It has been an exceptional year in terms of how two of the world’s best global models have performed. I am speaking of the American generated GFS or Global Forecast System versus the European generated/supported ECMWF or European Centre for Medium-Range Weather Forecasting. Both models are highly sophisticated and provide a vast array of products for forecasters around the globe. However, both have had their share of issues over the years when it comes to forecasting tropical cyclone events.
Let’s take a look at TS Debby back in late June. The GFS was forecasting a turn to the northeast as a trough of low pressure dug in to the eastern portion of the U.S. On the other hand, the ECWMF was forecasting a ridge to build and push Debby back to the west. Here is a paragraph from the NHC’s discussion regarding Debby on June 24:
“THE TRACK FORECAST IS EVEN MORE COMPLEX. THE GFS INSISTS ON A TRACK TOWARD THE NORTHEAST AS DEBBY BECOMES EMBEDDED WITHIN A LARGE MID-LATITUDE TROUGH. HOWEVER…THE ECMWF AND THE HWRF BUILD A RIDGE TO THE NORTH OF DEBBY AND FORECAST A WESTWARD TRACK. GIVEN THE WESTWARD TURN INHERITED FROM THE PREVIOUS FORECAST…AS WELL AS THE HISTORICAL STRONG RECORD OF THE ECMWF…THE NEW OFFICIAL FORECAST MOVES DEBBY INITIALLY A LITTLE BIT TO THE NORTHEAST TO REFLECT CURRENT TRENDS BUT THEN TURNS THE CYCLONE BACK TOWARD THE WEST OR WEST-NORTHWEST IN 24 TO 36 HOURS. A MAJORITY OF THE GFS ENSEMBLE MEMBERS NOW ARE CONSISTENT WITH THE DETERMINISTIC RUN…WHICH WAS NOT THE CASE YESTERDAY…MAKING A STRONGER CASE FOR THE EASTWARD SOLUTION. WE MUST BE READY TO MAKE A CHANGE OF THE FORECAST TRACK AT ANY TIME.”
Note the reference to the ECMWF and how it has a “historical strong record”. Well, as it turned out, the ECWMF was wrong and Debby ended up turning north in to the Florida panhandle.
Forecast for Debby showing west track towards Louisiana
Later that day, the new forecast, which was the correct forecast, shows Debby headed north in to Florida panhandle
Now let’s fast-forward two months to Isaac in late August. Here again we had two significant differences between the GFS and the ECMWF. The GFS wanted to take Isaac north and in to a trough after entering the Gulf of Mexico, putting a substantial threat to the Florida panhandle and even the west coast.
Meanwhile, the ECMWF showed a much farther west track towards Louisiana. Here again is a quote from the NHC discussion on Isaac from August 22:
“HOWEVER…THE ECMWF CONTINUES TO SHOW A STRONGER RIDGE…AND THUS SHOWS A MORE WESTERLY MOTION THAN THE OTHER MODELS.”
As it turned out, the ECMWF was correct in a stronger ridge and Isaac ultimately made landfall in southeast Louisiana.
So what does this have to do with the current situation with Sandy? For the first few days of the forecast period, not much. Right now, the GFS and ECMWF are pretty close to each other since the pattern is fairly straight-forward right now. Sandy is forecast to move slowly northward and bend east of north and pass over or very close to Jamaica. Both models “agree” on this.
Then, both models have Sandy passing over eastern Cuba and in to the central Bahamas by around 72 hours. It’s after this time frame that the two models part ways and have two vastly different outcomes.
The GFS basically moves Sandy slowly to the northeast and out of the Bahamas. While the ECMWF takes Sandy more north by day four but also fairly slowly.
At day five, the GFS has Sandy roughly half-way between Florida and Bermuda while the ECMWF is quite a bit back to west, near 30N and 75W.
It seems that the GFS wants to hand off Sandy in to the Atlantic due to lower heights in the middle layers of the atmosphere because of a large ocean storm, partly the remnants of Rafael. This provides and escape route for Sandy to take, like a rock gradually rolling down a slope, not too steep, but just enough to build momentum and begin rolling. Sandy is the rock heading down the gentle slope.
The ECMWF does not allow Sandy to feel the slope, so to speak, and keeps it farther back to the west, much closer to the Southeast U.S. coast. Then, all “you know what” breaks loose with that model.
By day six, a significant deep trough of low pressure digs in to the west of Sandy and causes the flow to turn more from the south out over the western Atlantic. Instead of pushing Sandy out like a broom sweeping across the floor, the trough instead scoops Sandy up and swings it back to the northwest and eventually in to New England as a powerful hybrid storm. Basically, the trough “captures” Sandy and pulls it in, not allowing it to turn east and out to sea like the GFS shows. The result is a remarkable storm for the mid-Atlantic and New England; the so-called “storm for the ages” that I alluded to yesterday.
Which solution will be right? I have no idea. I can see why both models show what they show but I also know that the GFS tends to have a bias in handing off tropical cyclones in to lower height fields more often than not. In other words, the GFS is taking the easy way out and sending Sandy in to the Atlantic. The ECMWF does not do this and instead keeps Sandy in the western Atlantic long enough for the big trough to dig in and capture it, creating a large ocean storm of epic size and strength.
Day seven of ECMWF showing powerful storm as Sandy gets drawn in to trough in the East
There is a lot of talk about this situation in the weather blogosphere. Opinions about both scenarios abound and some folks simply refuse to believe such a wild scenario with the ECMWF. However, considering its track record in the long term, it has to be considered and thus, once Sandy is in the Bahamas, the story will begin to unfold as to what ultimately ends up being the true course that it will take.
In the end, one of these two powerhouse global models will end up being correct. Which one that is will have enormous implications on the weather for millions of people along the East Coast of the U.S. towards the very end of the month. Stay tuned, this has only just begun….