Improved Computer Models Help Better Predict Hurricanes
May 31, 2013
By The Palm Beach Post
In August 2000, virtually every computer model tracking Hurricane Debby was calling for it to intensify. Managers already had begun evacuating the Florida Keys.
Then, somewhere over Hispaniola, Debby vanished, a mere two hours after posted forecast tracks had it striking either South Florida or the Bahamas.
Two months later, Keith strengthened in a mere 12 hours from a minimal storm to one almost as strong as Category 5 Andrew as it neared Belize. More than a dozen people died in Central America and Mexico.
While the science of forecasting where a storm will go has made dramatic leaps, "we haven`t made much progress over the decades" in predicting changes in a storm`s strength, said James Franklin, the National Hurricane Center`s chief of forecast operations.
That finally may be changing. Computer models have become more adept at using data from the Doppler weather radar mounted on planes that fly into tropical systems, Franklin said. That means forecast models could show improvement in measuring a storm`s intensity as early as this hurricane season, which starts June 1st.
The Doppler data - which now dates to 2010 and soon to 2008 - will let forecasters base their predictions off facts instead of what amounts to a phony storm, Franklin said.
"If you cannot accurately depict the arrangement of wind," he said, "you are almost hopeless in trying to predict that structure in the future."
In most cases, forecasters can`t provide substantial data on a tropical system`s birth or early life. So in the past, Franklin said, they`ve had to generate " `a standard vortex.` A bogus, if you will, is put in there.
"It may bear very little resemblance to the structure. It`s no wonder these models have struggled, because they`re starting with bad information."
The P3-Orion research planes that fly into storms have been armed with Doppler radar for years, Franklin said. But the computers and forecast models only recently have gained enough power and sophistication to digest and use the Doppler data.
"We haven`t had the computer power to ingest the volume of observations," Franklin said. "Now we have a snapshot and a model that can actually make use of that snapshot."
Forecasters feed that Doppler data into computer models run by the National Weather Service and the National Oceanic and Atmospheric Administration`s Miami-based Hurricane Research Division, as well as another operated by Penn State University.
Franklin called the results to date "intriguing" but added they pose a question: "Is it the Doppler data that`s giving us improved intensity forecasts, or can you do just as well with the flight data we already have?"
If the Doppler proves not to be crucial, it`s hard to justify the millions that would have to be spent to mount Doppler units on every research plane, especially in a time when researchers are fighting for every federal dollar, Franklin said.
While Franklin doesn`t expect substantial results this year, "we`re on the verge of seeing some significant improvements over the next few years." he said.
As a rule, forecasters bump up their strength forecasts 4 percent. Emergency managers go one step further, planning and evacuating for an entire category higher on the Saffir-Simpson scale, which ranks hurricanes in categories from 1 to 5.
"If the storm jumps up a category at the last minute, we want to make sure we`ve planned for that," Palm Beach County Emergency Manager Bill Johnson said. "Because of the clearance times that are required in order to evacuate the numbers of people that we have to, we need to be judicious in our decision-making."
The hurricane season runs from June 1 through Nov. 30.
Copyright Palm Beach Post (FL) 2013
Story image: FILE - In this May 7, 2013 file photo, National Hurricane Center Director Rick Knabb talks in Fort Lauderdale, Fla. AP Photo/J Pat Carter, File
What do you think of this story?
for comments or suggestions.