Masters of Weather : What They Don’t Know Can Hurt You

On April 29, 1995, a Saturday, afternoon thunderstorms developed just north of Dallas/Fort Worth International Airport. The storms were unremarkable and had been predicted. But when they were 15 minutes away, something unexpected happened. They started dropping hail.

On the east side of the airport, the hail was about 1 1/2 inches, just smaller than golf balls. “Where we have our hangar,” says Warren Qualley, chief meteorologist for American Airlines, “people picked up hailstones that were fiveinches long and three inches wide--as big as Nerf footballs.” But a lot harder.

The hail hammered 80 American and American Eagle airplanes, denting wings and control surfaces, shattering windshields. American called the storms’ damage unprecedented. It was, in the words of one weather official, “a monumental disaster.” American canceled 300 flights the next day, 150 the following day anddidn’t have all its planes back in the air until a month later. The storms cost American and American Eagle almost $1 million every day the planes were grounded--a total of $26 million. All because of a couple of afternoon storms that happened to be packed with fist-sized chunks of ice.

Just one bad weather moment for one company for one day.

Despite new, sophisticated National Weather Service radar in Dallas/Fort Worth, despite American Airlines’ full-time staff of 19 meteorologists, no one was ableto warn the company of a significant weather event three or four hours in advance, so it could have held planes elsewhere or moved them out of danger. While no one takes such an event lightly, in today’s weather world, the hailstorms still qualify, frustratingly, as an act of God.

By coincidence, a computer program in Norman, Okla., 200 miles north of Dallas, was tracking that same group of storms. Actually, the program wasn’t tracking the storms. It was creating them. The program, ARPS (for Advanced Regional Prediction System), is designed to mimic the behavior of the atmosphere. It uses math and physics equations to model the interaction of clouds, wind, humidity, sun and terrain. Give ARPS a location and a time of year, tell it what the weather there is now, and it will spin out a story about what the weather will be during the next several hours.

Such “models” have been around for 40 years. They are just software, even more ephemeral than the pieces of sky they try to capture. But models are an effort to create a crystal ball in a cathode ray tube--using math and computers to chasethe question: What will happen next?

ARPS, being developed and used experimentally at the University of Oklahoma at Norman, ran that Saturday morning and predicted the storms near DFW, creating pictures of them that looked very much like what materialized on color radar screens six hours later.

If the hailstorms’ ambush shows how much today’s meteorology cannot do, ARPS suggests what might soon be possible: The ability, for example, to forecast severe weather hours in advance, when it is still a diffuse cloud of humidity waiting for some convection to turn it into a storm.

ARPS is part of a transformation that is changing the way virtually every piece of weather data in the country is collected, interpreted, used and delivered. Already new technology is making forecasting more reliable, and warnings of potentially fatal weather--flash floods, tornadoes, hurricanes--more timely and precise. In Southern California, for instance, winterstorms in 1992 produced flash floods for which there was virtually no notice; during last winter’s storms, new radar allowed the weather service to give flood warnings three to six hours in advance.

Fueling this revolution is the National Weather Service’s decade-long modernization, which involves new satellites, radars, sensors, computers and newways of thinking. The whole program, which will eventually cost between $4 billion and $5 billion, is part of a wider meteorological advancement that has seen even routine forecasting improve. Today’s two-day forecast is as accurate as the 12-hour forecast was 15 years ago; the seven-day forecast is as reliable as the two-day forecast was in 1975.

The weather world--a reassuring place of rituals, traditions and a certain deeplyrooted conservatism--is not ready to acknowledge that a revolution is at hand: That within the next generation, forecasting may become less like cheerfully vague guesswork and more like science--precise, reliable, something you can act on. That hesitation is part of the very texture of weather culture, which has always been hermetic, and sensitive about its credibility. This is a universe where until 1938, U.S. weather forecasters were forbidden to use the word tornado, to avoid panicking people; where the most important weather data collected each day is still gathered by launching World War II-vintage weather balloons; where, until recently, NWS forecasters relied on 35-year-old vacuum-tube-equipped radars; a world where little--from clouds to hurricanes--is understood with confidence.

