Cyclone Silicon

On April 25, 1994, G.W. Burnett was sitting in the den of his ranch-style home in DeSoto, Texas, watching a Texas Rangers baseball game on TV, when a news bulletin flashed on the screen. A tornado had been sighted about 15 miles northwest of DeSoto. Although he lives in what Texans refer to as "Tornado Alley," Burnett, a 46-year-old fire department captain, wasn't worried. "We have these thunderstorms all the time down here, and there's always tornado warnings," he said.

But the local weatherman broke into the broadcast to show the path of the tornado, and Burnett realized it was headed his way. DeSoto's emergency sirens started blaring. Burnett told his wife and daughter: "Get your purse, get the cat, and get in the closet." Burnett heard a loud roar and glass breaking. His ears started popping. The house shook as debris slapped against the walls. "I didn't know if the house was going to stay intact," Burnett recalled.

Moments after the tornado passed by, Burnett learned just how lucky his family had been. The front windows of his house had been blown out, and the curtains had been sucked out onto the roof, which was stripped of its shingles. Burnett's pickup truck, originally parked in front of his house, had been tossed like a toy across the street and two doors down. About 50 of the 200 homes in the Creek Tree development where Burnett lives suffered major damage. But no one was seriously hurt. "Everyone [who] was watching TV had plenty of warning," Burnett said.

Folks in neighboring Lancaster saw the worst of the tornado, which was classified as an F4 on the 0-to-5 Fujita scale and had sustained winds of 240 mph. The National Weather Service had issued a tornado warning at 8:45 p.m. Lancaster's emergency sirens were activated shortly after 9 p.m., and the tornado touched down 18 minutes later. Margie Waldrop, who was the mayor of Lancaster in 1994, also heard the warnings on TV. The tornado missed Waldrop's house, but it tore through 800 others in the city of 23,000, staying on the ground for 15 minutes. The roof of city hall was sheared off, two schools were damaged, and nearly 400 structures were demolished. Despite the wreckage, only three people were killed, and about 20 people were seriously injured.

A few months after the tornado, Waldrop presented NWS' Fort Worth office with a plaque that reads: "We believe the horrors of that evening were decreased by the detection and early warning of the storm." The proclamation acknowledged "the advanced technology so instrumental in sparing lives, which might otherwise have been lost."

The technology that helped save Burnett, Waldrop and thousands of others in DeSoto and Lancaster is a computer system called Next-Generation Weather Radar. Nexrad, as it is known, was installed in the Fort Worth forecasting office just three months before the tornado hit Lancaster. Skip Ely, the meteorologist in charge of the Fort Worth office, said the Nexrad system allowed his team to spot the storm earlier and track its progress in much greater detail than would have been possible with conventional radar.

The $1.5 billion Nexrad system is touted not only as a lifesaver but as nothing short of a revolution in the way that severe weather is forecast. "It is the most accurate radar and the best method of pinpointing developing severe thunderstorms," said David Finfrock, a meteorologist for NBC affiliate KXAS-TV who was on the air broadcasting warnings and information about the tornado that April night. "[Nexrad] will actually see into the storm," he said.

What "Only God Could See"

The Nexrad system is officially named WSR-88D, which stands for Weather Service Radar 1988, the year it was put into service, and D for Doppler. More than 150 Nexrad units operate nationwide, and the individual systems consist of three main parts. The signature part of the system is a 30-foot-in-diameter antenna protected by a white fiberglass-and-foam sphere mounted on top of a 10-story tower. Nearby is a shelter that houses the radar transmitter, receiver, signal processor, and communications and maintenance facilities that process the information that the radar receives. The second element is a computer that processes and archives the data that the antenna receives. This computer is linked to the third part of the system, the Principal User Processor, which is the workstation used by forecasters.

This workstation includes a PC, a keyboard, a printer and a terminal used to display color-coded weather data similar to the images the public sees when watching local weather forecasts on TV. The Nexrad system also features warning bells and signs to automatically generate alerts for forecasters when severe weather is detected. The individual Nexrad units are hooked up to the Operational Support Facility (OSF) in Norman, Okla.

