[an error occurred while processing this directive]

Eyes on the storm

Advanced radar and other technologies are helping forecasters anticipate storms so warnings can be sent out faster.

By DAVE GUSSOW Times Technology Editor

© St. Petersburg Times, published June 7, 1999

Dennis McCarthy knew it was coming. Severe weather. Dangerous weather. Tornado weather.

The storm system evolved slowly. It didn't hit on the weekend as first thought. It struck on Monday, May 3, big, fierce and deadly. More than 60 tornadoes ripped through Oklahoma and Kansas. More than 40 people died.

"It didn't surprise us that we had a really big outbreak," said McCarthy, meteorologist in charge at the National Weather Service's forecast office in Norman, Okla. But, he said, "not too many would have said at that point that we would have a disaster along the lines that we had."

Technology helped McCarthy and his team of forecasters know that conditions existed for tornadoes to occur. Technology helped the forecasters give more warning to people in the path of the storm, to whom every minute was critical. But, when it comes to weather forecasting, nature rules, not technology, even though the hardware and software keep improving and getting more sophisticated.

"You cannot predict that a particular storm is going to take a particular path," McCarthy said. "You cannot predict it will produce an F4 (category of tornado). We aren't capable of doing that at this time. But we are very capable of monitoring the progress of the storms as they develop and evolve and monitoring the ones that start to rotate (to give) enough lead time that this is going to produce a tornado."

The deadly tornadoes in May caught everyone's attention, but the focus in Florida now turns to the hurricane season. Much of the same hardware and software used in the Midwest will be used to track hurricanes.

The National Hurricane Center tries to predict three things most important to get the public's attention about storms, said director Jerry Jarrell: a hurricane's path, how strong it will get and how big it will get, so people can evacuate in time.

"We're getting pretty good" at tracking a path, said Jarrell, but not as well at forecasting a storm's intensity and size.

Technology has helped reduce tracking errors by 30 percent in the past six years or so, Jarrell said, because of improved computer models. The error measurement is based on the predicted paths of storms compared with actual paths. During the previous 20 years, Jarrell said, tracking may have improved by only 1 percent a year.

Intensity and size are tougher, and sending planes into storms doesn't always give the right kind of information.

"Sampling is much more complicated than anyone thought," Jarrell said. "It just looked to us like one could go out and collect information around the storm and we would improve the models. As it turns out, the models are sensitive."

Collecting data from only a small area of the storm doesn't give a complete picture of what is occurring in a system that can cover hundreds of miles, Jarrell said. And that can throw off the models.

When it comes to predicting how strong a storm will get, "we're not doing any better than we were 20, maybe 30 years ago," Jarrell said. "Some would say we're doing worse," even with more information available.

While the hurricane center can track storms for days or weeks, tracking is harder for the meteorologists who follow tornadoes. "Their problem is 100 times tougher than ours," Jarrell said, because they sometimes have only minutes to get measurements and issue alerts.

* * *

The National Weather Service doesn't type up the text for weather warning advisories any more. Sure, a person runs the computer, but the computer does the work, sending out the warning automatically. It gets the information to the public faster and allows meteorologists to keep an eye on the storm.

The main problem with the system? The robot-like automated warnings on the radio. "A lot of people don't like the voice, but we're early in that game," McCarthy said. And it is only one of the new high-tech touches that forecasters have at their disposal.

Meteorologists rely on information from a variety of sources, such as radar, satellites and ground observation stations, to prepare their forecasts.

The National Oceanic and Atmospheric Administration, which oversees the National Weather Service and other agencies, has been modernizing its technology during most of the 1990s to improve analysis of the environment and storm detection and to get warnings to the public faster, McCarthy said.

Among the upgraded and new systems:

* The Advanced Weather Interactive Processing System, or AWIPS, takes data from many sources and helps forecasters analyze it. As part of a network connecting 113 Weather Service offices, it also is critical in getting severe weather warnings to the public faster by automating the warning advisory process.

* The Next Generation Radar, or NEXRAD (or WSR-88D in Weather Service lingo), uses sophisticated Doppler technology, which improves detection of circulation patterns that could turn into tornadoes or other severe weather. The computer-run system produces 3-D images that examine cloud formations higher in the atmosphere and help forecast how much rain a storm may produce.

The Doppler data has "given us the ability to see characteristics of storms in different parts of the country that we did not have before," said Jim Bellville, director of the NEXRAD Radar Operational Support facility, also in Norman. For example, severe storms in New England don't have the same structure as ones in the Midwest. "We didn't know that existed prior to Doppler radar," Bellville said.

What distinguishes NEXRAD from other weather radar systems is the size of the beam and the amount of energy it sends out. An old system would send out a beam about 2 degrees wide at about 500,000 watts. NEXRAD uses a narrower beam, a degree or less, at about 750,000 watts, producing a more detailed image. Old radar showed only one level of a storm; NEXRAD can show an entire storm, even showing the growth of raindrops.

Forecasters are refining the formulas for computer models of everything from hail to tornadoes.

How does the new equipment translate for the public?

On May 3, the day the tornadoes struck in the Midwest, early morning advisories indicated a slight risk of tornadoes. By noon, the risk was upgraded to moderate; by 3 p.m., the risk of strong tornadoes was high. That was almost two hours before the first tornado developed.

