Storm Kings

by Lee Sandlin

  The delegation didn’t believe, first of all, that the success of the operational forecast had been anything more than a lucky guess. And as for the follow-up warnings issued by Fawbush and Miller—those struck the meteorologists as worse than useless. The meteorologists pointed out that unlike the operational forecast, the later warnings had been not for an exact geographic point like an air base but for very broad and vague areas of the Great Plains. And yet, even so, when the near hits outside the prediction zones were discounted, their accuracy rate had actually been less than 50 percent. So Fawbush and Miller’s method would not lead to the issuance of clear advance warnings to specific communities. And that, as far as the bureau delegation was concerned, made the whole business a waste of time.

  The effective position of the bureau remained exactly as it was. Despite its own chief’s determination that the bureau do otherwise, no bureau forecaster was willing to take the risk of issuing a tornado forecast. The head of forecasting at the Washington, D.C., office even gave an interview in 1952 in which he declared that the policy would never change. “It is impossible,” he said, “to pinpoint a tornado by forecast.”

  Fawbush and Miller remained at the Tinker Air Force Base weather station for the next few years after the operational forecast. The station was renamed the Severe Weather Warning Center and given specific authority to predict tornadoes and other violent weather for military bases across the Midwest from the Ohio valley to the Rockies. During its first year of operation, the center issued 156 forecasts, and there were confirmed tornado touchdowns in or near 102 of its warning areas. In 1956 the center was transferred to Kansas City. Fawbush retired; Miller stayed with the center for another twenty years, first as a senior officer and then (after mandatory retirement from the air force) as a civilian forecaster in the Pentagon’s employ.

  Miller’s contempt for the Weather Bureau was not without its justice. In the years that followed the showdown between the bureau and Tinker’s weather officers, the bureau began a reluctant progress toward tornado forecasting, but it seemed determined to go about it in the most self-defeating manner possible. When a TV weatherman in Tulsa, Oklahoma, took it upon himself in 1951 to issue an on-air tornado warning, thus demonstrably saving many lives, the immediate response of the bureau was to demand that he be arrested. “Warnings for the general public,” the bureau said, “are, by law, a function of the United States Weather Bureau.” (Eventually, it let the matter quietly drop.)

  But its attitude began to loosen. A 1954 book, Tornadoes of the United States, by the bureau meteorologist Snowden Flora (this was the first book on tornadoes published in America during the twentieth century), gave a cautious but favorable summary of the tornado-forecasting guidelines worked out by Fawbush and Miller—exactly the same guidelines that the bureau had rejected a few years earlier. It also explained carefully why Fawbush and Miller’s now-famous tornado warnings were intended for the military and couldn’t be distributed to the general public. However, the author wrote, there was no need for the public to be concerned, because the bureau was now in constant communication with Fawbush and Miller; in fact, “one important forecaster center of the Weather Bureau in the Middle West has an unlisted telephone connection with Tinker Field for emergencies.” In every way—or so the implicit message was—the bureau was now in the thick of tornado forecasting.

  The bureau began encouraging the formation of amateur spotter networks in the Great Plains, along the lines of the military spotter networks set up during the war. The bureau sometimes disseminated warnings from spotters when they could be confirmed, and bureau spokesmen unbent enough to credit the spotters occasionally with saving lives. Then, too, the spread of radar technology to weather stations throughout the country made the amassing of reliable tornado data possible for the first time, and the bureau meteorologists became increasingly confident about identifying dangerous weather conditions that might lead to tornadoes. There were circumstances where even the most skeptical meteorologist would have to conclude that tornadoes were a strong possibility—in March 1952, for instance, when an exceptionally violent and unstable storm system crossed through the South. The bureau was sufficiently alarmed to issue a general tornado forecast for the entire region. The storm system spawned thirty-one tornadoes over the next two days.

  By the end of the 1950s, the bureau had evolved what it thought was a workable method for dealing with tornadoes. It issued “forecasts” several hours ahead of time when it believed there was a possibility of tornadoes in a particular area; this was essentially what Fawbush and Miller had been doing (or for that matter John Finley, almost a century earlier), in that the forecast areas were very large and no specific communities were identified as being at risk. The bureau then issued “alerts” for individual locations when there was a confirmed report, either from radar or from a spotter, that a tornado was on the ground. When the public proved to find these terms confusing, “tornado forecast” was renamed “tornado watch,” and “tornado alert” was renamed “tornado warning”—the terms still used today.

  The value of this system was open to dispute. The accuracy rate of the tornado watches was about one in three; that was scarcely better than Fawbush and Miller’s results, which the bureau had so vigorously scorned. The average lead time of the tornado warnings, meanwhile, was three to four minutes, which barely gave people ahead of the tornado time to take shelter. So what good was the bureau doing? When tornadoes failed to occur in the watch areas, they were attacked for crying wolf; when tornadoes were an imminent danger, they were attacked for staying silent until it was too late.