Most Americans at the end of the 20th century have become disconnected from the weather, regarding the day’s temperature, humidity and precipitation forecasts essentially as fashion advice. Weather is so much a part of the background noise of daily life--in every newscast, in every newspaper, in every long-distance call to Mom and Dad--that it has come to seem easy to grasp. In fact, it turns out to be a lot more involved than it looks in the hands of The Weather Channel, let alone Willard Scott.

Weather information is the hidden lubricant of the nation’s economy. During the past two decades, a $150-million-a-year industry has grown up around providing forecasts to farmers, railroads, construction companies, utility companies, state and county road departments. No road, bridge or building gets built without data about rain, snow, wind and flooding. The well-designed parking lot requires weather information--because an inch of rain falling on an acre is 27,000 gallons of water, so a 20-acre mall parking lot needs plenty of drainage. The airlines are huge consumers of weather data, with good reason. The federal government says weather is a factor in one-third of aircraft accidents. And the government estimates that weather delays alone cost airlines $3 million a day.

Over the last 10 years, meteorologists have gotten better at large-scale forecasts, predicting accurately across a couple hundred miles and a couple hours. But predicting weather that actually matters to the way people live remains infuriatingly difficult. The Dallas/Fort Worth hailstorms are a vivid reminder that with the weather, a few miles or a few hours is the difference between a curiosity and a catastrophe.

In the modernization, much of the attention is focused not on huge systems that sweep across continents, but on relatively small-scale events--from hurricanes down to individual thunderheads. Weather that matters, in other words.

Despite the progress, there remains plenty to be humble about.

The computer model that was able to predict the hailstorms? It got them right except for one thing. “It was two counties off,” says David Jahn, a researcher working on the program. Still, he and his colleagues are encouraged. “This is the best anyone’s done. We’re trying to figure out, why did we even do as well as we did?”

*

At 6 on a recent fall evening, two men walk out the back door of the weather service office in Norman, Okla., and head for a building nearby that resembles agarage with a small observatory attached.

Inside floats one of the nation’s most critical pieces of weather technology: a weather balloon. The balloon is latex, the washed-out color of Dijon mustard andabout the size of a small tree, with a thin neck and a bulbous body. One of the men attaches a packet of instruments to the balloon. The other man gets ready tolock a tracking antenna onto the balloon’s radio transmitter. The man with the balloon walks it out into an open area beyond the shelter, one hand around the neck, the balloon tugging in the wind.

At exactly this moment--4 p.m. Pacific time, 7 p.m. Eastern time, midnight Londontime--hundreds of technicians around the world are doing exactly the same thing. Twice a day, at noon and midnight Greenwich time, at 800 sites around the world,including 75 in the United States, weather balloons are launched.

In Norman on this evening, the wind is steady from the south. This is the only balloon that will be launched in all of Oklahoma. When the tracking technician is ready, the other man tosses the balloon into the air. It floats away quickly,its bag flexing and wriggling as it rides the wind out of sight. There is something primitive about the moment, an echo of Ben Franklin and his kite. For half a century, this is the way the nation, and the world, has gathered the mostimportant weather data.

The balloons--spaced roughly 300 miles apart in the United States--travel up 10 miles through the atmosphere, recording and transmitting four traits: temperature, humidity, pressure, wind speed and direction. Almost all weather can be deduced from these primary colors of meteorology.

The balloons measure each trait 40 times. The result is something meteorologists pine for: a three-dimensional picture of the atmosphere. At that moment, at thatplace, meteorologists know exactly what’s happening. They know the winds at eachlevel, they know the temperature profile, they know where moisture is accumulating and dissipating. They can compare what’s happening now to what happened at that same place 12 hours ago, they can compare it to what’s happening at that same moment at the other places around the country.

This data is fed into computers in Washington, D.C., which are modeling the atmosphere over the entire country. In essence, using information from 75 sites,meteorologists and their computers try to paint a complete picture of the blanket of air covering the United States, to a height of 10 miles. It’s a wonder they even come close.

