Windswept: The Story of Wind and Weather - Marq de Villiers (2006)
Chapter 5. The Art of Prediction
Ivan's story: Ivan hurried into hurricane status in three days, a full day faster than the official forecast. The transition came early in the day of September 3, when Ivan was still 1,210 miles east-southeast of the Lesser Antilles, at latitude io° and longitude 430. At five A.M. eastern time maximum winds were already 73 miles an hour, just marginally a hurricane, but by the end of the day the pressure had deepened sharply and sustained winds were already over 123 miles an hour—a strong Category 3 hurricane, nearly a Category 4. Such rapid strengthening at such low latitudes had not been observed before.
"Unprecedented" was the National Weather Service's measured word for the phenomenon. It looked as though the people of the Antilles, Puerto Rico, and Hispaniola all had a pretty good chance of a direct hit. The official forecast, based on numerous computer models, skirted Puerto Rico to the south and Jamaica to the north, and tracked right through Haiti and the Dominican Republic. If so, it would miss Cuba's south coast, but not by much. It was too early to say where it would make its U.S. landfall. Much more than twenty-four hours out and the tracks became guesswork. Highly educated guesswork, mind you, but still guesswork. The "normal" recurvature to the northwest and then north—a very common storm path—might not happen. A strong deep-layer anticyclone (another name for a high-pressure center) was hanging about in the atmosphere to the north, which, if it remained in place, as was probable, would prevent Ivan turning at all, in which case . . . what? The center's forecasters consulted their models and reported in a National Hurricane Center bulletin: "There is very good agreement through Day Five, with NOGAPS to the right of the AIDS envelope and the UKMET to the left of the AIDS envelope. The official forecast is slightly to the right of the previous forecast track, and is in very good agreement with GUNS and GUN A." I felt a bit like a kid with a new savings account whose bank manager was blathering on about financial derivative products and bond yields when all I wanted to know was how to take money out and put money in.
I would find out about these models later and how they both reassured forecasters and perversely, complicated their lives. But then the best forecasters, as I would learn, were those people who managed to shade guesswork into intuition, a very different order of information management.
Meanwhile, private yachts scattered into the Caribbean. A few went south, perhaps forgetting the dire lesson of Hurricane Mitch in igg8, which took a strange leftward turn at latitude 130 and slammed into the Honduran coast, trapping dozens of small boats and the tall ship Fantome, lost with all hands.1
By Monday morning, September 6, Ivan was still plowing steadily westward but had weakened some, to the middle ranges of Category 3, with sustained winds somewhere in the 120-mile-an-hour range. But the forecasters weren't happy. They reported that Ivan had "improved his appearance" and was now "better organized"—all of which anthropomorphizing meant that while an earlier concentric eyewall had decayed, a newer eyewall, tighter and clearer, had been formed. The storm was likely to strengthen, perhaps back to a Category 4, but it would now likely pass to the south of Puerto Rico and Hispaniola.
It was still being pushed southward by that protective subtropical ridge: The storm couldn't get around it to head north. Not good news for Jamaica or Western Cuba, but good for the Haitians.
I
In December 1703 an extratropical cyclone of immense ferocity hammered England. It was the worst storm in English history, killing more than 8,000 people, toppling the newly built Eddystone Lighthouse and drowning everyone in it, peeling the roof off Westminster Abbey, and damaging scores of cathedrals and church spires, including the sublime Ely. Within the first six hours of the storm, which lasted almost a week, the Royal Navy lost twelve ships and 1,700 men, a fifth of the entire fleet; some 700 vessels moored in the Thames were driven ashore in a massive tangled heap.
In the storm's aftermath, the muckraking journalist Daniel Defoe, who had just been released from several years' imprisonment both for debt and seditious libel, placed a series of ads in the London papers soliciting first-hand accounts of the storm. He published the results in his best-seller called The Storm. He received dozens of eyewitness accounts of massive damage and a few of miraculous escapes. Queen Anne described the damage to the Royal Navy fleet as a "Calamity so Dreadful and Astonishing, that the like hath not been Seen or Felt, in the Memory of any Person Living in this Our Kingdom." The Reverend Joseph Ralton of Oxfordshire, for example, recounted seeing an immense "Spout, or Pillar, very like the trunk of an elephant only much bigger, crossing a field and, meeting with a great old Oak, snapped the body of it asunder, and coming to an old Barn, tumbled it down." Defoe himself counted 17,000 trees down before he grew weary of the tally, and 2,000 brick chimneys demolished. A part of Queen Anne's palace fell in with a great crash, and the "Lead, on the Tops of Churches and other Buildings, was in many Places roll'd up like a roll of Parchment, and blown in some places clear off from the Buildings; as at Westminster Abbey … and abundance of other places."2
From a wind point of view, the interesting thing about the great English storm—a point made repeatedly by Defoe in his first chapter—was not so much the damage it caused, or what had happened, but that no one had even guessed it would happen. Put another way, everyone knew what had happened, but no one knew why.
It's not that people weren't interested, and weren't trying. Ancient weather lore was derived from careful observation of natural phenomena like clouds and by watching animal and insect behavior. The Farmer's Almanac notion that animals put on heavy coats before a particularly bad winter is not true; nor do squirrels increase their larders when severe winters are ahead; nor do frosts happen more often at full moon. But plenty of other signals are true, at least much of the time. For example, exposed seaweed will indeed swell ahead of bad weather, an effect that has to do with dropping atmospheric pressure.
Mariners particularly, whose safety depended on surviving storms, developed a whole litany of signals to foretell storms. Columbus, who in effect issued the world's first-ever hurricane prediction in 1502, based his anxiety on scudding wisps of cirrus clouds and on long swells from the southeast, signals of a storm that had nearly wrecked his expedition on his first voyage. Don Nicolas de Ovando, the new governor of the Spanish colony, ignored Columbus's warning, and ended up losing twenty-one ships to the hurricane that ensued.3
Columbus, like many a sailor before him and after him, recognized many different cloud types and shapes. It wasn't until 1803, though, that clouds were given their own taxonomy, when they were classified by Luke Howard. He arranged them into three general types: cirrus (curl), cumulus (heap), and stratus (layer). Other types of clouds, such as nimbus (rain clouds), were variations on these three basic types. Modern meteorology also classifies clouds by height—the prefix cirro denotes the highest clouds, 3.6 miles or higher; alto is the prefix given to clouds between 1.2 miles and 3.6 miles; clouds below that have no special prefix.4 High clouds are usually made up of ice crystals, medium clouds of water droplets and ice crystals, and low clouds only rarely contain ice. Clouds, if you can learn to read them, can be accurate, if occasionally ambiguous, weather predictors. Light, scattered clouds alone in a clear sky usually mean strong winds. Clouds lowering and thickening always bring deteriorating weather, while clouds increasing in numbers, moving rapidly across the sky, are often a warning of really bad things to come. In the midlatitudes, scattered clouds with blue skies to the west mean predictably fair weather.
In short order sailors learned to watch for signs of a tropical cyclone. The first signal was a broad trail of cirrus clouds, with heavy showers of rain (it is now known that these travel 300 to 350 miles, ahead of a storm). A sign that a storm was imminent was when that same cirrus departed rapidly, scattering in all directions, followed by a thin, watery veil that pervaded the air. By the time the next sign appeared—an ominous wall of cumulonimbus with layers going in different directions—it was too late to take evasive action, and the gear was lashed to the decks or stowed below, and storm-sails hoisted.
This scattering cloud is a true signal, known to sailors everywhere. Sebastian Smith quotes a fisherman in the Mediterranean port of Carro, who told him the trick of predicting a mistral: "There may well be no indications of bad weather. The air becomes beautifully clear, but then you get clouds like little balls. Clouds like plates, or some say cigars, are another sign. But stay alert, when these balls start to explode and scatter, the mistral will soon be upon you."5
In the LaHave Islands and around the fishing town of Lunenburg, the same signals hold true. An overly clear day is ominous; when you can clearly see the branches of the black spruce on a headland across a large bay, be cautious. You can also smell weather coming:
If with your nose you smell the day
Stormy weather's on the way.
Experiments have found much truth to this old doggerel. High pressure that accompanies fair weather tends to keep scents dormant. When a low pressure system replaces the high, scents are released.6
Folk sayings have persisted for good reason.
Mackerel sky
Mackerel sky
Never long wet
and never long dry
Mackerel skies mean changeable weather.
