interfering with development, thanks to an upper-level trough of low pressure.
Figure 1. Morning satellite image of Danny showing the exposed swirl of clouds where Danny's center is, well displaced from the heaviest thunderstorm activity to the east.
The intensity forecast for Danny
The upper-level low that has been keeping Danny disorganized has weakened and separated from the storm, leaving Danny in a region with moderate wind shear of 10 - 15 knots and a modest amount of dry air. These environmental conditions will remain roughly constant through Friday night. Slow to moderate strengthening of Danny to a Category 1 hurricane should result, and is called for by all of the reliable intensity models. By Friday night, a trough of low pressure will approach the U.S. East Coast and bring high wind shear of 20 - 35 knots through Saturday. Danny will be close enough to this trough that the trough may be able to feed energy to Danny as the trough converts Danny to an extratropical storm. As a result, Danny may not weaken as fast as one might ordinarily expect, given the high levels of wind shear expected on Saturday. A landfall in Cape Cod, eastern Maine, or Nova Scotia with 55 - 75 mph winds is a good bet.
Figure 2. Performance of the main models used to forecast Hurricane Bill. Forecasts for the time periods 24, 48, 72, 96, and 120 hours are shown, with the track errors for each models' forecasts in nautical miles (nm). The statistics are shown for the regular interpolated version of the models used by the NHC forecasters in real time to make their forecasts. The "Consensus" model is the NHC's TVCN consensus, which is the average of at least two of the other models shown here (but not including the Canadian model). The Canadian model had the best performance of any model for Bill, surpassing even the Official NHC forecast. The next best performing models were the GFDL and GFS. Last year's best performing model, the European Center model, was not available for this analysis. Image credit: Dr. Jim Goerss, Navy Research Lab, Monterey.
The track forecast for Danny
Wunderground provides a computer models plot showing the hurricane track forecasts of most of the major models used by NHC to formulate their official forecast (one notable exception: we can't show the European Center ECMWF model, since this model is not freely available). One of the most frequently asked questions I get is, "which model do you trust?" This morning we have several models like the NOGAPS and Canadian calling for Danny to pass very near Cape Hatteras, North Carolina, then over Cape Cod, Massachusetts. The rest of the models foresee Danny missing Cape Hatteras, and continuing on to an encounter with Nova Scotia instead of Cape Cod. According the 2008 NHC forecast verification report, the best performing model during the 2008 hurricane season was the European Model (ECMWF), by a wide margin. The ECMWF out-performed the official NHC forecast, and it is very rare for an individual model to do this. The next best models were the GFDL and HWRF, while the NOGAPS, UKMET, and GFDN did the worst of the major models. The Canadian model was not analyzed, but historically has been among the worst of the models for forecasting hurricanes.
So far this year, the ECMWF has also done well. Unfortunately, the European Center group does not make the output of their hurricane tracking module publicly available, so I cannot present any statistics of their model's performance. Somewhat surprisingly, the Canadian model has also done very well this year. The model received a major upgrade in its physics of the past year, and has performed extremely well in hurricane track forecasts for both the Eastern Pacific and the Atlantic so far this year. In fact, for Hurricane Bill, the Canadian model gave better track forecasts then the NHC did (Figure 2). Danny is a different storm than Bill, and it is possible that the Canadian model will do less well with a storm that is disorganized, like Danny is. Nevertheless, with the Canadian model consistently keeping a Danny's track close to Cape Hatteras and going over southeastern Massachusetts, residents of these areas need to be prepared for possible hurricane conditions from Danny. Given the recent reformation of Danny's center, and the possibility of yet another reformation later today, all of the track models must be viewed with more than the usual amount of doubt. Since the center reformations have been moving Danny's center to the north and east, it may be that the Canadian model's prediction lies too close to the U.S. coast.
For those of you wondering about specific probabilities of getting tropical storm force or hurricane force winds, consult the NHC Wind Probability Product. The 11 am EDT NHC forecast gave Cape Hatteras a 4% chance of seeing hurricane force winds from Danny, and Nantucket, MA, a 7% chance.
For more information on computer models used by NHC
Basics of hurricane forecast models (Dr. Jeff Masters, wunderground.com, updated 2007)
Description of computer models used for hurricane forecasts (NHC, updated 2009)
Description of computer models used for hurricane forecasts (NOAA/AOML)
Figure 3. Tropical wave 94L off the coast of Africa.
Invest 94L off the coast of Africa
A well-organized tropical wave lies a few hundred miles southwest of the Cape Verdes Islands, near the coast of Africa. This wave was designated 94L by NHC this morning. Shear is low, about 10 knots, and waters are warm enough to support development. The dry Saharan Air Layer is relatively limited in extent and intensity, so dry air may have only a small inhibiting effect on the wave. Expect some slow development of this wave as it moves westward over the next few days. NHC is giving this system a low (less than 30% chance) of developing into a tropical depression by Saturday. This system is moving rather slowly, 10 - 15 mph, and it will be at least a week before it approaches the Lesser Antilles Islands. The GFS model develops 94L into a tropical depression early next week.
Special note on using the Canadian model
While the Canadian model has been doing well with hurricane track forecasts this year, the model still does a poor job forecasting the genesis of new tropical cyclones. The Canadian model has a false alarm rate perhaps three times higher than any other model, so one should not believe the Canadian model's regular predictions of new tropical cyclones springing up. You can access output from the Canadian model at Environment Canada or at Florida State University or Penn State.
I'll have an update later today.
Comments will take a few seconds to appear.
Light Snow Mist