  To be fair, the bureau was relying on volunteers to spread the alarm; their watches and warnings were distributed to local governmental authorities, newspapers, and radio and television stations, which were then supposed to alert the public. This was effectively a guarantee that warnings would arrive late or not at all. In most of the country, even in Tornado Alley, there were no emergency alert networks; few local TV stations in the country would interrupt their regular programming for any news bulletins involving severe weather. Even when the watches and warnings did get out, the most common response among the public was bafflement. No one understood the bulletins; even when warning sirens were sounded, no one knew what they meant.

  The bureau believed the real problem was that the public didn’t understand what tornadoes were. In the late 1950s and into the 1960s, bureau public affairs officers, school systems, law enforcement and other agencies of public safety, and newspapers and television stations began educating people about tornado safety. This was how a whole generation of American midwesterners first heard about opening the windows when a tornado approached and taking shelter in the southwest corner of the basement. They heard that tornadoes reached a top speed of five hundred miles an hour and that there was a vacuum at the heart of the funnel cloud that made houses explode. They heard, in fact, an accretion of old folklore, forgotten theories, and once-plausible but long-superseded guesswork dating back to John Finley’s time and before, all of it passed off as the current scientific thinking only because there really wasn’t any current scientific thinking on tornado safety. It was the sort of authoritative nonsense people had always associated with the Weather Bureau, the kind of thing that made those like Robert Miller wish the bureau would leave tornadoes to the real experts.

  Certain great figures in the study of tornadoes, it has been observed in the Bulletin of the American Meteorological Society, “possess an ability to view the complex phenomena of the atmosphere with limited reliance on the theoretical underpinning of meteorology.” The subject of the observation might have been James Espy, or John Finley, or even Benjamin Franklin, but in this case it was Miller. Miller, over his years at the Severe Weather Warning Center, acquired a reputation as the greatest practical meteorologist who ever lived. But he had never been particularly well-grounded in meteorological theory. From his success with the operational forecast onward, he relied on his intuition. He simpl
y knew the shape of a tornado-spawning storm system when he saw it, even if he could not articulate the reasons why. Forecasters like Miller, the Bulletin concluded, were something like “human analog machines, with phenomenal memories of synoptic weather regimes.”

  Miller taught many cadets and junior weather officers in his years at the center, and they all remembered the extraordinary visual clarity he brought to the study of violent weather. They particularly recalled his weather maps, which he would draw freehand, with marvelous simplicity and force: they were, one forecaster remembered, “like architectural drawings.” But his skill encompassed the specifics of daily forecasting as well. He had evolved through years of trial and error an almost infallible sense of how to transform a stream of raw data into a lucid and accurate forecast. In the 1960s, he was ultimately prevailed upon to put together a memorandum outlining his forecasting principles; he intended it only for in-house use, but it immediately began circulating throughout both the military and the civilian meteorological communities and is still considered a standard reference today. It is a brilliantly practical treatise on weather forecasting, but the most striking thing about it is that its underlying principles are essentially just ad hoc rules of thumb.

  This is what made his guidelines for tornado forecasting so problematic. His own intuitions about tornadoes were still essentially based on the rules that he and Fawbush had arrived at in 1948: they depended on crude, direct observation of vertical instability and upper-air wind shear in a severe thunderstorm. Miller’s contemporaries, on the other hand, were attempting to devise models of what they called tornadogenesis, involving complex patterns of downdrafts, vortices, and outflow. While both approaches had their value, neither seemed to get anywhere near a solution to the essential problem: determining which thunderstorms were actually going to produce tornadoes, and when, and where. That goal appeared to be out of reach—perhaps permanently. In all his years as a forecaster, Miller never again succeeded in making another tornado prediction as specific and accurate as his first operational forecast. But then, neither did anybody else.

  The new theoretical approaches to meteorology left Miller at best indifferent and at worst enraged. His students observed that he didn’t want to hear about any rival ideas and didn’t want his students or assistants reading about them, either. His teaching method consisted of a relentless drilling in his own forecast guidelines. His goal was to train his students to visualize the large-scale three-dimensional shape of violent storms with exactly the same clarity as he did.

  The sureness of his hand and the lucidity of his thought were accompanied by other personality traits that people found more difficult. Even those who came to feel affection and admiration for Miller thought him an intimidating, sometimes terrifying presence (particularly after Fawbush, always a tempering force, retired from the service). He was infamous for his brutal temper, his competitiveness, his toughness with his underlings, and his disdain for authority. He had a raging intolerance for error and a contempt for contrary views. He was infamous in his office for bellowing insults at those he had to work with; his kindest word for them was “dummies.”

  His attitude toward his old nemesis the Weather Bureau hardened into something like permanent fury. When the Weather Bureau set up its own office for storm prediction in Kansas City, called the Severe Local Storms Unit (known by the acronym SELS—the acronym remained even after the name changed to the National Severe Storms Forecast Center), Miller fiercely resisted cooperating with it. He routinely refused to share any data at all with SELS until he had issued his own predictions. He still blamed the Weather Bureau for rejecting his ideas, and the ones they hadn’t rejected he believed they’d stolen. He wasn’t necessarily wrong about this, either: in the early days, one prominent regional observer had been taking Miller’s tornado forecasts, rewriting them, and surreptitiously passing them along to local newspapers and television stations as his own off-the-record insider information. On Miller’s desk at the Severe Weather Warning Center, several people remembered, was a green telephone that was the direct unlisted line to his counterpart at the SELS office. They dreaded hearing it ring. Miller never picked that phone up willingly and often ended his calls by hurling it at the wall.