If the moon's face is red, water ahead
This red can mean dust pushed ahead of the high winds of a low, bringing moisture.
Rainbow in the morning is the shepherd's warning.
Rainbow at night is the shepherd's delight.
A rainbow refracts light, breaking into colors; rainbows in the morning to the west usually indicate rain coming; at sunset a rainbow usually means the rain departing. As early as 1660 these portents were closely watched. "About noon we discovered one of those phenomena called a weather-gall or ox-eye because of its figure. They are looked upon commonly at sea as certain forerunners of a storm. It is a great round cloud opposite the sun and distant from him eighty or ninety degrees; and upon it the sun paints the colors of the rainbow, but very lively. They appear, perhaps, to have so great a lustre and brightness because the weather-gall is environed on all sides with thick and dark clouds." But this observation, by a Jesuit father crossing the equator, came with some skepticism: "However it be, I dare say I never found any thing falser than the prognostics of that apparition, I formerly saw one of them when I was near the continent of America, but which was followed, as this was, with fair and serene weather that lasted several days."7
Red sky at night, sailor's delight;
Red sky in the morning, sailor's warning.
Jesus, an early forecaster, made this prediction, or something very like it, in Matthew 16:2-3, to advise the fisherfolk on the Sea of Galilee, and sailors have been using it ever since. It sometimes works, and sometimes doesn't. Last night around my house there was a lurid red sunset across the bay, and this morning should have dawned calm enough for the lobstermen to be out setting their traps. Instead a warm front rolled through from the Midwest, bringing rain and blustery winds. A sailor named David Kasanof once expressed his skepticism that a sailor would show delight at any sign of redness in the sky, for redness denotes moisture. "Delight," he said, "is an inappropriate state of mind for anyone who has had the poor judgment to go to sea under sail. Under sail, I would suggest a state of alert suspicion bordering on paranoia as the appropriate mindset."8
In fact, in north and midlatitudes in the northern hemisphere, where weather systems move from west to east, red evening sky will bring clear weather about 70 percent of the time, and red mornings will bring foul weather about 60 percent. But in the Caribbean, where weather systems come from the east, the doggerel is useless.
Another old reliable in the northwestern Atlantic is the halo around the sun or the moon, a harbinger of bad weather. (Actually, a halo around the sun when bad weather is already here means it is over, and fine weather is coming.) These halos are caused by light refracting through the ice crystals in cirrostratus clouds; in a perfect example of apparently useless knowledge for its own sake, science has found that the crystals must all be hexagonal, less than 20.5 micrometers across, and producing a ray that is displaced 220, no more no less. Larger halos, known as 460 halos, are produced when light enters through one side of such a crystal and exits near the bottom.9 But the notion that halos are harbingers of bad weather is true, much of the time. Studies have shown that in two out of three times, rain or snow will occur within eighteen hours after a halo appears.
All these folk sayings, derived from long experience, are now much less useful than they once were. Atmospheric pollution has contaminated the signals and made them more erratic.
And of course satellite surveillance and accurate weather forecasting has made them, to some degree at least, redundant. A year or so ago I fell into conversation with a very old man on the wharf at Tancook Island, in Mahone Bay. His family had been lighthouse keepers in the area for three generations and he'd been a ferry boat skipper, and I figured him for a rich store of wind and weather lore.
"What'll the weather be this afternoon, Warren?" I asked.
"Oh," he said, squinting at the sky, "it'll blow a little, then some rain."
"How do you know?"
"Oh," said Warren again, as bland as could be. "I heard it on the radio this morning."
II
Weather forecasts are "right" somewhere between 50 and 80 per cent of the time. Depending on your definition of "right," and on the time frame, it could be worse than that, or better. In climatically unstable or variable regions, the five-day forecasts are notoriously unreliable, and it can be curious how often the fifth-day prediction seems to promise sunny or particularly pleasant weather. But before you sneer, think of the difficulties. The American Meteorological Society described it this way, its defensive tone entirely justified: "Imagine a rotating sphere that is 8,000 miles in diameter, with a bumpy surface, surrounded by a 25-mile-deep mixture of different gases whose concentrations vary both spatially and over time, and heated, along with its surrounding gases, by a nuclear reactor 93 million miles away. Imagine also that this sphere is revolving around the nuclear reactor and that some locations are heated more during parts of the revolution. And imagine that this mixture of gases continually receives inputs from the surface below, generally calmly but sometimes through violent and highly localized injections. Then, imagine that after watching the gaseous mixture, you are expected to predict its state at one location on the sphere one, two, or more days into the future. This is essentially the task encountered day by day by a weather forecaster."10
This more or less inspired weather-related guesswork goes back a long way into recorded time. It is known that the Mesopotamians, the people who gave the world the Hanging Gardens, were trying to correlate short-term weather changes with cloud cover and haloes around the sun and moon as early as 600 B.C. The Chinese, in their more formal and courtly way, tried to codify the weather even more, and around 300 B.C produced a calendar dividing the year into twenty-four segments, each associated with a particular weather pattern. Aristotle's four-volume Meteorologica, in which he dealt not just with wind but with thunder and lightning, hail and clouds as well, remained the standard text until the seventeenth century.
But weather forecasting as we know it, which is essentially a means of tracking wind and air systems and their effects, began in Europe, especially Germany, in the eighteenth century, when networks of towns shared weather observations. It started more formally in America and in Britain in the mid-nineteenth century. The Smithsonian Institution, newly founded, was by the mid-1850s having weather data telegraphed to its offices from those parts of the country that had been reached by railroad, and therefore the telegraph. In the same decade it cautiously, and very much after the fact, began to compile the first national weather maps. The Civil War put a brief stop to these efforts, but in 1865 a series of strong winter gales sank a number of ships on the Great Lakes, prompting the resumption of weather-data collection. A year or two later Cleveland Abbe, director of the Mitchell Astronomical Laboratory in Cincinnati, established a weather telegraphy service. The U.S. Army joined in, and then the Smithsonian again, and by 1870 President Ulysses Grant ordered the establishment of a formal army-run weather service.
Across the Atlantic, the British Meteorological Office was founded in 1854 as a small department in the Board of Trade, and by 1861 was already issuing gale warnings to shipping by telegraphing predictions to harbormasters, who would then hoist appropriately colored cones up a mast. Their forecasts were ostensibly for forty-eight hours, though they acknowledged that their day-two predictions were, to put it kindly, erratic, because most British weather came in from the Atlantic, where few observer stations were located. These forecasts persisted for a decade and then abruptly stopped, over the protests of the sailors who had been using their output.
The same thing was happening elsewhere in Europe, and by the 1870s data from weather observing stations all across the globe led to the construction of the first crude multinational weather maps. Which in turn led to the development of synoptic forecasting—the compilation and analysis of weather data from many different regions in the same period. In September 1874, the official weather map showed a hurricane for the first time.
Around the turn of the twentieth century efforts were made to develop what was called numerical forecasting—that is, forecasting the weather by solving mathematical equations that described the physical laws involved. This wildly optimistic notion—the real complexity of weather data had not yet been recognized—was first expressed by Norwegian weatherman Vilhelm Bjerknes in 1904, the year before Einstein wrote his paper expressing the special theory of relativity.
A short time later British mathematician Lewis Fry Richardson tried to put Bjerknes's ideas into practice and, working furiously for six months, produced a six-hour forecast for Munich. The futility of producing a forecast six months after the event happened was not lost on Richardson; nor was the fact that his forecast was wrong in almost all aspects. Rather bravely, Richardson reported on his failure in his 1922 book, Weather Prediction by Numerical Process. He suggested, tongue firmly in cheek, that the difficulties could easily be overcome: To predict the weather before it actually happens, you would need a roomful of people, each computing separate sections of the equations, and a system, not yet invented, for transmitting the results as needed from one part of the room to another. He guessed no more than 64,000 mathematicians would be needed.
The next real technical advance came in the 1920s, with the addition of huge amounts of high-altitude data. This was made possible by the invention of the radiosonde, a small lightweight box tricked out with weather instruments and a radio transmitter. Radiosondes were sent aloft tethered to helium balloons, climbing to almost eighteen miles before bursting. On the way down, they transmitted wind velocities, temperature, moisture, relative humidity, and pressure information to a ground station. Even now, for regular weather data collecting, radiosonde probes are the workhorses. Literally hundreds of them go up every day. Twice a day, every day, the little weather offices in places like Abidjan and Dakar and Niamey in West Africa, and on the Cape Verde Islands, and for that matter in Honduras, Cuba, a scattering of Caribbean islands, and yet again all up the eastern seaboard as far as Newfoundland, in Greenland and Iceland and the British Isles, helium- or hydrogen-filled balloons carrying a small payload of instrumentation are released into the atmosphere. At noon Greenwich time every day all this data is transmitted to regional offices, where there are any, then to national ones, and then the data flies across the oceans. Within minutes the computers in the national weather centers in Halifax and Miami and Ottawa and London and Hong Kong all have the same data to crunch.