“So, every 12 hours, we know what the atmosphere looks like within plus or minusa few hundred miles,” says Jeff Kimpel, professor of meteorology at the University of Oklahoma. “But the things that affect humans occur on a much smaller scale. People want to know what the weather is in the suburb of L.A. where they live. What’s the weather going to be at my house? What’s the weather going to be at the eighth tee at noon? We predict on those same smaller scales, and the humans have to infer. And that’s where the art comes in. It’s very hard to predict what you can’t measure.”

That conundrum captures what has been the central problem of weather for centuries. The weather evolves. To imagine what might happen next, you have to know what’s happening now. And that’s harder than it seems.

Before the telegraph became widely used in the 1840s, weather was strictly a local affair; there was no way for observers to know what was going on in two ormore places at once. The telegraph inspired the creation of the first official U.S. weather bureau, as part of the Army in 1870. By the 1920s and 1930s, meteorologists were so eager for information about the atmosphere that, for a while, there were routine airplane flights just to gather weather data. The federal government in the late 1920s also maintained six “box-kite stations”; ondays when the wind was right, the kites gathered atmospheric data.

Radar, perfected during World War II, gave the first picture of the atmosphere beyond what a person could see. The military pursued radar for its ability to reveal enemy aircraft and considered the clouds and rain that showed up on screens to be irritating “noise.” The nation’s first weather radars were not developed until 1957. Those radars are still in operation, vacuum tubes and all. The first U.S. weather satellite was launched in 1960, and they came into routine use in 1966.

Radar and satellites both offer pictures of what is going on--the satellites by looking down, the radar by sending a beam into the clouds. The images are useful, but they provide little actual data on the weather’s primary colors--temperature, humidity, pressure, wind. Figuring out the weather this way is a little like a doctor trying to figure out what’s wrong with a patient by looking only at pictures of the patient, without taking pulse, blood pressure or temperature.

In fact, simply walking outside to look around remains an important way the weather service gathers information, and the century-old ritual is a routine part of weather culture. Every hour, between 10 minutes and 5 minutes before thehour, a trained weather observer consults the barometer, the wind gauge, the rain gauge, the thermometer, and weather permitting, ducks out to check the sky.This happens at only 250 sites in the country.

So weather prediction has been built on these few shards of information--the surface observations, the satellite pictures, the basic radar data and the 75 weather balloons launched twice a day at a cost of about $250 a launch. The weather people pump this information into the fastest computers available, squeezing every possible interpretation out of it, applying ever more baroque formulas to make it yield as revealing an image of the air as possible.

The effort has paid off. From 1930 to 1939, tornadoes killed an average of 195 people a year. By the 1970s, that number had been cut in half. In the 1980s, it was cut in half again, to an average of 52 people a year. The decrease owes mostly to dramatic improvements in warning precision and timeliness. The numbersare even more compelling when you consider that during the same period, the country’s population doubled.

But, as everyone in the weather world will tell you, forecasting that really matters, “neighborhood” weather, has remained tantalizingly out of reach. The reason is clear: not enough information.

More water, as vapor, flows over the state of Arizona every day than flows down the Mississippi River, for instance; one wouldn’t presume to predict the daily moods of the Mississippi with information from just a couple points along its course. Forecasters feel the same way. Out in the Atlantic Ocean and the Caribbean, where the hurricanes spawn and mature, there is only one daily balloon site to probe the atmosphere, on Bermuda. One balloon for roughly 8 million square miles of ocean. No balloons are launched between California and Hawaii.

Meteorologists refer to this as “data-sparsity,” and it is one reason the ARPS model got the location of the hailstorms near Dallas wrong on that spring morning. It only had so much information about the atmosphere, and it was tryingto envision storms that were a few miles across using data from points much farther apart than that.

John McCarthy, a research meteorologist at the National Center for Atmospheric Research (NCAR) in Boulder, Colo., says that in terms of gathering data about the atmosphere, “the capabilities of the weather service have been marginal for 30 years.” You can’t understand, and you can’t predict what you don’t know about it.

“We were fishing for minnows,” he says, “with a chain-link fence.”

