|Late in the Cold War, the United States Navy decided it would be a good idea to survey the altitude of the ocean surface, all over the world, to within a few inches. The point was not to measure waves. The ocean is not flat even where it is calm: it has hills and valleys that depart by as much as a few hundred feet from what we think of as sea level. The slopes of these features are so gentle--they extend over tens or even hundreds of miles--that no ship ever feels them. Yet the Navy decided that submarine commanders, of all people, would benefit from precise measurements of this imperceptible topography.
Why? Because the study of bumps on the ocean surface is a reliable kind of phrenology. it reveals deeper truths about the ocean. Small, shifting bumps are created by the shifting fronts between water masses- -between the warm Gulf Stream and the cold Atlantic, say--and those same fronts scatter sound, thus creating sonar shadows that can hide a Red October. The larger and more permanent hills and valleys are created by something else entirely: by Earth's gravity field, which varies slightly from place to place. Knowing those variations helps a submarine stay on course when it is underwater and sailing blind. And when the time comes to launch a missile at Minsk, knowing the precise direction of gravity at the launch site--it does not always point straight toward the center of Earth--is essential. If the missile starts out on a slightly wrong heading, it will miss its target, thousands of miles away.
So in 1985 the Navy launched Geosat, a satellite that measured the height of the sea surface by bouncing a radar beam off it. In a near- polar orbit, 500 miles high, Geosat circled the spinning Earth, painting it with a tight mesh of densely packed radar tracks. The satellite worked flawlessly, and it yielded the most comprehensive set of gravity measurements ever. For the Navy, the payoff was a substantial reduction in a missile's margin of error--which meant a better chance of hitting Minsk. The Navy was not very interested in making a beautiful map of Earth's gravity field; and even less was it interested in using such a map to chart Earth's most remote frontier--the unseen topography of the seafloor. But David Sandwell and Walter Smith were interested in precisely that. And when the Navy finally declassified the Geosat data last summer, Sandwell and Smith wasted no time in creating the map you see here.
Let us be perfectly clear about one thing: this is a map of gravity, not of seafloor topography. Where the map is blue-green, the rate at which Earth's gravity accelerates a falling object ("little g" in the equations of physics) has more or less its average value of 9.8 meters per second squared--or 980 "gals," as physicists say, in honor of Galileo, who first measured the acceleration. In the bright orange areas of the map, gravity is at least 60 milligals--about 60 parts per million--stronger than average. In the darkest purple areas it is at least 60 milligals weaker.
The map shows tiny variations in gravity, then--and yet to anyone who has ever seen a map of seafloor topography, its broad outlines will look familiar. That shouldn't be surprising: insofar as mountains tend to have more mass than valleys, topography generates gravity. And insofar as one person can be said to have invented the idea that small bulges detected on the sea surface by a satellite could reveal the presence of large mountains on the seafloor, the credit should probably go to a geophysicist named William Haxby of the Lamont-Doherty Earth Observatory. In the early 1980s Haxby took data from a NASA predecessor of Geosat, called Seasat, and made a much cruder version of the map shown here. "People looked at the data," Sandwell recalls, "and said, 'Oh wow, we can see seamounts and fracture zones and all types of features on the seafloor.' That was when people really woke up to this idea."
Sandwell, now a marine geophysicist at the Scripps Institution of Oceanography, was a graduate student when Seasat flew in 1978. It was a good time to be starting out: Seasat opened the door to a whole new field of research, but it never finished the job. Its ground tracks were never less than 50 miles apart, which was why Haxby's map was relatively crude. Geosat's tracks were between 2 and 4 miles apart, and it could resolve features down to about 6 miles across. Sandwell has spent his career, in effect, getting ready for a satellite like Geosat, and once Geosat flew, waiting for the Navy to give up the data. Smith started working with him as a postdoc in 1990 and later moved to the National Oceanic and Atmospheric Administration's Geosciences Laboratory in Silver Spring, Maryland.