  Miller retired from the Severe Weather Warning Center in 1975. He spent the next few years working as a weather consultant to private corporations. He didn’t stay anywhere for long: he was in Maryland, then California, then Louisiana. For a good part of the time he was working under a NASA grant to study commercial applications of satellite weather data. He was never happy in this part of his career and was a poor fit with the white-collar corporate culture of his employers. He left professional meteorology at the end of the 1970s.

  He reemerged in 1994, when he made a public appearance at the National Severe Storms Laboratory in Norman, Oklahoma. He was in extremely poor health, but he roused himself to give a speech about his famous forecast; the speech was described as “spirited” and one that “displayed his rough-hewn nature coupled with his undeniable charm.” Shortly after his return home his health collapsed, and he was permanently confined to bed, where he died four years later. He had spent his last active years drafting his memoirs, which he titled “The Unfriendly Sky,” and working part-time at a hobby shop specializing in military memorabilia.

  19

  Visible Effects of the Invisible

  On September 26, 1948, a waterspout formed in Ariake Bay, off the island of Kyushu, the southwesternmost island in the Japanese archipelago, and rapidly came ashore. That area of Kyushu is today thoroughly urbanized, and the ancient farming villages only survive as the names of railroad stations, but in the 1940s it was still rural: the spout roared up and down hillsides that were terraced with rice paddies, and it shredded the scattered farmhouse roofs for six miles inland.

  A few days later an expert arrived to survey the damage. His name was Tetsuya Fujita. He was twenty-seven years old; he’d been trained as a mechanical engineer and was now an assistant professor of physics at the Kyushu Institute of Technology. He had made a special study of damage patterns in catastrophic events. His most significant work had been a survey of the most catastrophic event in the island’s history, the bombing of Nagasaki; he’d led a special team there in the weeks after the blast and analyzed the pattern of burn marks on the wreckage. (The dangers of radiation poisoning were still only dimly understood at that point; Fujita never showed any ill effects from his time spent at the disaster site, but several members of his team did get severely sick.) He had been able to deduce most of the physical properties of the bomb from his study; he had even managed to work out that it had detonated at roughly five hundred feet in the air.

  But Fujita’s real love was meteorology—in particular the physics of extremely violent weather. After the Nagasaki survey he’d gotten a grant from the government to train science teachers in the basics of the subject, which wasn’t then much studied in Japan. He began his own researches with the thorny problem of the conflicting and variable surface winds often recorded during severe thunderstorms. He spent days at a mountain weather station, watching the storms cascade and surge and billow below him, to see if he could understand their internal logic. That was where he made his first major discovery: convective updrafts in thunderstorms were accompanied by equally powerful downdrafts. (He didn’t know it until much later, but the same discovery was being made at around the same time by American meteorologists.)

  Fujita’s dream was to study the fundamental questions of meteorology full-time. He didn’t see any opportunities for this in Japan; in the years after the war, America was the only place where basic meteorological research was being funded. But then America was also the country with the most violent and interesting weather. So Fujita wrote up his detailed observations of the damage trail of the waterspout, translated them into English (a typewriter with an English keyboard cost him more than a month’s salary), and sent his report unsolicited to the meteorologist Horace Byers, a pro
fessor at the University of Chicago. He picked Byers only because he’d found the name on a meteorological paper that somebody in his college had thrown into a wastebasket.

  In Chicago, Byers read Fujita’s paper and was impressed with it. The two men struck up a correspondence. Fujita went on to get a doctorate in science from Tokyo University in 1950. A few years later, Byers was able to arrange for the University of Chicago to invite him to join the faculty.

  Fujita departed for America in 1953. He landed in San Francisco and took a train across the Rockies and through the Great Plains to Chicago; it was his first glimpse of the vast theater of the atmosphere where he would spend the rest of his life. But what he remembered most about the trip was that he’d only been able to bring twenty-two dollars with him through customs at San Francisco. In order to survive during the three-day train ride, he’d spent all his money on a supply of Fig Newtons and Coca-Cola.

  Fujita arrived in America at exactly the right time: in the boom years after the war, American universities were flush with government funding and were building up impressive new departments in the hard sciences. There was also the unexpected technical bounty of radar equipment; the government was making available to university research labs at bargain rates the radar units from countless decommissioned military planes and ground weather stations. The flood of new data from these units meant that in the 1950s and 1960s, the basic conceptual model for violent storms was being rethought. It was increasingly understood that storms weren’t static events but complex and rapidly evolving dynamic systems, with their own underlying logic and life cycle. The term that began to be used in the 1960s was “storm cell”; a storm front was seen as a kind of tidal wave, with individual thunderstorms flowering and dying all along its length like the curlicues of its crest. Certain cells were so strong and well organized that they became self-sustaining—repeatedly dwindling and regenerating again as they were carried across the landscape: these were called supercells, and it became clear over the years they were studied that these were the cells that most often spawned tornadoes.