The synoptic maps ("synoptic" here means a general overview) you see on your television weather channel or published in newspapers are based on this data.
Over the years, certain transborder conventions have been developed. Everyone measures the same things at the same altitudes. For example, data from all stations measures atmospheric pressure at 500 millibars. "Normal" weather would show this pressure at about 18,000 feet, so the critical datum is at what altitude and at what geographic node this pressure exists when the measurement is taken. If it is higher at 18,000 feet, this represents a high-pressure zone. If lower, it obviously means a low-pressure zone or front. The pressure gradients are shown on the surface maps by a series of curved lines called isobars. Where the lines are close together, the winds are very strong. Where they loosen up, the winds are light. On upper level charts, where the lines are closest together, are the jet streams. By the same consensus, weather maps show the wind direction as parallel to the isobars, with lower pressure to the left, looking downwind. Wind speed is calculated, in knots, by counting the number of isobars, analyzed at every 4 millibars, that fall within a spacing of 50 latitude, and multiplying by ten. For example, if 3 X isobars lie across 50, the wind is 3.5 X 10 = 35 knots.11 Consecutive sets of maps show the track of the different weather systems.
Numerical-calculator enthusiasts were given a new lease on life with the invention of the first computers in the 1940s. Late in the decade, mathematician John von Neumann at Princeton put together a team of colleagues and a few meteorologists to have another go at the problem. The team's director, Jule Charney, figured he could overcome Richardson's data swamp by using the computer and at the same time filtering out whole sets of data, such as sound and gravity waves. Indeed, in April 1950, Charney's group made a series of successful twenty-four-hour forecasts over North America, and by the mid-1950s, numerical forecasts were being made on a regular basis.
A decade later, on April 1, i960, the first polar-orbiting data-collecting satellite, TIROS 1, was launched. It worked for less than four months, but it gave the world's weather people the first ever pictures of Earth and its cloud cover12
Weather forecasting remained an art. But it now had real numbers, and real-time pictures, to back up its intuitions.
III
Part of the long struggle to understand and then to forecast weather was to accurately measure wind—and then to find some way of depicting it that others would readily understand. The new meteorological offices springing up across the industrialized world needed some way of describing wind force to their clients, at first merchant mariners and sailors, but then all kinds of industrial and shore-based users. As data collection became more widespread, standardization became more and more necessary. It was no use for a telegraph operator on the Prairies to refer to a gale, without some notion of what that gale meant, or how strong it really was. A gale may be a storm to some, just a fresh breeze to others. Notoriously, what fishermen from Gloucester, Massachusetts, considered a bracing wind would send yachtsmen from New York scurrying back to port—or that's what the Gloucestermen said, anyway. In any case, whoever they were, sailors needed something more precise than a "brisk breeze" (or, as some say of gales in New England waters, "a breeze o' wint") to describe what they were likely to experience.
By the late twentieth century scales had been devised for winds of all kinds—regular winds, hurricanes, and tornadoes. We even had scales for windchill and other esoteric matters. But the first of all these was the Beaufort scale.
This is one of these simple measures that seems to have been around forever. Most people know what it is meant to show—a simple numerical relationship to wind speeds, based on real observations of wind's effects. But its history is a little more complicated.
The eponymous Beaufort was Rear Admiral Sir Francis Beaufort, Knight Commander of the Bath, who was born in Ireland in 1774, son of a country parson, an emigre from France with a doctorate in law and a mischievous penchant for acquiring massive debts and then dodging the debt collectors. Young Francis apparently always expressed a yearning for the life at sea, and at the age of thirteen his father acquired him passage on an East Indiaman bound for China and the Indies by way of the Cape of Good Hope. That the vessel was wrecked in Java apparently didn't deter the young man, and he joined the Royal Navy as a midshipman on the Aquilon. His naval career wasn't exactly illustrious, but it was nevertheless quite successful, and by 1800 he had risen to the rank of master and commander. In 1805 he took command of the Woolwich, a forty-four-gun naval vessel that mostly acted as a supply ship for a force that was at the time attacking Argentina. It was there, in the chartroom of the Woolwich, that he devised the first version of the Beaufort scale.
"Hereafter," he wrote in his journal (now kept in a box in a filing cabinet in the Met Office in London), "I shall estimate the force of the wind according to the following scale, as nothing can convey a more uncertain idea of wind and weather than the old expressions of moderate and cloudy, etc." The scale he devised ranged from o, calm, to 13, storm. Within a few years he had allocated the numbers more specifically, to relate them in a way that sailors everywhere would understand. He had reduced the categories to twelve, which ranged from "light air, or just sufficient to give steerage way," through breezes and gales to "hurricane, or that which no canvas could withstand." (For the full Beaufort scale and other wind measurements, see Appendix 2.)
Sir Francis Beaufort
So useful was this scale that by 1838 it was made mandatory for log entries in all Royal Navy vessels, and civilian vessels soon followed suit. One of the first recorded times the scale was put into daily practice was on Darwin's voyage on the Beagle.
Beaufort went on to become official hydrographer to the Admiralty, and died in 1857. His obituary made much of the charts he had devised for the Admiralty, and of his splendid work in ensuring safe passages for vessels everywhere, but made no mention whatever of the Beaufort scale. It was just one of his chores, something useful he did along the way. Useful, but at the time not worth recording.13 It is still, however, being used, in one form or another. The curious thing about Beaufort's original scale is that it doesn't mention the wind speed. The only velocities mentioned are those that might ordinarily be achieved by a man-of-war under full sail at various conditions—Beaufort wanted his readers to look at the ship, not the wind. His numerical system was entirely arbitrary, but the ship's behavior each number was attached to was not—his descriptions would have been very well understood by sailors who had spent years in vessels similar to the Woolwich. They were all Royal Navy men who had put in their time blockading Europe and ranging to Africa and the Indies in sailing ships with quite recognizable characteristics. Beaufort's descriptions are couched in terms of the ship's behavior under sail—the first four for how easily a ship could be propelled, those of five through nine in terms of her sail-carrying capacity, and the rest for mere survival at sea.
The scale jumped from this—a shorthand reckoner for experienced sailors—to something still used by weather offices because of two small technological improvements and one accident.
The two gadgets were the first practical telegraph, invented by Samuel Morse in 1835, and the cup anemometer (wind gauge) invented by T. R. Robinson in 1846. The accident was more than just a small accident—it was a naval catastrophe.
In the fall of 1854 the French and English—for once on the same side—were fighting at Sebastopol when the fleets transporting almost all the supplies they needed for the coming winter siege were struck by a sudden savage gale. On the morning of November 14, the English lost no fewer than twenty-one supply ships, and the French almost as many, a maritime disaster that was rivaled in its intensity only by the almost complete destruction of a French invasion fleet aimed at New England more than a hundred years earlier, in 1746, off Sable Island, no more than two hundred miles from where I now live. The uproar that these losses caused was at least partly responsible for prompting the British Admiralty and the French Marine to jointly sponsor a weather network, in the hope of forecasting future storms before they could wreak similar havoc. This was the ancestor of the World Meteorological Organization.
Since the Royal Navy was involved, it made sense that their Beaufort numbers were incorporated into the new weather office data. But there were obvious problems—the new weathermen in Boston or Belfast or Bratislava, many of whom had never seen the sea, never mind a man-of-war, had trouble agreeing with each other on definitions. The result was confusion, made worse by the proliferation of wind scales. By 1900 there were more than thirty sets in vogue, some disagreeing by more than 100 percent. "It was no longer clear just what the old force scale meant, and few men survived who were competent to judge what the behavior of an 1805 man-of-war would be."14
The first solution was to produce a landlubber's version of Beaufort's observations, in which his "light air giving steerage way" was changed to "light air, direction shown by smoke but not by wind vanes," and his hurricane was changed from "that which no canvas can withstand" to the more stark "devastation occurs."