*

Norman, Okla., may well be the most weather-crazed, weather-conscious place in America. A college town 10 miles south of Oklahoma City and smack in the middle of “tornado alley,” this part of the country may have the most consistently violent weather anywhere on earth. For most folks here, there are just two seasons: football season in the fall, storm season in the spring. It’s good they don’t overlap.

Fast-food chains run contests seeking the most dramatic weather videos. TV stations engage in weather wars--tornado videos here are as prized as celebrity videos in L.A. All three stations have their own Doppler radars, and they had them long before the local NWS forecast office did.

Gary England, dean of the TV meteorologists in Oklahoma City, presides over a staff of four full-time and 10 part-time meteorologists at KWTV. He also has 23 volunteer teams on call to chase tornadoes and severe weather and to bring back the pictures. If England declares a Priority One situation--meaning a tornado is on the ground--every news staffer is at his disposal. “We’ll interrupt the pope, the president, God or O.J. for that,” he says.

Dennis McCarthy, who heads the NWS Forecast Office in Norman and is not related to the NCAR’s John McCarthy, likes to tell the story of a lunch with senior officials from weather service headquarters in Washington, D.C. “It was a day wehad a lot of severe weather and watches and stuff, and the restaurant had its TVs on,” he says. The stations, during severe weather, run not just a text crawlalong the bottom of the screen but also a small map of Oklahoma with the affected counties highlighted. One of the headquarters types was marveling at the on-screen display when the waitress came up.

“She didn’t realize who we were or anything, and she says, ‘Oh, there’s nothing to worry about. That’s just a tornado watch. It doesn’t mean there’s any danger.It’s not a warning yet.’ These guys were just flabbergasted that a waitress at Chili’s could explain the difference between a watch and a warning.”

Norman, along with Boulder, is one of the hubs of the U.S. weather world. Acrossthe street from McCarthy’s office is the federal government’s main severe stormsresearch lab, with 100 scientists. It will soon be joined by the national severestorm forecast center. Both are part of the same agency as the weather service--the National Oceanic and Atmospheric Administration (NOAA). Nearby is the University of Oklahoma, which has a vigorous weather research and teaching center.

Most people in the weather world seem to have passed through Norman or Boulder at some point in their careers, including the current head of the NWS, Joe Friday, who earned all three of his degrees in Norman. The weather world is small and generally collegial. Everyone knows everyone. The leaders are all in their 50s and 60s, which has allowed them to span much of the development of modern meteorology and forecasting. They got in when the models and the satellites and radars were just coming along.

The people who do the weather every day, either as forecasters or as researchers, are a culture unto themselves--they love the weather, in all its sameness and explosive changeability. Aside from routinely wearing a tie, which almost no one in the weather world does, Dennis McCarthy is the embodiment of that culture.

McCarthy, tall and lanky, started in the weather service 22 years ago and is themeteorologist-in-charge at the NWS office in Norman, responsible for the forecasting in 58 counties. At 50, he is like a Garrison Keillor version of a weather forecaster: thoughtful, unflappable, gently ironic, with a clipped Midwestern way of talking, often finishing stories by saying “and all that” or “and stuff,” as if to say, “There’s much more, but I’ll spare you.”

McCarthy’s eyes light up when he talks about being “right in the middle of the severe weather, the flash floods, the storms. The forecast office is where it all happens, and all that.”

Once you’ve been in a couple of weather service offices, you have no trouble recognizing the next one. They tend to be quiet places, about the size of a ranch home, and since they are staffed 24 hours a day, they have the feel of a fire station--ready, watching, waiting. The operations center is the soul of each forecast office, an open area the size of a large living room. Around the perimeter is an oval of aging blue computer consoles. Forecasters sit facing out, usually on chairs with wheels. They often roll from computer to computer tocheck things out, and they occasionally roll over to consult with one another. Virtually every operations center has stacks of paper weather maps, often hung on wall-mounted racks.

The people are almost as similar as their offices: watchful, reflective, a little nerdy, predominantly male. “I can walk into a weather service office the world over--in Beijing, in India--and it’s all the same thing,” says John McCarthy. “We are a subculture, irrespective of language or nationality.”