In all the years since Seasat, the way you measure sea-surface height from a satellite hasn't changed. The satellite beams a radar pulse at Earth and times how long the pulse takes to bounce back. Since the pulse is known to travel at the speed of light, that measurement reveals how far it is from the satellite to the sea surface.
The next thing you need to know is exactly where the satellite itself is in relation to the center of Earth--which, surprisingly, is much harder to determine. Geosat's velocity and thus its altitude were measured by tracking stations that listened for the Doppler shifting of its radio signals as it passed over them. (A slower speed means a higher orbit.) But tracking stations exist only on land, and on its long journeys over the ocean, the satellite didn't stick to a perfectly constant orbit; it was always being dragged by the atmosphere, for instance, and buffeted by the solar wind. To know a satellite' s path when it is out of sight, you have to calculate all those forces. Researchers have gotten a lot better at that over the years as they have gained experience tracking all sorts of satellites. As a result, Sandwell and Smith could be more precise in 1995 about where Geosat was in 1985 than the Navy could be at the time. That's one reason their gravity map is so sharp.
Once you know (a) how far it is from the satellite to the sea surface and (b) how far it is from the satellite to the center of Earth, calculating the sea-surface height is easy: you just subtract (a) from (b). The slope of the sea surface at any given location then tells you the direction of gravity: it must point exactly perpendicular to the slope, rather than ran along it. Otherwise the water would just run downhill. "The ocean surface acts like a gigantic carpenter's level," explains Smith. "When something's level, there's no direction on its surface that's downhill--downhill is perpendicular to the surface."
Through the action of gravity on water, then, the sea surface becomes like an attenuated visual echo of the seafloor, piling up over mountains, dipping down over trenches. "If you put a mountain on the seafloor, " says Smith, "the extra material represented by the rocks in that mountain add their own gravity to the overall field. If you're right above the mountain, the added gravity pulls down in the same direction, and so it adds to the magnitude of gravity. But if you're off to one side of the mountain, the gravitational field of the mountain pulls toward the mountain, and so the effect is to change the direction of gravity just a little bit."
From the direction of gravity everywhere at the sea surface, Smith and Sandwell could calculate its magnitude everywhere. Then it was just a matter of choosing an attractive color scheme and putting the data on a map.
To anyone interested in the ocean, it is refreshing to see the land for once reduced to a featureless black, and the ocean alive with bold oranges, greens, and purples instead of a vapid aquamarine. But what exactly does the map show? It shows, better than any map has ever done, the basic fabric of the seafloor and how it is created by plate tectonics. The grim-looking shadow zones of dark purple all around the Pacific Rim are deep-ocean trenches, where plates meet their end by diving into the mantle. Northeast of New Zealand the spectacle is almost sad: one sees a long line of volcanoes, recognizable as individuals, all marching toward their doom in the Tonga Trench. Although these volcanoes, the Louisville Seamounts, had already been discovered by survey ships, about half the ones on Sandwell and Smith' s creation had not. Thousands of mountains that would be eminently skiable if they happened to be on land had remained unknown to us; only now is it possible to say that any peak on Earth taller than around 3,000 feet has been put on the map.
An even more striking feature of the gravity map is the deep chasms, known as fracture zones, that cut across the Atlantic, Pacific, and Indian Ocean basins. Like all the seafloor, they are created at midocean ridges, where two plates diverge and hot lava wells up from the underlying mantle. As the young seafloor spreads away from a ridge, into the eternal snow of sediments that falls everywhere in the ocean, its volcanic essence is buried, the way a man's youth is buried as he ages under layers of flab and care; middle age for the seafloor is endless plains of mud. But the fracture zones, which cut across a midocean ridge at right angles and dissect it into distinct segments, are a link to what once was. How exactly they form is still the subject of debate. For whatever reason, less magma tends to well up at the end of a given ridge segment than in the middle, and the result is a chasm--one that is continually being extended as the seafloor spreads away from the ridge, and that remains visible because it is too deep to become filled with sediment.