But in the end this wouldn't do either. In 1912 the International Commission for Weather Telegraphy began the search for real wind speed numbers to attach to the Beaufort observations. A set of equivalents was accepted internationally in 1926, and revised in 1946. By 1955, wind velocities in knots replaced Beaufort numbers on weather maps. At the same time, a gust was defined as "any wind speed of at least 16 knots that involves a change in wind velocity with a difference between peak and lull of at least 10 knots lasting for less than 20 seconds." A squall is more intense, "having a wind speed of at least 16 knots that is sustained at 22 knots or more for at least 2 minutes (in the U.S.) or one minute (everywhere else)."
And so we arrive at the modern Beaufort wind scale, which ranges from dead calm through light air (1 to 3 knots), to hurricane, measured at "greater than 64 knots" (74 miles an hour).
This was much more useful, and usable. Still, modern sailors often use Beaufort's own ship-oriented point of view. As Scott Huler put it in his book Defining the Wind, "Sailors tend to define the Beaufort scale by simply looking at the sea—how high the waves are, what the surface looks like. As for the higher numbers—10, u or 12—who cares? It's just surviving anyway."15 The fishermen on the eastern seaboard tend not to use a scale at all. They judge the wind by the swell and the break of the sea and by the sound of the wind in the rigging, and know from the pitch of the boat and the pitch of the sound when it is time to head for home. Even I have learned to judge a Force 9 gale ("41 to 47 knots") by the sound the spruces make in the wind.
If it is useful to know the speed, and therefore the effects, of gales and storms, how much more useful to have a measure of the planet's most powerful natural force, the Atlantic hurricane and Pacific typhoon? Curiously, though hurricanes were known in the Caribbean from the beginning of the exploration era, and though destructive hurricanes had battered the American coast for centuries (the 1900 hurricane that slammed into Texas all but destroyed Galveston), and though some of these storms actually made their way up the continent as far as the Great Lakes and occasionally back across the Atlantic to Europe, it wasn't until 1973 that a scale could be agreed on.
In the end, it was the construction industry and not the meteorologists that pressed for a solution, and the scale that was finally adopted was devised by Herbert Saffir, a building engineer, and Robert Simpson of the National Hurricane Center, which had been established in Miami. So the scale is called the Saffir-Simpson hurricane scale. It is a 1 to 5 rating based on the hurricane's intensity at the time of sampling. Category 1 hurricanes range from 74 to 95 miles an hour; Category 5s, the most severe, start at 155 miles.
It is a curious fact of wind that even in the very worst storms, and even in an open field with no obstructions, the velocity at ground level is effectively zero. When hurricane forecasters refer to surface wind speeds they really mean velocities at a standard height of 33 feet above the surface; from 33 feet wind is assumed to increase in speed with height, and generally does (see Appendix 7). The Saffir-Simpson scale, then, refers to sustained wind speed, measured over a full minute, at 33 feet. Gusts can be considerably higher.16 Ratings assigned to hurricanes by the weather service are used to give an estimate of the potential property damage and flooding expected along the coast from a hurricane landfall, but wind speed is always the determining factor, as storm surges depend to a considerable degree on the slope of the continental shelf in the landfall region, on the elevation of the nearby land, and on topographical peculiarities— is there, for instance, the possibility of a funnel effect, which would push a surge higher than normal?
Anything above a Category 2 is considered a major hurricane, likely to do considerable damage to buildings and the landscape. (For the complete scale, and a representative sampling of major storms', see Appendices 3, 4, 5, and 6).
Only three Category 5 storms have ever hit continental America. The first was an unnamed storm that struck the Florida Keys in 1935, when the barometer fell to an unbelievable 892 millibars (26.35 inches). This Labor Day storm killed more than four hundred people; some of the victims were quite literally sandblasted, reduced to bones, leather belts, and shoes.17 The second was Hurricane Camille of 1969, which struck the Mississippi coast with sustained winds of over 190 miles an hour and a storm surge of 25 feet above the mean tide levels—a three-story wave rolled through Pass Christian, knocking over apartment buildings, and one appalled survivor who had retreated to his attic was forced to break the window and swim to a nearby transmission tower, from which he saw the water submerge the peak of his roof. He lived two miles from the ocean. Camille is still the most intense storm ever known to have made landfall in America; the winds were so strong—probably 200 miles an hour from Long Beach to the ironically named Waveland— that entire sections of the Mississippi coast vanished. Although there had been plenty of warning, and evacuations had gone on apace, hundreds of people were killed and more than 14,000 homes completely destroyed. It probably didn't much encourage the survivors when President Nixon ordered the dropping of a hundred thousand pounds of the pesticide Mirex on the ravaged communities in an effort to destroy the plague of rats that followed.
Hurricane Andrew of 1992 was the third. Originally classified as a Category 4, Andrew blew away the antennae and radar disks of the National Hurricane Center, which subsequently, prudently, moved inland. The storm's rating was upgraded to a 5 more than a decade later, in late 2003.
Katrina, the storm that mauled New Orleans in 2005, was briefly a Category 5, but was Category 4 at landfall. Ivan, the storm I had been tracking, reached Category 5 not once, not twice, but three times. No other storm on record has done that.
A wind-intensity guide for tornadoes exists too. Because of their explosive but transitory nature, they are really only categorized after the fact, by the damage they have caused.
The scale was first written in 1971 by Theodore Fujita of the University of Chicago, together with Allen Pearson, then director of the National Severe Storm Forecast Center. It is called the Fujita scale, and for those of us with weather anxiety, it makes ominous reading, even if Fujita was, in the upper reaches, somewhat stretching the limits of adjectival vocabulary. Fujita 1, or "moderate" tornadoes, range from 74 to 112 miles an hour, and will peel off roofs, overturn mobile homes, and push cars off roads. Fujita 5s are described as "incredible" and their winds range from 261 to 318 miles an hour, "with strong frame houses lifted off their moorings, cars flying about, trees debarked and even steel reinforced concrete severely damaged." And yet Fujita described one grade above "incredible," which he called "inconceivable." These, if they ever occurred, would carry sustained winds of 319 to 379 miles an hour, but no one will ever know for sure, because all measuring devices would be destroyed, along with pretty well everything else in their path. (For a fuller description of tornadoes and the Fujita scale, see Appendices 9 and 10.)
A quarter of all tornadoes are marked as "significant" (F2), and only 1 percent are Fujita 3s or above, the most violent categories.18
Other useful measurement schemes devised in the twentieth century include the Mach number, the Reynolds number, and, most useful of all, at least in more northerly regions, the so-called wind chill scale. The Mach number, named after Ernst Mach (1838-1916), is used mostly by the military, by NASA, and, as a knowing aside, by former passengers of the transatlantic Concorde flights. It is, simply put, the speed of a moving object compared to the speed of sound; no winds, not even the most violent tornadoes, approach Mach 1. The Reynolds number, named after English engineer Osborne Reynolds, is on the face of it somewhat more esoteric. It's a way of expressing that how winds move through and around objects depends on their speed, density, temperature, viscosity, and compressibility. A wind's Reynolds number indicates whether the flow will be laminar (streamlined) or turbulent. It's widely used in aircraft and car design and to define wind flows around buildings.19
The wind chill, on the other hand, has immediate connections for anyone out in a winter gale. It has always been tricky to measure, and has generally been met with a certain degree of skepticism, because to some degree it measures how people "feel" in cold winds, rather than how cold they really are. But it is important, because winds do exaggerate coldness, and from about — 35° Celsius, severe frostbite sets in within ten minutes—and much faster in high winds.
On calm days, the human body is somewhat insulated from ambient temperature by warming a thin layer of air close to the skin, the so-called boundary layer. Wind interrupts the boundary layer, exposing the skin directly to the air. It takes energy to warm up a new layer, and if successive iterations are blown away, the body feels— and gets—colder. Wind has another effect: It makes you feel colder by evaporating skin-surface moisture, a process that draws away yet more heat. When the skin is wet, it loses heat much faster than when it is dry.
The original wind chill formula was derived from experiments conducted in 1939 by Antarctic explorers Paul Siple and Charles Passel. Their table, first published in 1945, was devised by hanging a small plastic cylinder filled with water on a pole and measuring how long it took to freeze in various wind and temperature conditions. Their formula was modified somewhat over the years, but it still had two basic flaws as a practical guide. Human bodies shiver, for one thing—a plastic cup just doesn't react to cold in the same way. In addition, the measurement was done at the conventional height of 33 feet, where the winds are much stronger than at an average human height. So their table was unnecessarily severe.