It is a culture with a flavor of the military, partly because many longtime forecasters came out of the military, partly because weather people have such a strong sense of mission, partly because the weather world is infested with acronyms. Nested acronyms, where one of the letters stands for another acronym, are a favorite of the weather service. The joke is that the acronym for the agency’s parent, NOAA, stands for the National Organization for the Advancement of Acronyms.

The weather service and the weather world are insular, and forecasters change slowly. The NWS is struggling to bring its modernization to fruition, spending almost as much each year--about $400 million--on moderni0zation as it does on routine operations. The job of every weather service employee is being redefined. Because of technology, the number of field offices will drop from 254to 118 by the end of the decade, and you can sense the anxiety, and sometimes the resistance, in almost every office. Meanwhile, the weather forecasting has to get done, 24 hours a day.

“Change is hard,” McCarthy says, “but when you first do it, it’s really hard.” It is in the operations center at Norman, the nation’s most modernized weather service forecast office and a prototype of the new forecast office, that the changes become apparent. For one thing, as McCarthy says, “There are no maps hanging on any walls.” And the most distinctive design element of NWS offices, the big blue computer consoles bearing Ford Aerospace nameplates, the backbone of operations since 1980, gather dust in a corner. Norman forecasters use advanced desktop workstations.

In Norman, it’s not data sparsity but data overload that’s the problem. The modernization will bring a million times more bytes to Norman every day comparedto 10 years ago. McCarthy sits at a workstation and pulls up a live radar scan for Oklahoma, layered over a surface map with as much detail as he wants. He canadd data from a network of observing stations in Oklahoma to see how the radar and the surface weather compare. He can add a satellite picture. He can animate the satellite and the radar in coordinated film loops. He can add maps and predictions from computer models. Thousands of forecasting combinations are available, all with a few clicks, all in color.

Using the new system is like jumping from keeping financial records by hand in oversized journals to keeping them on a computer with Quicken. There’s so much more information and so many ways to manipulate and display it, even the abilityto calculate things you couldn’t before--at first, it’s hard to tell what provides the most meaningful picture. It is from this deluge of information thatthe opportunity arises for the NWS to provide a much more densely textured, a much more useful picture of the weather.

Historically, the weather service has issued a variety of separate forecasts, called “products,” for airports and pilots, farmers, boaters and firefighters, and a general forecast for the public. In each office, the “products” are parceled out among forecasters, who sit next to one another, analyzing the same sky and issuing their particular forecasts. Weather service offices have focusednot on understanding the details of the atmosphere but on issuing their products.

Now McCarthy is trying to get his forecasters to think in more holistic terms, to focus on the weather, not the products. In his office, one person, at one workstation, issues all the forecasts. A second forecaster acts like a fireman, whose job, says McCarthy, “is to pay attention to the unfolding weather, to focus on any hazardous situation.”

The idea is to use the new data to come up with a detailed picture of how the atmosphere will behave, then paint that picture using numbers: the temperatures and the humidity will do this, the winds will do this, the cloud cover and the precipitation will do this. The computers do the rest, producing a forecast aimed at farmers or pilots.

That’s exactly the kind of change the weather service hopes the new systems willinspire. Not just a change in the mechanics of the way forecasts are issued, butan end to the compartmentalized way of thinking about the weather. Indeed, the hope is that the new system will inspire a lot more actual thought about the atmosphere.

“This is a little radical,” says McCarthy. “Some people who have been forecasters a long time say, ‘I can’t adjust my thinking from the aviation forecast to the public forecast.’ We don’t want people thinking like that.”

Just as important, the new system can be easily adapted to increasing precision:forecasters can use their grids to specify the weather in as much detail as theyare able, down to parts of cities and counties. Indeed, the system will highlight areas where information is sparse; it will challenge forecasters to beas specific as possible. Tornado warnings can be issued by streets, rain forecasts for particular neighborhoods.

This is the possibility that excites Dennis McCarthy.

“What we can do is just phenomenal. And this is important, too: We’re not doing what we’re capable of. People expect instant perfection. But with these new capabilities, as we get ‘em in place, as we get to understand them, it will be phenomenal compared to what we used to be able to do.”