Nexrad's technology is based on Doppler radar, named for the "Doppler effect." The natural phenomenon was first described around 1840 by Austrian physicist Christian Johann Doppler, who noted how the frequency of sound and light waves seems to increase as objects approach each other. A common illustration is how a train whistle sounds higher in pitch as a train approaches and lower in pitch as it passes and moves away.

Nexrad uses this phenomenon to measure the direction of wind and how fast it is moving. Nexrad sends out 3 billion microwave bursts each second and then measures how long it takes the bursts to be reflected to Nexrad's antenna by precipitation, dust or other objects in the atmosphere. Nexrad's computers crunch that data to provide images and other useful data to forecasters.

With Nexrad, NWS has significantly improved the amount of warning time it can give the public for severe storms. The average lead time for tornado warnings has gone from five minutes in 1986 to 10 minutes in 1996, and for all severe storms, it has increased from 12 minutes to almost 20 minutes— enough of an increase to save lives in some instances. In May 1995 the weather service was able to warn central Iowa residents that nine tornadoes were in the area. No one was killed, and only two residents were seriously injured.

A year later, the weather station at the Ellsworth Air Force Base outside Rapid City, S.D., issued a warning on a severe thunderstorm and potential flash floods. As a result, an ambulance carrying a critically ill patient was safely routed around hail and flooded areas to reach Rapid City Hospital.

James Wilson, a senior scientist at the National Center for Atmospheric Research in Colorado, said Nexrad has had "probably the single biggest impact on the National Weather Service of anything in the past 50 years."

Rex Reed, the chief of the Nexrad OSF's engineering branch, said, "This system can see stuff that only God could see before. It is amazing the leap that has been made with this radar."

Thirty Years in the Making

In 1964 the National Oceanic and Atmospheric Administration, which oversees NWS, set up the National Severe Storms Laboratory on an old Navy base in Norman. The mission of the center was to improve the methods for detecting severe storms.

The lab used surplus military Doppler radar units for its research, and in 1971 the first 10cm Doppler weather radar, a Nexrad prototype, was put online in Norman. (Most weather radars operate at wavelengths of either 5cm or 10cm.) Although the radar proved capable of identifying mesocyclones and discovering the vortex signatures that form inside tornadoes, the NSSL did not have the computer power to process the data in real time. Instead, the data had to be recorded on reels of eight-track magnetic tape, which NSSL workers would load into a car and drive to a NASA lab in Houston to analyze.

Not until 1973 would the NSSL have its own computers to analyze the Doppler data in real time. In April of that year, the equipment was used to display a tornado as it was forming in Ada, Okla., about 60 miles southeast of Norman. That image was recorded on a primitive black-and-white CRT, and for the first time meteorologists could see the circular rotation of the wind as it was moving. Everyone at the NSSL who was working on the development of the new radar system was convinced that it could dramatically improve warning times.

In 1976 the Joint Doppler Operational Project (JDOP) was established at the NSSL to evaluate how Doppler radar could be used to develop real-time forecasts and to lay the groundwork for establishing a national system of Doppler radar.

The project was a joint venture of NWS, the Defense Department and the Federal Aviation Administration. The agencies eventually became partners in developing Nexrad. In the JDOP tests, meteorologists detected the internal structures of thunderstorms, often as far away as 155 miles. They repeatedly saw mesocyclones forming and issued tornado warnings up to 30 minutes before the tornadoes touched down.

On the basis of the JDOP tests, the decision was made to move forward with the development of a national Doppler weather radar system, and the first funds for the project were included in the 1981 federal budget. In May 1983 two bidders, Raytheon Co. and Sperry Corp., were awarded contracts to develop pre-production models for what would eventually become Nexrad.

Stormy Weather for Nexrad

Although Nexrad's benefits to weather forecasting seemed clear, the system's birth would prove far more difficult than its conception. Sperry, which had become Unisys Corp., was awarded the contract for production of the Nexrad system in 1987. The first Unisys prototype for Nexrad was erected in Norman the following year.

