PONTE VEDRA BEACH, Fla. – The wait for Hurricane Irma’s turn north was nail biting.
Models showed Irma turning along Florida’s east coast followed by a pendulum swing to the Gulf.
Recommended Videos
Even as close as 48 hours before landfall great uncertainty on Irma’s exact landfall caused millions of vulnerable Floridians to flee in the Nation's largest evacuation in history.
The track resulted from a game of tug of war. The battle was between a northward pulling mid-latitude trough over the United States and the westward push of the Bermuda High.
Forecast simulations were off the mark except for for one, the new experimental model called the HWRF-Basin got the sharp turn and continuous track through the peninsula right.
Models initially showed Irma moving up the east coast however a jet stream dip south of Canada never weakened the Bermuda ridge and its influence pushed Irma westward toward the Gulf coast.
Results from NOAA’s Atlantic Oceanographic and Meteorological Laboratory suggest the HWRF- Basin (B) was superior to others due to the way it handles multiple individual storms. Instead of looking at all storms in a static snapshot over the Atlantic, the HWRF-B targets multiple movable hi-resolution grids on each system in a method called nesting.
It is proving to be an ideal way to predict hurricanes by using an area that covers ¼ of the globe with several sets of movable nests for multiple storms in the tropics.
The models does have a weakness forecasting systems outside the tropics in the mid-latitude.
What worked well with Irma was the HWRF-B’s method of “seeing” and replicating the hurricanes initial condition and the interaction with Hurricane's Katia and Jose which altered the general weather flow across the Atlantic basin.
If Katia never formed, Irma would of turned earlier and more slowly taking a path closer to Hurricane Matthew. The opposite is shown without Jose's interaction steering Irma faster with a later turn.
While the HWRF-B won gold for Irma it is part of a broader switch in the forecast community to ensemble prediction systems. These choose the best average without focusing on a single track and provides a greater sense of uncertainty.
The HWRF-B is still in experimental study and may eventually be transitioned into the GFS upgrade planned over the next several years in the FV3GFS.