And all that.

*

If in Norman it is the awareness of the weather that is so striking, in Boulder it is the technology. Boulder is like an R&D; park devoted to atmospheric research. Major research centers are strung from one end of town to the other. Last year, the city magazine dubbed Boulder “the capital of the atmosphere.” There are marvels around each corner--either in someone’s computer or in someone’s imagination.

Sandy MacDonald heads a weather forecast research lab for NOAA in a building in the heart of Boulder. MacDonald is one of the local visionaries--he needs little prodding to start talking about weather forecasts being available on pocket beepers or dashboard computers--but on this Thursday afternoon, MacDonald is thinking about Sunday’s Denver Broncos football game.

MacDonald tunes up his workstation, which is running a more advanced prototype of the software being used in Norman, and pulls up a model that looks at the weather a week ahead. “Ten years ago, I couldn’t have showed you this,” he says.The seven-day model has a snowstorm looping through Denver on Sunday afternoon.

“My son wants me to take him to the Broncos game [but] I’ve been saying since Monday that we’ve got a shot at a really good storm on Sunday. I’m not sure if it starts during the Broncos game or not.”

The software model is an extraordinary achievement. It loads itself with as muchinformation as it can get about the state of the weather right now, then runs that data through equations and predicts the behavior of the atmosphere--around the entire world--for each of the next seven days. Some models now even use the predictions of other models to save work. The short-term models are good enough that the longer-term models can use their predictions as if they were “reality.”The whole system is a wonder of physics, math, computing and, sometimes, wishful thinking.

On this day, the TV weather forecasters aren’t venturing predictions about Sunday’s snow. But MacDonald is. “I think this storm will put down a couple inches,” he says, looping his maps back and forth on the screen. “It could put down six or eight inches of snow.”

In fact, the Sunday snowstorm arrives before halftime of the Broncos game, dropping five inches of snow and throwing Denver’s new international airport into chaos.

Four floors below MacDonald, Gregg Pratt helps manage a computer system that blends weather data with air-traffic control data to provide a picture of flying conditions over the United States and the Caribbean. On a screen, in realtime, Pratt can call up a display showing every airliner aloft over the United States, with the plane’s flight number, destination, speed and minutes until touchdown. Layered over the slowly moving airplanes are contours of weather data, so air traffic can be routed around the weather. The system is about to add 30-minute forecasts, so air-traffic managers can see where problems may develop.

Just a 10-minute car ride northeast is the National Center for Atmospheric Research, with hundreds more weather scientists. One building houses a school for teaching NWS meteorologists the latest science and techniques. As part of its modernization, the weather service has created a new job, the Science and Operations Officer (SOO), who will search out tools and technologies to help hisoffice solve local weather problems. It is the weather equivalent of adding Spock from “Star Trek” to the operations center of each forecast office.

Eric Thaler, 35, is the SOO in the Denver forecast office. Young and lean, Thaler represents an evolution of the weather culture. He conveys the air of a tightly focused software engineer. In his office, he is already promoting a new forecast tool called the Divergence of Q, that couldn’t even be calculated without the new computers. Says one Denver forecaster who regularly uses Div Q: “I wasn’t even taught this in college.”

At the center, meteorologist Morris Weisman has developed a CD-ROM to teach forecasters how storms develop. They can walk through 50 scenarios and see radarscreen data showing how each storm evolves, or fails to evolve. It’s like a doctor being able to go through various simulations of patient symptoms and outcomes.

In Boulder, the problem of “data sparsity” is under assault. Many researchers would like to do away with the antiquarian weather balloons, finding them almostan affront in the digital age. The NOAA lab has developed a network of Doppler radars that look straight up, providing a continuous and finely detailed profileof the winds up to 60,000 feet.

NOAA research meteorologist M.J. Post is pushing a serendipitous discovery that moisture in the atmosphere can be measured, using a basic global positioning receiver. Turns out these receivers are sensitive enough that, anchored in one place, they appear to move up and down ever so slightly depending on how much humidity is in the column of air above them.