In the 1970s the scale was modified by Robert Steadman of Texas Technological College in Lubbock, who proposed a scale that included not only wind but the intensity of sunlight, clothing worn, and other factors.20
And there it rested until the turn of the millennium. The push for a still more accurate measure, unsurprisingly, came from north of the U.S. border—Canada is not as dismally frigid as many Americans profess to believe, but, yes, it is colder. A Canadian-sponsored international symposium on wind chill attracted participants from thirty-five countries, and in 2001 a team of U.S. and Canadian scientists cobbled together a new wind chill index. Data were collected by the research agency of the Canadian Department of National Defence, which added to human knowledge by "volunteering" soldiers for a stint in a refrigerated wind tunnel, where they were exposed to a variety of temperatures and wind speeds. These hardy soldiers were, at least, dressed in winter clothing, but had their faces, the most vulnerable part of the human body to extreme cold, exposed directly to the air. The volunteers also walked on treadmills and were tested with both dry and wet faces.
The new index, now in use by both U.S. and Canadian weather offices, is expressed in temperature like units. The base comparison is to the way the skin would feel on a calm day. If the outside temperature is - 10° Celsius, and the wind chill is listed at -20, it means that your face will feel as cold as it would on a calm day when the temperature is —200 Celsius. Or you can do it the other way: If the temperature is — 10° Celsius and the wind speed is 30 kilometers an hour, the chart will tell you the wind chill is —20 (see Appendix 12). The wind chill, for the technically minded, is expressed in watts per square meter.21
Weather, wind, and storm projections first went truly public in Britain. The British Met Office began issuing forecasts to the public through the press in 1879. Its efforts were warmly greeted. Said the London Standard, "It may safely be conjectured that, unless the authorities of this most completely conducted department had already verified their forecasts within the not extravagant limits of time which are now mentioned, they would not assume this new responsibility."22 This was somewhat overstating the case, as users were soon ruefully to understand. But the forecasts were useful enough that they have never been discontinued. And then, on January 11, 1954, the weather took to television, when George Cowling presented the first "in vision" weather forecast on BBC TV He used an easel and "treatment to walls and background" that cost the Beeb £50.23
Well-coiffed weather people have been with us ever since.
Almost every television newscast in almost every country now contains a synopsis of the current and predicted weather. Many countries now have channels specifically set aside for weather news. Despite the cynicism with which these were initially greeted—after all, a weather channel demands drama to sustain its ratings, and drama means storms and probably exaggerated warnings about bigger, better, more frequent, and more violent storms at that—millions of people have come to rely utterly on their programing to plan travel and other activities. Even cynics, among whom I count myself, click their way to the weather channels when a major storm is thought to be imminent. In dozens of countries the national weather services uses commercial television and radio to disseminate their message; jointly, the weather industry has devised sophisticated but accessible computer graphics to mimic atmospheric weather conditions, and millions of people with only a modicum of scientific learning can now talk knowledgeably about isobars and frontal systems and follow the news about jet streams as eagerly as they follow whether Tiger Woods has won yet another golf tourney.
For those industries where weather is critical, commercial weather advisory services have sprung up. For example, a company called Compu-Weather in 1975 launched what it called "forensic weather services" in support of the insurance, legal, and engineering professions. Its idea was to give careful analysis of what the weather had been at what was guardedly called "the client's point of loss," and it offered same-day service for busy litigators. It would also offer, for a fee, expert witnesses who could be expected to deflect property and liability claims. As an adjunct, it would also provide "24/7 weather decision assistance" for film and TV studios, "to deliver safe and efficient on-location weather shoots." Will Tom Hanks need an umbrella today or not, or will his umbrella merely be blown away in a gale? Odysseus could have used a service like that.
IV
Because of hurricanes' enormous potential for destruction, predicting them has always been a special case of weather forecasting, and in many ways has driven the constant search for new and more sophisticated weather technology. Available data were always depress-ingly sparse. Even after it was known that hurricanes were wandering cyclones, tracking their paths and intensities was at best a hit-and-miss business. At first, of course, this was because the proper technology simply didn't exist. There were no high-flying aircraft or satellites, for one thing. Indeed, when the U.S. Weather Bureau was first set up there were no aircraft at all, and data radioed from ships was only incorporated into forecasts in 1909. But for a while the lack of proper data at the turn of the twentieth century, and the consequent failure to accurately track storms, was also a factor of the politics and personalities of the American weather service itself.
The best hurricane predictions at the time were those of the Cubans, led by the flamboyant genius Jesuit Father Benito Vines, who mixed intuition and meticulous observation to derive his often eerily accurate storm predictions. But although American relations with Cuba were then cordial, a turf war was under way in Washington— the head of the U.S. Weather Bureau, Willis Moore, had actually banned the use of the word tornado in forecasts, fearing it would create panic and through panic would come criticism, something he was ill-disposed to accept, engaged as he was in an attempt to centralize forecasting through his own office. That the Cubans often did things better than his people did was infuriating, and Moore instructed his people to ignore them or even to sabotage them. Vines died before the major storm that destroyed Galveston and killed thousands of its citizens, but his work was carried on by his successors, and the Cuban rivalry with Washington persisted, with fatal consequences—the U.S. bureau discounted Cuba's alarming prognostications about the Galveston storm, and as a consequence failed to warn the Texans in time.24
Aircraft changed the way hurricanes were perceived. Radiosondes, the workhorse of the weather-prediction industry, were more or less useless in hurricanes. Hydrogen or helium balloons could only be released fore and aft of a hurricane—they'd simply be blown away otherwise—and so there was a yawning gap in the knowledge of what was actually happening inside major storms. Forecasters were restricted to land-based observations and the occasional report from a hapless vessel caught in the storm itself, though the crews were usually too busy saving themselves to spend much time updating the weather service. Radar, developed during the Second World War, helped to some degree. The National Hurricane Research Project was founded in the United States in 1952, and first used radar imagery to track a storm off Cape Hatteras in 1955. But radar was then land-based and fixed, useful for last-minute track changes but not for forecasting. Still, during Hurricane Isabel in 2003, the National Oceanic and Atmospheric Administration (NOAA) combined airborne sensors, offshore monitoring stations, and land-based radar to assess the storm. A series of towers with Doppler and SMART radars (Shared Mobile Atmospheric Research and Teaching) were deployed and were able to report minute-by-minute data back to the National Severe Storms Laboratory in Norman, Oklahoma—the eye wall actually passed between two of the towers.25
But it was really aircraft-deployed "dropsondes," first used in the early 1950s but not widely deployed until the 1990s, that revolutionized storm data collection. Dropsondes are aircraft-borne radiosondes; aircraft flying high overhead can drop them through a storm on small parachutes, and before they are destroyed, they can collect the same data the land-launched balloons do—pressure differentials, wind speeds, temperatures, and humidity.
The first storm photographed from space was Hurricane Ana, as early as 1961, but it wasn't until the late 1980s and early 1990s that satellite imagery from dedicated orbiters was employed to capture weather data, and for the first time forecasters could see, in real time, the actual patterns they were plotting on their maps.
More recently NOAA, the parent body of the weather service, together with NASA, launched a network of weather satellites with varying technologies and capabilities. One of the most promising of these, at least for wind measurement, is a technique called Synthetic Aperture Radar—SAR for short. SAR measures wind by calibrating every pixel value in the radar image to what is called "absolute radar backscatter." That is, it measures precisely the size and frequency of local image interference, and matches that to wind speeds and directions. A trial in the Gulf of Maine in 2000 yielded a finely grained map down to a twenty-five-yard resolution. The same year a satellite called Radar Sat was launched carrying the new device, rather unimaginatively dubbed Scan SAR, which enabled scientists at the Jet Propulsion Laboratory to extract data for coastal winds over many hundreds of miles with a resolution of only hundreds of yards. Its use in hurricane-watching would be evident: It would enable the hurricane center to see at a glance small regions of very high winds, as a supplement to aircraft and other measurement.26 Three new wide-swath ScanSARs were launched in 2002, the European En-visat, the Canadian RadarSat-2, and the Japanese ALOS.
Nevertheless, SAR is far from perfect. It has difficulty seeing through heavy cloud cover and in high winds, just the conditions in which accurate data become most necessary.
