Why Forecasts Are Getting Cloudier

By Philip Elmer-DeWitt

The National Weather Service's new $3 million radar outpost in Norman, Okla., proved its worth on its first day of operation last March. That evening a series of thunderheads rolled across the southern Oklahoma hill country. One storm cell appeared -- at least on conventional radar -- to be relatively benign. But not to Nexrad (for Next Generation Radar), a new detection system that is powerful enough to track a swarm of insects moving across a wheatfield 50 km (30 miles) away. The domed instrument peered into the swirling winds and raindrops inside the clouds and saw a tornado aborning. The Weather Service flashed an alert to the surrounding community. Two houses and $1 million worth of property were destroyed that night by the cyclone, but there were no serious injuries. "You can never prove you've saved a life," says Ron Alberty, director of the Nexrad facility. "But I'm convinced many people's lives have been spared this spring in Oklahoma."

For weather forecasters, the radar station in Norman represents the bright edge of what is technically possible. It is the first of a proposed network of 160 stations that will eventually blanket the U.S. with high-power radar, vastly improving the accuracy of predictions. The network is part of an ambitious $2.25 billion modernization of the National Weather Service, almost a decade in the making, that also features a fleet of advanced satellites, a mosaic of automated weather stations and a high-speed information network linking them all together.

Unfortunately, the Norman outpost has also become a symbol of broken promises, missed deadlines and unfulfilled potential. Two more radar systems, one set for installation near Cape Kennedy in Florida and another outside Washington, are still sitting in packing crates, victims of a bitter contract dispute between the agency and the manufacturer, Unisys. Meanwhile, virtually every other part of the modernization program is either over budget, technically flawed or facing stiff opposition in Washington. The program could cost up to $1 billion more than originally estimated and is not likely to be completed until 1998, several years later than planned. In the meantime, the agency is forced to rely on outdated equipment that is deteriorating so rapidly it could leave large sections of the U.S. with no radar and satellite coverage at all.

At a Senate hearing last week, government officials admitted that they had "underestimated the complexity" of the overhaul and pleaded for restoration of millions of dollars that Congress might cut from the Weather Service's budget. Congress members have not only balked at the soaring cost of the program, but have also raised pork-barrel concerns about plans to reduce the number of NWS offices around the country from 249 to 115 -- a reduction made possible by the greater power of the new technology. "It's a minor version of the military-base closings," says one NWS official.

The Weather Service is in drastic need of renovation. The 100-year-old agency has become a technology museum. Its forecasters still launch old- fashioned balloons -- 70 of them twice a day -- to take readings in the atmosphere. They use refrigerator-size computers that have less power than the average desktop machine. And they depend on radar equipment that runs on World War II-type vacuum tubes. This creaking system is dangerously prone to breakdowns. In one notorious instance in the winter of 1988, the radar sentinel in North Carolina was out of service for 10 days, during which a batch of tornadoes tore up the state, injuring 157 people, killing four and wreaking $77 million worth of damage.

Even when the aged system is working, it has a blind spot for what meteorologists call "mesoscale" events, measured in minutes and tens of miles: tornadoes, flash floods, squall lines and thunderstorms. Some Weather Service offices do not issue a tornado warning until a human actually sights a twister -- by which time it is often too late to get out of harm's way. False alarms of flash floods have become so common that they are usually ignored.

Even under the best of circumstances, weather prediction is an inexact science. Because the upper atmosphere is subject to countless fluctuations, mathematicians say the theoretical limit for a reasonably accurate forecast is less than two weeks. But within this time frame, a number of innovations have enhanced the meteorologist's prophetic powers. Supercomputers build mathematical models that show the interaction of wind, sun, temperature and humidity across the entire globe. And Doppler radar -- the technology at the heart of the Norman station -- is adept at spotting the destructive midsize squalls that have traditionally taken forecasters by surprise. By bouncing microwaves off the tiny droplets in the center of a cloud and picking up the echoes, Doppler systems can map the relative velocity of wind currents within the cloud. High-velocity winds and a high level of organization can signal the formation of a mesocyclone -- a precursor to a full-fledged tornado.

In the mid-1980s the NWS put together a plan to make use of the new technologies. Since then the program has encountered nothing but turbulence. Among the problems:

FLAWED SATELLITES. In 1986 the Weather Service ordered five advanced satellites from NASA to replace three that were either out of commission or nearing the end of their life cycle. One of the three died of old age two years ago. Another was lost in space. The third is scheduled to run out of fuel in mid-1993. Meanwhile, the new satellites, like so many NASA products, have run into trouble: they are $500 million over budget and three years late, and they have developed a mysterious flaw that makes their temperature soundings unexpectedly weak. A race is on to correct the problem, but if the old satellite dies before a new one is launched, the U.S. will lose its ability to monitor broad weather patterns across the country, a situation NWS director Elbert ("Joe") Friday calls "a national emergency."

RADAR WARS. When the Weather Service put out bids for the Nexrad system in % 1988, the choice came down to Sperry (now Unisys) and Raytheon. Sperry, which promised to build 121 machines for $386 million, was the low bidder. But two years into the job, the company insisted that it needed an additional $250 million to complete it. The government refused to pay, and the company refused to make any more radars. Now, with the Weather Service logging a record year for tornadoes (1,033 so far this year), the program is still stalled in court. A decision on whether to pull the plug on Unisys is expected within weeks.

COMPUTER MORASS. The Weather Service finally replaced its main number- crunching supercomputer -- a clunky Control Data machine -- with a slick new Cray Y-MP last year, and has been upgrading the software for its radar and satellite stations. To speed the dissemination of data and forecasts between its central office in Camp Springs, Md., and weather stations around the country, it is building AWIPS, the Advanced Weather Interactive Processing System. However, AWIPS is already a year late. Meanwhile, a report by the National Research Council in May cast doubt on the ability of the NWS's small staff to manage its other complex new programs.

How did the Weather Service get into such a mess? Part of the problem is bureaucratic: the NWS falls under the sway of the Commerce Department, which has never shown much understanding of or interest in the science or technology of weather prediction. Pinched by tight budgets and layoffs over the past decade, the agency was very nearly shut down under the Reagan Administration, which in its zeal to privatize government operations briefly proposed selling off the Weather Service's satellite network to the highest bidder. Public outcry forced the White House to scrap its plans.

What the budget cutters forgot is that the Weather Service is one of the few government operations that give every American a tangible benefit for his tax dollar. Not only do picnicgoers count on the predictions to save them from a sprinkling, but thousands of businesses depend on the NWS for their very survival -- from airlines plotting the most efficient flight path to utilities trying to meet peak-load demands. Farmers, fishermen, oil drillers, construction companies, snowmakers, moviemakers, grain speculators and baseball umpires all have an urgent interest in accurate weather predictions. With hats in hand, NWS officials tried to impress this upon the Senators last week. And while further technical delays seem inevitable, the betting is that funds for modernization will be found. Or, as the Weather Service might put it: the outlook is overcast, with skies slowly clearing.

With reporting by David Bjerklie/New York, Wayne Greene/Norman and Dick Thompson/Washington