Between 1988 and 1991, disagreements between Unisys and NWS over Nexrad's specifications and performance, as well as missed deadlines and cost overruns, nearly killed the project. Unisys filed a $200 million lawsuit against the government to recoup the costs of accommodating specification changes on the system. David Lewis, director of Unisys' weather information services department and technical director for the Nexrad project, said the company was frustrated by the frequent specification changes made by the weather service and feared Unisys would lose money on the program.

Elbert W. "Joe" Friday, who served as the head of NWS from 1988 to 1997, recalled that because of the disputes between the government and Unisys, "the decision was almost made to shut down" the Nexrad program in favor of buying off-the-shelf Doppler radar units that were less expensive but also far less powerful and versatile.

But Friday remained optimistic and continued to push for the system. "I never thought it was going to fall apart, although I was disappointed at times that things were going so badly," he recalled. "I thought it was absolutely imperative that we move forward. I didn't think there was any alternative."

In 1991 Friday and other Nexrad supporters received some luck that would build support for the system. A year before, 10 Nexrad systems had been installed nationwide, but the contracting dispute between Unisys and NWS had kept the systems from being turned on. But on March 20, 1991, Unisys and NWS agreed to turn on the Nexrad unit in Norman for a 30-day test. The next day, NWS spotted a tornado in Ada and was able to give residents there ample warning. About a month later, the system showed three tornadoes in a single line on the radar screen. The NWS forecasting office issued tornado warnings that had 15 to 20 minutes of lead time.

"It was the classic kind of storm, [but] with the conventional radar there would have been no warning, or it would have been very minimal," said Dennis McCarthy, the meteorologist in charge of NWS' Oklahoma Forecast Office. Forecasters used Nexrad to track the storm almost up to Kansas City, illustrating the unit's range as well as its high resolution even at great distances.

Based on Nexrad's performance, McCarthy's office was granted another 30 days to test the system, and then another. The spring of 1991 was an active storm season, and the Nexrad unit exceeded forecasters' expectations. On May 26 an intense thunderstorm and tornado hit near Woodward, Okla. "We had almost an hour of lead time to issue a warning," McCarthy said. "It was unprecedented."

From that point, Nexrad was turned on for good. That summer, the Norman forecast office had a continuous stream of visitors who wanted to see Nexrad in action, including members of Congress, their staff members and officials from the Commerce Department, the FAA and DOD.

Meanwhile, in Washington, D.C., the contract disputes between NWS and Unisys raged on. Thomas Giammo, a Commerce official who had acted as the government's chief negotiator in the dispute, told NWS and Unisys that it was in everyone's best interest to see that the Nexrad system was delivered. "I said, 'You have no choice; you can't kill this thing,' " Giammo said.

Unisys and NWS agreed to rewrite the final specifications for the system. In 1991 Giammo convinced Unisys to settle its $200 million suit against the government for about $40 million and then complete the contract. "That's probably the biggest thing I've ever done in my entire life," said Giammo, who retired from Commerce in 1993.

Despite problems along the way, Lewis said he is extremely proud of his work on the Nexrad system. "It's not too often you get to work on a system that touches so many lives," he said. "Not getting this system fielded would have been a tragedy."

For G.W. Burnett, who still lives in the same house along Tornado Alley, the tragedy of not listening to NWS' early warning is clear. "They made a believer of me," he said. "Now when they put out a warning, I pay attention."

-- Killian is the author of The Freshmen: What Happened to the Republican Revolution? and is the former editor of National Public Radio's "All Things Considered."

* * * * *

Nexrad Components

Hardware: Microflect tower, Westinghouse Corp. transmitter, Lockheed Martin Corp. receiver/signal processor, Concurrent Computer Corp. computers, Hewlett-Packard Co. workstations, Motorola Corp. modems, Ramtek Corp. graphics processors, Hitachi monitors, Tektronix printers.

Software: Concurrent Computer OS/32