Another promising, though still in 2005 speculative, technique is the analysis of what are called "ocean microseisms." These are simpler than they sound: When seismometers capable of measuring ground vibrations were deployed decades ago, in the early part of the twentieth century, it became apparent that the ocean itself was giving off a continuous seismic hum, the product of the earth's response to wave-on-wave interactions. More recently it was realized that storms can be located and tracked using this seismic data. Because almost seventy years of archived information exists, "this approach allows, for example, the strengths of El Nino conditions to be assessed for times when [other] ocean data were unavailable."27
By 2004, space-borne "scatterometers" were slowly building up a real-time description of global wind patterns in a system calibrated by matching results with the evidence of wind buoys. A scatterom-eter is a device that sends microwave pulses to the earth, and uses the backscatter to measure the surface roughness. On land, the device has been used mostly to map things like vegetation cover in the Sahara-Sahel region, and to track shifts in polar ice. At sea, the backscattering is caused by ripples and waves, and can be measured down to a few inches. As the Jet Propulsion Laboratory (JPL) puts it, "the idea of remote sensing of ocean surface winds was based on the belief that these surface ripples are in equilibrium with the local wind stress"— which for nonspecialists means that the direction and height of the backscatter can tell you the direction and strength of the winds.
Scatterometers do not always work smoothly, either—ambiguities are encountered in interpreting wind direction, which require several casts at different angles to resolve, and rain can still fuzzy the images—but they are nevertheless the only instruments currently deployed able to give real measurements of ocean surface wind speed and direction under both clear and cloudy conditions, day and night. As W Timothy Liu of the JPL wrote in Backscatter, "They give us not only a near-synoptic global view, but details not possible using numerical weather prediction models. Such coverage and resolution are crucial to understanding and predicting the changes of weather and climate."28
NASA first deployed the technology in a satellite called QuikScat, launched in 1999. The scatterometer on board used pencil-beam antennas in a conical scan, and was able to cover some one thousand miles in a continuous swath that reached 93 percent of the world's oceans in a single day. The device's standard resolution was fifteen miles, but repeat scans in special cases were able to reduce the resolution to about half that.
The satellite was put to early use. The National Hurricane Center had declared Hurricane Floyd to be a tropical depression on September, 7, 1999, but QuikScat had found a surface vortex with the required wind speed a full two days earlier, and was able to track the vortex all the way back to the African coast—the future Floyd emerged from the Sahara to the sea on September 2. "Because such vortices, in their early stages, are too small to be resolved by numerical prediction models," Liu says, "and have no clear cloud signal, the scatterometer, with its high spatial resolution, is the best means, if not the only means, of early detection of hurricanes and the study of their genesis."29
There's a wonderful feeling of empowerment, and a somewhat more hubristic feeling of omnipotence, in looking at a satellite-eye scatterometer picture of the Atlantic Basin. In one taken after Floyd gathered strength and bulled his way into the Caribbean, and subsequently posted on the JPL Web site, you can clearly see the whole area dominated by a massive high-pressure system whose anticyclonic flow was creating strong northerlies along the coast of Spain and Morocco, implying strong upwelling in the ocean. Hurricane Floyd is clearly visible doing his business west of the Bahamas. And you can already see Tropical Depression Gert, later Hurricane Gert, forming a counterclockwise spiral in the central Atlantic. In the northwest Atlantic, off New England and the Canadian Maritime Provinces, nothing much was happening on that day. There would still have been tourists on the beaches around Cape Cod, though of course the satellite images didn't show them. Here and there, no doubt, lobsters were being boiled and broiled and roasted and consumed with the usual accompaniments, mostly beer mixed with beach sand. At that moment, our connections to the Caribbean, and to Mitch and Floyd and Ivan and his cohorts, seemed mercifully slender.
V
The difference between modern national hurricane centers is revealing. The American center in Miami is relatively new; located in an aesthetically challenged and charmless concrete and steel structure hunkered into the earth, its roof bristling with data collection devices, disks and domes and antennae of various description. So secure is the building that the only way to tell if a major storm is raging overhead when inside is to watch the computer screens that will tell you so—or to emerge like a startled rabbit into the open air above, to be buffeted about in person. It is a place built to withstand the most severe winds imaginable—and the denizens of the center can imagine and have experienced winds of awesome ferocity.
The Canadian Hurricane Centre in Dartmouth, which is governed by Environment Canada's weather service, is a very different beast. It was until recently at the top of a small eighteen-story high-rise, for one thing. And "at the top" means just that—it is higher than the elevators go, and you have to trudge the last story up what looks for all the world like an emergency-escape stairway. While the U.S. center is like a bunker, the Dartmouth operational center is located in a large, airy room surrounded by curtain walls of glass, with magnificent views down along Halifax's extraordinary harbor past McNab's Island, which sits like a cork in its mouth, and thence out to sea. High-rises are not good places to be during hurricanes, and indeed when borderline Category 2 Hurricane Juan struck the city in 2003 the staff had to evacuate, not so much because they were alarmed for their safety but because the building's sprinkler system had failed, and thereafter bureaucratic regulations took over. Don Connolly, a Halifax broadcaster who was in constant phone touch with the center as he put the grim news out to his listeners, recalls blanching when he heard that the Hurricane Centre was evacuating. If they were bailing out, why not him? Through the large windows in his studio he could see the trees in the park across the street toppling in the gusts, but his own building was only four stories tall, and though it shook some, it stayed put. So did he.
The U.S. center in its bunker and the Canadian one in its eyrie reflect the realities that they each face. Miami doesn't get hamme red every year but the chances that it will be are quite high. Dartmouth does suffer occasional hurricanes, and they can be more than usually unpredictable when they arrive. ("As hurricanes head north they become a different sort of beast," Chris Fogarty told me. "There is so much we still want to know.") Still, they are not common—the Maritime Provinces get a landfalling hurricane about every two or three years, and, if you include the region's maritime waters, one or two every year, but by the time they reach northern waters they are generally not much more than a moderate Category i. Juan, which rode a northbound jet stream directly from east of Bermuda, scooting so quickly up the ocean that the colder ocean temperatures had no chance to spin it down before it hit land, was a small and extraordinarily violent storm, and the day after it blew off across the gulf of St. Lawrence rumors asserted confidently that it had actually reached Category 3, but in fact it just made it to a 2.
Despite their differences, the operating centers of both institutions are quite similar—computer workstations where the forecasters pass their shifts, color-coded maps showing the current season's activity matched to past history, and, everywhere, photographs of what hurricanes have wrought, a constant reminder that the predictions they are forced to make can affect not just the livelihoods but in some cases also the lives of the people they serve, and the usual jumble of papers, clipboards, and old coffee cups. The maps are color coded in the same palette the NHC uses for its public advisories—green for tropical depression, yellow for tropical storm, red for hurricane, each active track fronted by a bulbous nose spreading out from the storm's current position, indicating the zone of uncertainty in guessing its probable direction. These seventy-two-hour track and intensity forecasts are issued four times a day for all storms in the North Atlantic and northeastern Pacific. They show predicted longitude and latitude, intensity (maximum sustained winds), and predicted path to a tenth of a degree.
By November of 2004, the end of the hurricane season in the Atlantic, the year's map showed nine red tracks. Most of them, because of a strong midseason ridge of high pressure, kept to a westerly bearing and, graphically and ominously, converged on Florida. Ivan, the storm I had been tracking, was shown clearly, but after it passed through Florida, it lost its red color, even on the Canadian map.
On one wall of the Miami center, a composite has been pasted, showing all the named storms from 1871 to 1998, a can of deadly worms writhing across the ocean, menacing in its general predictability (most of the storms went west, curved north, and then caught the prevailing southwesterlies) and even more so for the apparent randomness of each storm's path—some dived down into South America, others headed out to the mid-Atlantic, the occasional one even bullied its way to Baja California; they hit Texas, Louisiana, Mississippi, Florida, the Carolinas, Bermuda, New England, in a patternless wave of awesome destruction.
Sometimes the tracks even circle back on themselves, though this happens more commonly in the Pacific—to the east of Australia, where a regular procession of anticyclones arises every year, there were nine single cyclonic loops, four double loops, and one triple loop among ninety-three tracks, during a period of fourteen years. In the Atlantic, Hurricane Jeanne did a strange double backflip in 2004 and circled around twice before abruptly haring out to sea.