Perhaps most immediately intriguing is the possibility of using commercial jetliners to gather the same information as the balloons--a thousand times more frequently, at a fraction of the cost. Some jets already carry a few weather sensors, and the information is used in forecast models. Rex Fleming, a NOAA meteorologist, heads a project to increase the number of airlines, to get more frequent measurements, and most critically, to add humidity sensors. Says colleague Tom Schlatter, “This is the single most important data source over the U.S. for short-term predictions.”

In Boulder, the implications of reliable weather forecasting become real: Precise information would change the way people live. A dashboard display in your car would tell you what the weather would be three hours ahead of you on the interstate. Your television would turn itself on to warn you of an approaching tornado, displaying a map of the tornado’s expected track. Road and utility crews would deploy equipment and personnel with absolute certainty aboutthe severity of the snowfall that’s coming. Hurricane evacuation orders--which cost an astonishing $650,000 for each mile of coast--would be issued for the specific coastal communities where a hurricane ultimately comes ashore.

At NCAR, researcher Bruce Carmichael demonstrates a computer program that shows how close some of this is to being used every day. With virtual reality technology, the program combines weather data and aviation data to offer a three-dimensional view of the weather as planes approach an airport. The program paints a picture of the landscape, the airport, the runways, the clouds, the rain, all updated continuously as the scene or the position of an airplane changes. Rain clouds stack up over and around the runways in their precise positions. Dangerous microbursts appear on the screen as rotating red and white disks. The program allows one to quickly fly simulated approaches into the airport to check the weather along specific flight routes. This program could beused in the control tower by flight controllers. It could also be beamed directly onto screens in airplane cockpits for pilots.

How accurate is it? “It’s as accurate,” says Carmichael, “as the radar is.”

*

There’s more politics to the weather than you’d imagine. Consider Wichita Falls, Texas. At the southern end of tornado alley, Wichita Falls has a small NWS office that will be closed as the agency cuts the number of field offices in half. But the weather service has not been allowed to shut the office because residents have objected to receiving their tornado warnings from 100 miles away.

No matter, even with the office open, the warnings have been coming from Norman for almost three years.

“The staff there--they don’t do a single blessed thing,” says Dennis McCarthy, who supervises Wichita Falls. “They are doing zero work. They take care of people who walk in, and they answer the phone.

“I met with the congressman for that area, and I was this close to him,” says McCarthy, who leans closer in a rare moment of intensity verging on anger. “And I said, ‘I got federal employees down there. They got radar, they got phone lines, they got rent, they got heat--and they don’t do anything. And I’m paying for all that, and I’m a taxpayer too, and I don’t like it.’ ”

The National Weather Service’s modernization has a half-dozen problems like Wichita Falls--budgets, politics and protests--and additional problems the weatherservice has brought on itself. The formal modernization is smartly designed, thescience and technology impressive, the early results auspicious. But the process has been a bureaucratic bog. Every one of the four main technology programs--the weather satellites, the national radar network, the automatic sensors, the new software--is or has been plagued by technical problems, contract problems, deployment delays and cost overruns.

The GOES weather satellites--which are providing more detailed images four times more often as the satellites they replaced--are five years behind schedule and almost 10 times over budget.

The new NEXRAD Doppler radar network is a major advance over conventional radar, allowing forecasters to see inside clouds and storms, revealing the circulation of winds and moisture. The network may be able to provide accurate rainfall totals everywhere it scans. But development of NEXRAD has dragged on for 15 years; perfection of radar during World War II took one-third that time.

The NWS is working to deploy automatic sensors, called ASOS, to take over the basic observing now done by humans. With 868 sites, ASOS will triple the number of locations where weather data is gathered routinely. But even if the sensors’ problems can be solved--six of eight instruments don’t work reliably--it will haveto overcome its early history of failure. Pilots and air-traffic controllers don’t trust it. “ASOS is an unmitigated disaster,” says Mike Smith, president of WeatherData Inc., a commercial forecasting company in Kansas, and well-known weather service gadfly. “The temperature readings are wrong, the visibility readings are wrong, the wind readings are wrong.”