Across the gravity map the fracture zones streak, and like frozen whoosh lines they visualize the growth of oceans and the drift of continents. Look under the bulge of West Africa and you can see at once how that continent has motored away from South America over the past 110 million years--"motored," that is, at a speed of an inch or two a year (see enlarged map on page 60). Look in the southern Indian Ocean and you see the history of Australia's drive for independence from Antarctica; its halo of fracture zones seems almost to be hurling it into the confusion of the western Pacific. Farther to the northwest, the striated floor of the Arabian Sea shows the trail left by India as it plowed into Asia and raised the Himalayas. The fracture zones show the gravity map for what it is--a still from an action-filled video that has been running for hundreds of millions of years, with no prospect of an end.
The midocean ridges themselves, where seafloor geology begins, are visible on the map, too. The Midatlantic Ridge snakes down the center of that ocean from Jan Mayen off Greenland to the latitude of Cape Horn; near Iceland, where its volcanic effusions are so prodigious that it becomes land, it coincides with the most fiery of gravity highs. Under South Africa, the Southwest Indian Ridge shoots into the Indian Ocean like a fizzling rocket, or perhaps like the trail of some giant and cartoonish deep-sea mole.
Sandwell and Smith, naturally, are drawn to subtler features of the map. South of New Zealand and Australia, for instance, they can point to several places where one ridge segment, presumably with a richer supply of molten rock from the mantle, is growing longer at the expense of its neighbor, and in the process is creating a ghostly V of disturbed terrain that trails behind it like a motorboat wake (see enlarged map on page 61). In this same region the mapmakers have also found ragged, "Atlantic style" ridge segments, their crest cleaved by a deep canyon, directly next to segments that have the gentle, rounded crest characteristic of the principal Pacific ridge, the East Pacific Rise. Since these two radically different types of seafloor topography have been attributed to differences in spreading rate--the East Pacific Rise. typically spreads at six inches per year, around six times faster than the Midatlantic Ridge--it is not at all clear how they can coexist at at neighboring segments, which must spread at the same rate.
"This map is going to focus our attention on some places where we had not usually gone with ships, because they're in remote areas in the Southern Ocean, they're far from ports, and the weather down there is uncomfortable," says Smith. "Those areas probably hold the key to how the plate-spreading system actually works. It's having this global view that is really so exciting. Were going to have to rethink all our hypotheses that were based on limited knowledge of a few easy- to-reach places in the Atlantic and the Pacific."
Although the gravity map reveals subtle details of the midocean ridges, the ridges also highlight its limitations as a guide to seafloor topography. The East Pacific Rise, which winds more or less due south from Baja California down past Easter Island, is one of the most heavily studied areas of the seafloor these days--and yet it is next to invisible on the gravity map. The reason is its fast rate of spread, which allows the newly formed plate to reach a great distance from the ridge crest while it is still hot and thin. A hot, thin plate is a weak plate, one that cannot support the weight of all the mountains on top of it: as a result it bends downward, and in so doing it displaces heavy mantle rock with fight crustal rock. "You have this elevated ridge axis that acts like a mass excess, and ordinarily it would create a positive gravity anomaly," Sandwell explains. "But deeper into Earth there's a mass deficit that exactly cancels the excess. They're matched perfectly because the interior of Earth is fluid, and so all this stuff is floating. It's Archimedes' principle: if you have a cork sitting in the water, part of it will stick up and part of it will go down. And if you calculated the gravitational effect due to that floating cork, you would get almost zero, because of the cancellation."
In general the gravity map matches topography only over relatively short horizontal distances, a hundred miles or so. Individual volcanoes show up clearly because the plate is stiff enough to support their relatively small mass without bending much, and so the mass excess is not compensated by a mass deficit. But huge mass excesses spread over very long distances do not show up well at all--the map does not make clear, for instance, that the midocean ridges rise 10,000 feet or so above the surrounding abyssal plains. In the abyssal plains, moreover, the map reveals its second weakness: there it tends to show too much topography. Where the seafloor is really flat the satellite map instead shows gravity highs created by hills and mountains buried under the blanket of lightweight sediments.