A few African-born Caribbean systems even track back across the Atlantic after catching the midlatitude westerlies. Some reach Britain—London's 1703 storm, during which Queen Anne was escorted to a wine cellar under the Palace of St. James, and some windmills were destroyed through friction-caused fire because they were rotating so fast, is an example of what was probably an African, then a Caribbean storm in its track, before it turned back across the ocean. It had very likely completed its extratropical transition; it was no longer a true hurricane, but it would have coincided with another low to cause a weather bomb—an explosive pressure change defined as a drop of 24 millibars in twenty-four hours with a central pressure below 1,000 millibars, a state that can cause massively high winds very like hurricanes (bombs happen every year in New England and Atlantic Canada, usually more than once). Norwegian explorer Fridtjof Nansen experienced the remnants of a tropical storm in 1888 on the Greenland ice cap. The 1900 Galveston storm was still severe as it passed over Europe and disappeared into Siberia, where no records were kept of its passing. Some of these storms travel more than six thousand miles before they expire, causing havoc in an area several hundred miles wide.
Each workstation in each hurricane center is linked to the datastream coming in from quite literally hundreds of sources—weather buoys, remote sensing stations, aircraft and air traffic control weather data, reports from ships at sea, ham radio operators from Canada through Bermuda all the way across to the Cape Verde Islands, satellite and QuikScat data, Doppler and SAR radar outputs, and many others, including conclusions reached by hurricane and typhoon centers as diverse as those in Honolulu, Tokyo, Dartmouth, London, and Paris. There is even an acoustic model of a hurricane, containing data recorded from the apartment balcony of Dennis Jones, a scientist with the Canadian Department of Defence, and it's an eerie thing to listen to—you can hear the gusts breaking trees, for example, and you can "see" the gusts through their accelerated acoustic signature.30 And each workstation has access through the broadest of bands to the various prediction models that have been put together so painstakingly for so many years.
These models, whose keepers are the Tropical Prediction Center and the National Center for Environmental Prediction, range from simple statistical tables to sophisticated three-dimensional equation simulations. They are of two kinds, track models and intensity models, designed to answer the two key questions: Where is the storm going? How strong will it be? These are the two most difficult—yet critical—things a forecaster must decide. Where is landfall to be? How rapidly will it intensify—or weaken? You know, because you have learned it to be true, that the maximum wind speeds are approximately proportional to the square root of the difference between the central pressure and the surrounding pressure, and the models will show you that. But now you need to know the speed of change, and no formula will yield that up. How to account for the explosive deepening that some storms undergo? Surface sea temperatures? The surrounding ridges and troughs? Something analogous to the intersection of wave trains that can cause rogue waves at sea? A jet stream crossing a storm's path at a critical angle? Before satellites and the models crunched on supercomputers, forecasters had relied entirely on data from ships, and to a lesser extent from planes, and matched those against known historic data; but their forward projections past forty-eight hours were frequently wrong by three hundred miles or more.31
Peter Bowyer, of the Hurricane Centre in Dartmouth, explains some of the difficulties: "A tropical storm is a disembodied unit in a much larger flow of air. Think of it as a cork in a river. The river meanders to and fro, and the cork will go pretty much wherever it is taken, wherever the river goes. In our latitudes"—he's talking of Atlantic Canada, and latitudes in the midforties to -fifties—"the flow of the river is very strong. In the tropics it can be very weak, sometimes so weak that it is no longer the dominant moving force. Then other factors, including rotation of the earth, come into play and can induce secondary rotations within a storm system, and those rotations can themselves become the storm's driving mechanism. So if the atmospheric movement above a hurricane is so weak it is hardly moving, just the difference in rotation between one side of the storm and the other will induce a forward motion and cause the storm to veer. Also, any kind of terrain, such as the mountains of Hispaniola, will change the course of the storm. A steering flow in the upper air is easier to predict because it is generally well defined."32
One of the critical factors in predicting Atlantic hurricane paths is the Bermuda high, the ridge of high pressure more or less permanently anchored in the midlatitudes. How strong it is, precisely where and how stable it is, will often be the critical factor in a storm's course. South of the ridge the prevailing winds are easterly, which is why storms off Africa head west toward America. They will tend to turn north because of the Coriolis force, but the Bermuda high makes it difficult to predict exactly where, or how sharply that recurvature will happen. Once through the ridge, storms hang a right into the south westerlies, and then can move fairly smartly northeastward. Also, the westerlies can accelerate storms very quickly—they can travel from South Carolina to Newfoundland in twenty-four hours, which gives very little warning time to, well, people like me.
Here we have to risk getting lost in a thicket of acronyms: tracking models include CLIPER (from CLI-matology and PERsistence), GFDL (Geophysical Fluid Dynamics Laboratory model), AVN (AViation Run model), NOGAPS (which, if you really need to know, is the U.S. Navy's "global spectral forecast model with 18 sigma levels, a triangular truncation of 159 waves, parameterizations of physical processes and a tropical cyclone bogussing scheme"; it uses complex motion equations to monitor air circulation around the globe) and UKMET (run by the British Meteorological Office, that is, UK MET). CLIPER is perhaps the simplest of the models, merely a historic track of hundreds of previous storms.
Models that attempt to predict a storm's changing intensity include the GFDL, which is also used for tracking storms, and SHIFOR (Statistical Hurricane Intensity Forecast), which uses climatological and persistence predictors to forecast intensity change. SHIPS (Statistical Hurricane Intensity Prediction Scheme) looks at the difference between the maximum possible intensity and the current intensity, the vertical shear of the horizontal wind, the persistence (that is, the previous twelve hour intensity changes), and other factors. A version of SHIPS is available for the Atlantic and East Pacific.33
Some of these models look only at radar data, others look to history, still others are based on broader, global, meteorological patterns. They all do it differently, and the forecasters, or weather analysts, will still have to make the judgment as to which model, or array of models, to follow in making their predictions. A forecaster has to get a feel for what works best in what circumstances, and not so well in others, and come up with a plausible synthesis. For example, most dropsondes measure the winds at the 10,000-foot level, and forecasters have to estimate how far to scale down estimations of surface winds—most multiply the figure by 0.9, but some use other measures, and the NHC has been criticized in the past for underestimating surface winds.
The forecasters are all scientists, with the data-processing training that the term implies, but they have also learned to rely on an ability to sense patterns from scattered information in a way that computers cannot hope to match. This apparent lack of rigor drives engineers crazy. How can you be creative and rigorous at the same time? The hiring profile for a weather analyst would be a mathematician, unflappable under pressure, fast to make judgments, well schooled in the models, and a good communicator. Analysts' judgments can have enormous consequences. When they draw the path-predictions on their computer screens, placing dots where they believe the storm will be in twelve, twenty-four, thirty-six, forty-eight, and seventy-two hours, it is this path that is issued by the center in its public advisories, and emergency preparedness services in a dozen places either stand down or go into heightened alert as a consequence; the same report might make the difference between thousands of people packing up to evacuate or staying home, a decision that can have dire consequences, sometimes life-or-death consequences. In the text explanations that accompany the bulletins, the forecasters can hedge somewhat and second-guess themselves. Many do, and will advise when the predictions are more than ordinarily uncertain, but they know as well as anyone that many of their readers only look at the path on the map and behave accordingly. "If I change one little thing in one of the models," Chris Fogarty told me, "it can change the landfall data by ioo kilometers [60 miles] or more. That can make the difference between destruction and escape." The forecasters have to learn to live with their guesses, to accept their mistakes, to coach their readers constantly not to take specific projections forty-eight hours or more into the future as gospel worth gambling on, and then … to do it all over again next time. Burnout, not surprisingly, is common.34
VI
Sometimes weather "customers," particularly those who are especially vulnerable to storm winds, need more hand-holding than national hurricane centers are able to provide. When you're out at sea in a small boat, you listen to the marine forecasts as often as you can. It can be wonderfully pleasurable sailing the open ocean in a small boat, but terrors lurk too, and yachts will scatter like chaff when they know a hurricane is coming. To supplement the official forecasts, a curious network of amateurs has emerged, people who provide direct and personal, and therefore doubly reassuring, links between weather forecasters and sailors. David Jones, a name nicely out of maritime legend, is one of them. He is a British accountant turned Caribbean weatherman who created the Caribbean Weather Network for yachters in 1993. Based on Tortola in the British Virgin Islands, he transmits on single sideband twice a day, at seven thirty A.M. and five thirty P.M. He gives the official forecasts for the Caribbean but adds his own gloss, his own interpretation of a U.S. Navy forecasting model available on the Internet. In short order, yachting folk came to believe he was generally a day or two ahead of the National Hurricane Center.35
There are several other individuals doing the same kind of duty:
Schooner Arcadia Schooner Arcadia, this is Southbound II Coastal. Do you copy?