The software to pull all the new technology and information into usable form in a single workstation has been under development for 10 years. The prototypes running in Norman and elsewhere are just that--prototypes. A first version of the actual software is now promised for the end of the year, with full deployment set for 1999, three years late.

Taken together, the problems have eroded morale in the agency and enthusiasm forthe modernization. Many of the problems were avoidable; many are the result of trying to design and install complicated systems using budgets controlled by politicians.

But context is everything. Americans tend to think of their government as lumbering, oafish and hopelessly out-of-date. But each of the four major modernization projects breaks new technical ground. Government scientists invented the radar; the GOES satellites, marvels of technical achievement, have dramatically improved coverage over the country. Both are having exactly the kind of operational effect everyone hoped, and if the automated sensors and the software ultimately work as well, they too will have been worth the trouble.

Because the problems are easier to understand than the science, the weather world often feels misunderstood, if not mocked. During the Reagan administration, there was talk of privatizing all U.S. forecasting and warnings, leaving to commercial firms the launching of satellites, the creation of a national radar network and the tracking of hurricanes. These days, the commercial weather industry is vigorous and innovative, in part because its starting point each day is the huge volume of raw weather data provided by the National Weather Service.

Including the delays and the cost overruns, the NWS’ modernization will cost between $4 billion and $5 billion. Even Mike Smith says that for the $2.50 a year it costs each American to run the weather service, “it’s the biggest bargain around.”

The modernization is happening at a time when people are increasingly aware of the weather, mostly because the weather has seemed so unruly. Last year, two significant reports, including one from the United Nations, concluded that global warming is a reality and could lead to climatic chaos--oceans rising, coastal areas submerged, ecosystems transformed. Last year was also the busiest hurricane season in six decades. One of the 19 named storms--Hurricane Opal--caused $3 billion in damage. Opal came ashore sooner and more powerfully than anticipated--a miscue partly the result of data sparsity. This month, a report will be released indicating that in the first five years of the 1990s, insurance companies have paid out 3.5 times as much money in weather-related damage claims as they did in all of the 1980s.

Scientific understanding of climate is, if anything, less sure-footed than its understanding of weather. But if the world is in for a period of increasingly extreme and dangerous weather, the timing of the modernization will have been fortunate. Meteorologists will need every increment of insight the new tools can provide.

As modernization takes hold during the next decade, as forecasters and researchers learn to exploit the new technology, the weather world could experience two significant changes. First, people could come to expect accurate forecasts, to rely on them. Second, as researchers burrow into the new data accumulating about the atmosphere, they will develop better explanations for what triggers many basic weather events--tornadoes, hurricanes, serious winter storms. There is already so much new information, says meteorology professor Kimpel, “We’ve gone from not having enough data to having enough so that we needa whole new wing of meteorology to handle it.”

As researchers better understand the details of the atmosphere’s physics, they will be able to give forecasters more powerful techniques. So the modernization should be self-reinforcing: The technology will be the source of its own enhancement.

One hundred years ago, medicine was in the same position meteorology is in now. In the 1890s, disease and accident overtook people, and there was little that could be done. Doctors weren’t held in much higher esteem than forecasters are now, and medicine was still a clumsy combination of gentle wisdom, comfort and carpentry. The X-ray had just been discovered; anesthesia was just coming into widespread use; modern antibiotics were a generation off. Doctors were just getting the tools necessary to understand what made people sick. Those tools allowed medicine to blossom in the 20th century, transforming the way people live.

Meteorology at the turn of this century is on the threshold of its own golden age. Meteorologists are finally getting the tools to understand more than the broad outlines of the atmosphere’s moods. The details--the weather that really matters--are finally coming in reach. Forecasting will never be perfect. The atmosphere obeys the laws of physics, but so does your golf game. The physics of the atmosphere are so complicated that if things are only a tiny bit different today than a year ago, the weather this February will likely be very different from last February. And forecasters are far from being able to account for tiny differences. Meteorology, though, has an easier task than medicine. At least for the moment, all anyone wants is an accurate diagnosis.