"We felt the watches and the warning and the information flow were tremendous," said McCarthy, the meteorologist. "With all of that, we still had an unfortunate death toll. Some of these things are so strong and have such intensity and magnitude, it's very hard to do much about it."

* * *

Deep Thunder passed its first test at the 1996 Olympics in Atlanta.

The project, developed by IBM and NOAA, gave an early-morning forecast that thunderstorms would be in the area but that they would not affect the closing ceremonies. So organizers went ahead with their plans.

The forecast was right. How did it pinpoint where storms would or wouldn't be?

Deep Thunder uses an RS/6000 SP supercomputer, from the same family as the machine that beat chess champion Garry Kasparov, and advanced 3-D models. It uses the same information gathered by the National Weather Service, with a major difference: It focuses on much smaller geographic areas than the Weather Service, and in more detail.

If the costly system were available in the Tampa Bay area, Deep Thunder researcher Lloyd Treinish said, it would be possible to say whether a thunderstorm will go over Tampa Bay, downtown Tampa or St. Petersburg, even before it has formed. Narrowing the forecast to individual neighborhoods might be possible.

By contrast, the Weather Service produces forecasts that are regional and national in scope, then adapted for local forecasts. It divides the country into thousands of grids covering about 18 miles each. Even with supercomputers, it takes time to process the data, and weather changes can slow the process further.

To get those focused, pinpoint forecasts than one can see on TV or elsewhere requires more computing power. And Deep Thunder, which IBM is preparing for commercial uses, is expensive -- ranging from $500,000 to $1-million for each supercomputer setup. The first full deployment is in China, with a system that covers more than 2-million square miles. It also is being used in some places by the National Weather Service and the Air Force weather agency.

But it has limitations: It doesn't do long-range forecasts. It can warn of severe weather, but not that tornadoes will occur, according to IBM's chief meteorologist, Zephiris Christidis.

IBM sees numerous commercial possibilities for the system. Businesses whose fundamental decisions can be affected by weather -- insurance, aviation, agribusiness, utilities -- and local and state governments are potential customers.

"These are all commercial types of applications," Christidis said, "but it's clearly being driven by science. Science is enabling more intelligent business decisionmaking."

Deep Thunder isn't IBM's only venture into weather-related projects.

Last year in Orlando, it rolled out a prototype of TeamBuilder. It allows governments, utilities and others that need to form emergency teams to get in touch with key personnel quickly.

One of its first tests was during Hurricane Georges last year. When the city activated the program, it automatically sent out phone, pager and e-mail messages to people on emergency teams. The system gives each worker his or her duties, can send any necessary paperwork electronically and can keep a log of activity.

Lt. Roger Huder of the Orlando Fire Department said the system allowed him to do tasks that had been handled by three people, as well as help coordinate efforts across departments and agencies. "To me, it's not something you would use just during hurricanes," Huder said.

For example, Houston has a center for police, fire and public works agencies that can monitor and control daily problems around a city. It can intervene in traffic, blocking interstate on-ramps if there is an accident and controlling traffic lights. Orlando is considering a similar system.

"Little things happen daily. To be able to control some of these things is very important," Huder said. "The technology gives us tools to do it in a reasonable way."

IBM researcher Mike Greenwood, who came up with the TeamBuilder idea when he saw utility workers near his New York home struggling during a snowstorm, saw even more possibilities for the system as he watched Orlando use it.

"I see a direct correlation between a system for building and collaboration of teams to save lives as very similar to winning in business," Greenwood said. "The more efficient you are at teaming, the quicker you can respond to bids, the quicker you can restore outages, the more you can please the customer."

* * *

Not everyone pays attention to weather warnings.

While Deep Thunder allowed the Olympics to go on as scheduled, one person died during the thunderstorm miles away.

"What we were doing in Atlanta had never been done before operationally," Christidis said. "They didn't know what to expect."

IBM's Treinish said: "If they had believed that information, if they had used it to the degree they could have used it, they might have been able to save a life."

Jarrell at the Hurricane Center worries that residents of the Keys will have different reactions to a storm this year after last year's bout with Georges.

Those who left ahead of the storm will remember the hard time they had getting back to their homes and may not evacuate if another storm threatens, Jarrell said. Those who stayed during Georges may be the ones to evacuate next time.

One expert predicts a heavier-than-normal hurricane season, but to Jarrell that doesn't make a difference for how people should prepare or react.

"You can be hit just as easily on a light season as a heavy season," said Jarrell, who is retiring this year. Dangerous weather requires "constant vigilance," as last year's disasters with Georges and Mitch showed.

"I looked very carefully at what we did," Jarrell said. "And I was amazed at how well we had done. The track and intensity were not great, but the weather (forecast) was."

Back to Tech Times
Click for TampaBay.com, your entertainment section and more

Action | Arts | Business | Citrus | Columnists | Floridian
Opinion | Entertainment | Floridian | Hernando | Pasco | Sports
State | Tampa Bay | Travel | World & Nation | Taste

Back to Top
© Copyright 2006 St. Petersburg Times. All rights reserved.