Sandwell and Smith are now in the process of making a real map of seafloor topography that corrects for these limitations of the gravity map. Their method, essentially, is to use actual depth measurements made by shipboard sonars to calibrate the satellite gravity data-- and to replace those data altogether wherever gravity is an unreliable guide to the depth of the seafloor (see maps on pages 62 and 63). The work involves comparing both types of data in 200-mile squares over the entire planet. The two researchers expect the project to take a year: even though it is computerized it is, like all car-tography, laborious and time-consuming.
Sandwell and Smith, however, don't think of themselves as cartographers. "I never set out to make a map as an end in itself," says Smith. " A map was simply a tool for research. There are a lot of geologists who make their living specializing in a particular area of Earth-- they become experts on the geology of the Hawaiian Islands or something like that. But my attitude has always been that you should not make generalizations about the whole from studying one area. So global mapping is just a side effect, really, of trying to search for phenomena that are universal. And if you want to be universal about geology, you've got to study the ocean floors-because they're 70 percent of Earth." Says Sandwell: "I guess I just like being able to explore the oceans this way. It's sort of like having a satellite mission to another planet."
And yet, as good as it is, Sandwell and Smith's map still doesn't provide as good a view of Earth as we have of, say, Venus. A few years ago a spacecraft made a radar survey of Venus not unlike the Geosat survey of Earth. That spacecraft, called Magellan, had a great advantage: Venus is dry. So Magellan could bounce its radar beams directly off the Venusian surface and make pictures that revealed features as small as a few hundred feet across, in contrast to the six-mile resolution on Sandwell and Smith's map.
Nor are satellite maps of the seafloor likely to get much better in the future. In addition to the Geosat data, Sandwell and Smith incorporated data collected by a European radar satellite, the ERS-1, in their map; using as many measurements as possible of the same areas allowed them to do a better job of averaging out the random noise of waves. To make a substantially sharper map, says Sandwell, you would need to put a satellite in orbit for a decade, and there are no plans to do that. Even then you would run up against a more fundamental limitation: when you try to look at seafloor gravity through two to three miles of water--the average depth of the ocean--your vision is inevitably blurred. If two hills are only two to three miles apart, the bulges their gravity produces at the sea surface will merge, making them indistinguishable. "You can't beat that limitation," says Sandwell, "unless you drain the oceans."
Or unless you use sound, which, unlike radio waves, travels through water. The best sonar instruments today can map a swath of seafloor six miles wide with one pass of the survey ship and to a resolution of a few hundred feet-comparable to the Magellan data for Venus. In other words, they pick up just where the gravity map leaves off; it shows the whole globe, but no details finer than six miles across. Marine geologists are already using the gravity map to guide their sonar surveys toward interesting-looking features of the seafloor.
The problem with survey ships, though, is that they are slow, while the ocean is big. Only a few percent of the seafloor has been mapped in this way--by civilian ships anyway. Naval ships are another story. At the U.S. Naval Oceanographic Office in Bay St. Louis, Mississippi, there is a huge cache of sonar data. Collected by a fleet of 8 to 12 Navy survey ships over the last four decades, it is the reason the Navy had no need to do Sandwell and Smith have done with the Geosat data. "We have a very large quantity of sonar data for a substantial portion of the world ocean," says Edward Whitman technical director in the office of the Oceanographer of the Navy. "The never intended to use Geosat as a mapping tool. It provides useful information, but the resolution of it was not good enough for tactical mapping."
The Navy never intended to release the Geosat data either. It was prodded to do so by Vice President Al Gore and by a committee of experts set up by Gore to review the potential scientific utility of Navy data. That same committee has also urged it to release at least some of its sonar data. The Navy is now considering the request. On the one hand, its submariners are understandably reluctant to give away valuable information to a future adversary, whomever that may be. On the other hand, the Navy itself now sees its future in shallow coastal waters, fighting Persian Gulf-style wars. Its survey ships are now working in places like the gulf and the Mediterranean. Given the way the world has changed, perhaps the Navy will see its way dear to turning over the deep seafloor to the rest of us.