Hurricane Dorian Will Spare Florida: Why So Much Uncertainty with This Storm?

It is becoming increasingly clear that Hurricane Dorian is going to make a sharp right turn as it approaches the eastern coast of Florida, resulting in only modest impacts to the Sunshine state.  Hurricane force winds (sustained at more than 64 knots) and heavy rains will probably remain offshore.

Early forecasts had predicted landfall of a category 5 hurricane on the central Florida coast,  with potentially devastating impacts, but now the real risk has shifted northward into the Carolinas.

Some critics have already started complaining, suggesting that there were too many cancellations and closures before the storm track was clear.  Folks along the Florida coast have stripped the food stores, gas stations are out of fuel, and even the Orlando Airport was closed.

 Hurricane Dorian at 2:40 PM PDT on 31 August

Was there a failure of weather prediction technology for this event?  Did the European Center Forecast outdo the American FV-3?    Did local governments over-react?

My conclusion:  this has been an inherently difficult storm to forecast, with more uncertainty than many recent storms, some which were predicted well over a week in advance (e.g., Irma).

The latest forecasts of all major centers are converging on a solution in which the storm approaches the Florida coast, but swings northward just offshore.  Let me show you the usual cream of the crop model, the European Center forecast.

The forecast for 5 AM Monday PDT, shows the storm approaching the Bahamas.

 A day later, it moved to the northwest...quite slowly.

 5 AM Wednesday it is crawling northward.

And Thursday morning it is just off the southeast tip of North Carolina

The strongest winds of a hurricane are in the right front quadrant--to the right of the direction of motion.   And thus the strongest winds are generally kept offshore.  Same thing with precipitation.

The accumulated precipitation totals through Friday morning show the heaviest precipitation offshore.  No problem at all for Florida and Georgia.   An issue for coastal North Carolina.

As I shall describe in a second, this is an inherently difficult forecast situation with high levels of uncertainty, particularly once the storm gets near Florida.  A measure of the uncertainty is apparent in the ensembles--in which multiple forecasts starts slightly differently and are run by both the European Center and the NWS.

To illustrate, here is the latest ensemble prediction of the position of the hurricane center during the next five days..  No much variation (spread) initially as the storm heads west. And then nearly all members make an abrupt right turn....but there is a lot of spread in the exact time when the turns start, resulting in a big difference in the low center positions in five days.  The U.S. ensemble is similar, but closer to the coast.

Why is there so much uncertainly?  It has to do with steering currents.  Hurricanes tracks--or trajectories in time---are controlled by the large scale flow around them....which is called a steering current.

For strong storms like Dorian, the steering flow can be represented by the flow speed and direction averaged from around 5000 ft to 30,000 ft (or roughly 850 hPa to 300 hPa in pressure).  If the flow becomes weak or if there is a large change in steering flow in distance, hurricanes can stagnate or move erratically.  

The current (2 PM Saturday)steering flow analysis by NOAA CIMMS is consistent with westward movement of the storm (the arrows show the flow direction and the hurricane is indicated by the red marker).

But this changes during the next few days, with the flow weakening over the region and the subtropical high moving eastward (steering flow prediction by HWRF model for Monday at 5 PM PDT shown).  The result is a northerly component to the steering flow and the shift to the right.

But let's be clear...this is all on the edge, and it would not take a large prediction error to allow the storm to drift further to the west...and thus onshore.   

The weak steering flows make this inherently a difficult forecast, even for the best of the current models.   A verification of position errors by Professor Brian Tang of U. of Albany suggests the U.S. FV-3 model (dark blue color) and been similar in skill to the European Model (purple color) for 12-96 hr forecasts, with the Euro being significantly better at 120 hr.  The position errors are quite large compared to recent hurricanes. But these are early times.