The voice is that of Herb Hilgenberg. He is hunched over a transmitter in the basement room that is the studio for Southbound II Coastal, his private commercial radio station. To his right, a computer screen is filling with an image of the globe as an updated satellite weather photo is downloaded, pixel by pixel. Over to his left, another computer is twittering away, compiling the raw data, isobar by isobar, that Herb has been transmuting into yachtsmen's gold—accurate weather data.
He toggles a switch. The airwaves hiss and crackle.
Schooner Arcadia, do you copy, please?
His face, amiable in repose, is furrowed in concentration.
"They're out there in the blackness of the ocean, all alone," he'd said that morning, speaking of his listeners. "The ocean can be very large, when you're all alone and an easterly gale is blowing. It's reassuring for them to hear a familiar voice." We'd been sitting in his sunlit kitchen, staring outdoors past the two satellite dishes tucked away in an L in the house, sucking data from the satellites orbiting overhead. "I'm talking to them directly; they know who I am."
Another sailor had said to me the day before, "Without Herb, you're blind. You don't know what's coming. It's like driving along a country highway without headlights. Herb is your beacon."
Southbound II this is Schooner Arcadia. Good afternoon, Herb. How copy please?
Loud and clear Arcadia. Wliat is your position and conditions?
Our position is 37°24 N and 74°03 W, barometer 1012 and falling steadily, wind is northwest and light, wave height two feet from northwest.
The Arcadia, with captain Dennis Greenwood, is part of the great "they" that constitutes the devoted—not to say obsessive— listenership of the "Herb Show." "They" are the crews and owners of yachts and small boats from the Canaries to the Caribbean, from Venezuela to Newfoundland. Day after day, week after week, month after month, Herb sends his voice out into the Atlantic, and by doing so he helps to save lives, deal with and minimize crises, head off tragedies, track storms, and give early warning of dangerous lows and hurricanes. His listeners, the free spirits of the Atlantic, have come to trust him with their lives and their possessions. And though many Caribbean boaters have become wary of giving their exact position on air, for fear of hijackers and pirates, they know Herb needs impeccable information, and they give it to him. Many of them have learned to their cost that to doubt him is to risk having their boats mercilessly mauled by an Atlantic storm.
Arcadia, I'm coming to you out of sequence. Arcadia, that second low I mentioned yesterday is moving into a position north of you. It is very intense and dangerous. I must urge you to alter course for Bermuda.
Okay, Herb, thank you. We will head for Bermuda now.
On an average night Herb will talk directly to twenty, thirty, forty yachts. For every person he talks to, maybe a hundred are listening. Every night maybe a thousand people tune in to 12.359 (or sometimes 8.284) megahertz to hear the steady, knowledgeable, competent, informative, comforting voice of Herb Hilgenberg.
Sailor Hans Himmelman, from Halifax, put it this way: "When a sailor is down to his last amp of battery power, I'll bet you he's using it to talk to Herb."
Herb is a weather router, and his single sideband radio broadcasts help ocean travelers navigate the always unpredictable weather of the Atlantic.
For this he gets paid—nothing.
And he does it from, of all places, Burlington, Ontario, a thousand miles from the sea.
Herb learned to sail in the challenging waters around St. John's, in Newfoundland, where he grew up, but he moved away from the Rock when he took his engineering degree and then an MBA from the University of Toronto. In 1982 he built his own boat and went to the Caribbean. "On the way, we got hit by a November storm. We were leaving Beaufort in North Carolina, and we weren't prepared … We had no idea it was coming. The U.S. weather service hadn't warned us. It took us six days to get out of it."
A few months later he got a job in Bermuda, "the center of the yachting world. It's the perfect refueling and provisioning port. And every week, you'd hear some horror story about a boat in trouble, demasted in a storm, its sails in shreds, taking on water … Caught unprepared."
Herb started to accumulate more and more knowledge about weather. "With an HF receiver, a single sideband radio, a modem, you can pick up a lot of raw data. I started to call out daily, but informally, forecasts just for my friends." In 1987 Hurricane Emily made a turn over the Turks and Caicos Islands. The weather services forecast it would stall and peter out offshore, but Herb disagreed. His data showed the eye would pass right over Bermuda. He battened down, told his friends to do the same. Finally, the U.S. Navy issued a warning at six A.M. The storm hit at eight. The eye passed just where Herb had forecast.
After that Herb became a constant presence in amateur radio circles and worked with the Bermuda Emergency Services Organization. Every Monday he'd be part of a routine broadcast about weather conditions, and, when hurricanes were about, he'd go on the air every six hours. People would listen. And so it went. "Two or three boats every day would talk to me. By 1989 I was running a daily schedule—I would cover people who became friends. In 1990 it became more hectic, and I was talking to fifteen or sixteen boats a day. It was taking time, more than two hours prep time and an hour on air. Even some fishing vessels in the Caribbean."
In 1992, Herb was talking to a couple of "his" boats in the general vicinity of the annual Newport to Bermuda yacht race. He warned them a gale was coming, though nothing adverse had been forecast. What he didn't know is that the U.S. Navy Training Squadron had a couple of boats in the race, and they took note of the fact that Herb, alone among forecasters, had got it right—and had undoubtedly prevented a few near disasters. After the race, the navy called and asked him to lunch. After that the NOAA people would download Herb's forecast daily—and Herb got access to sensitive satellite data he's still coy about explaining fully. It's no surprise, then, that hanging in his basement studio is a plaque from the Navy Training Squadron that says, "To Herb Hilgenberg, for Best Analysis of North Atlantic Weather and Sea Conditions."
That's where the "Herb Show" came from. In 1994 the Bermuda economy went sour and Herb found himself at age fifty-seven without a work permit or a job. The offers flowed in—come to Florida, to Norfolk, to Annapolis, to the Bahamas, Tortola … Give us your weather service, we'll look after you.
But Herb returned to his suburban Burlington bungalow, and within a few months was back on the air. Four hours prep time now, and three on air.
"It's even better here," he says. "The propagation is actually better. I can talk to boats all the way from the Cape of Good Hope, to Greenland, and over in the Pacific to Hawaii. I get e-mails, phone calls, from England, Europe. It's so big. At any time there are fifty, sixty boats out there, needing me."
For none of this does Herb charge money—that would spoil the special relationship he has with his clients, that of a kindly but, when necessary, scolding uncle. He does get voluntary contributions from grateful customers, enough to almost cover his $10,000 to $12,000 annual expenses, but he won't take money from people who expect a service in return. Those checks he sends back. Nor did he accept a contract from an astonished Lloyds insurance company, whose investigators had noticed that Herb's listeners tended to make fewer damage claims.
Here are a few excerpts from the Arcadia's log:
Dec. 22. Afternoon. We turned for Bermuda as soon as we heard Herb's warning. We have learned not to ignore his forecasts. We arrived safely in Bermuda this morning and tied up inside Ordnance Island. A few hours later we were hit by the winds of one of the most intense lows the North Atlantic has seen for years.
Dec. 26. Afternoon. Steve and Mary arrived at noon. They left Buzzard's Bay two days ago. Their weather service hadn't warned them. Their boat was damaged and they were exhausted, safe only due to years of experience.
Dec. 27. Noon. At 3 am this morning the dockmaster awakened me to help dock a 65-foot sailing vessel. Its French crew looked haggard, exhausted, grim-faced. The main sail was shreds hanging in their rigging, the ship very battered. One of their crew was taken off by ambulance with a fractured neck. Hanging over the stem rail was the empty safety harness of a companion who had been lost overboard in a knockdown.
[Dennis, the skipper, added this unnecessary but heartfelt notation:] This [man overboard] is the nightmare of every ship's captain. Once again I avoided it. Thanks to Herb. The French told me they'd taken their advice from a paid meteorological service in the States. They'd been told it was safe to sail.
But even Herb can't win 'em all. One of his boats, the SV Sparrow, got caught in a hurricane. "We talked every twelve hours for fifteen days, it was exhausting. Finally the boat made it through. But in the end, the owner was so strung out that she fell asleep and ran aground on a reef, losing everything."36
Even when you know a hurricane is coming, even when its path has been accurately predicted and its wind speeds known, even then there are times when there is nothing you can do but hunker down and wait. The power of hurricanes is unmediated, unremitting, affected by neither hope nor belief nor artifice nor device.
Of course, humans being what we are, that hasn't stopped us from trying.