Falling Into the Gap

A new theory shakes up earthquake predictions.

By Joshua FischmanOct 1, 1992 5:00 AM


Sign up for our email newsletter for the latest science news

There’s not a lot moving in Landers this evening except for the sun, which is quietly slipping behind the hills. Fading pink light silhouettes the Joshua trees that dot the flat, bare ground all through this small California desert town. Their stumpy limbs cast long, misshapen shadows across dirt roads, across the closed gas station and the closed post office. One of these shadows stretches toward a single-story wood and stucco house. The lights are on, and so is the stereo, blasting an old Grand Funk Railroad song across the high desert. We’re comin’ to your town / We’ll help you party it down, roars the onetime Michigan garage band.

The garage attached to this house couldn’t hold a band. One side of it is tilted up at a crazy angle, the wood near its foundation splintered and torn, the roof sagging at the middle. And stretching across the ground, pointing right at the garage, is a long, twisting scar in the earth. One edge of the ragged divide is a good foot and a half above the other. At places where the scar cuts through a patch of scrub, the bushes on one side have been shoved four feet ahead of their former companions.

This is but one of many cracks that have shattered Landers. The ruptures and chasms stretch for about 40 miles through the desert to the north. Four days earlier, just before 5 A.M. on June 28, Landers was rocked by the biggest earthquake to hit California in 40 years. Just three hours later, another big quake shook the town of Big Bear, 20 miles to the northwest.

Tonight the quakes have produced activity of a different sort. Ten miles south of Landers, seismologists from all over California have convened at a makeshift command post in a county health building in the town of Yucca Valley. Earthquake researchers from the U.S. Geological Survey have pinned large maps of fault systems on the walls. Armed with pointers, they are briefing scientists from Caltech, UCLA, and other institutions on ruptures they’ve spotted during reconnaissance flights in small planes. Throughout the room scientists huddle in small groups, eating cookies, drinking coffee, and discussing early findings. A knot of worried Yucca Valley residents stands in the doorway, hoping to learn if and when a new tremor will strike.

Near the front of the room UCLA geophysicist David Jackson is taking steps to find out. Jackson has recruited a group of volunteers to take ten receivers tuned to the Global Positioning System out into the field. The GPS is a flock of satellites whizzing overhead that beam down signals able to pinpoint each receiver’s position to a fraction of an inch. Jackson wants to use this information to determine how the ground has moved around the faults near Landers and Big Bear and whether that motion means the faults can produce other quakes in the near future.

Most seismologists would say no. According to the theory that has guided most attempts at earthquake prediction, quake risk on a particular fault is low after a big tremor, since the quake releases pent-up strain in the planet’s crust. It takes time for that strain to build up again, so only long-dormant faults are prime hazards. Though the Landers and Big Bear faults could--and indeed would--produce sizable aftershocks in the weeks following the first quakes, they shouldn’t be capable of anything like the magnitude 7.5 and 6.5 rumblers that have people all shook up.

But Jackson thinks that recently rocked areas like Landers are the most likely, not the least likely, to be rocked again by something the same size--or even bigger. I think that when you’ve had one earthquake, you have to be prepared for another one, he says during a break from his planning session. That’s when I’m most worried. Big quakes, Jackson says, come in bunches, and the ground doesn’t need a long warm-up period to release them. He’s just published an analysis of ten years’ worth of earthquakes to back up his claim that the prevailing wisdom should be turned on its head. If he’s proved right, geologists will have to reevaluate not only the way they assess earthquake risks but also some very basic ideas about the way quakes are created.

This notion has created sizable aftershocks of its own. This is a landmark paper, really good work, says Seth Stein, a seismologist at Northwestern University.

It’s hogwash! responds Stuart Nishenko, a seismologist with the National Earthquake Information Center in Golden, Colorado. Dave’s gotten a lot of attention by pushing this, but there’s really nothing to it.

He’s a troublemaker, says seismologist James Savage of the U.S. Geological Survey’s California office, but he’s a good troublemaker. He’s smart as hell and very original.

Seismologists, like frustrated lovers, long to know when the Earth will move, so attempts to predict quakes garner considerable attention. There are two reasons for this. Obviously, the predictions are chances to use science for direct public benefit; lives and property can be saved. But on another level, the predictions are reality checks on an often abstract discipline. Earthquakes tell us what’s happening inside Earth, Jackson says. It’s very easy to generalize about things, but here’s a case where the statements that you make are tested by nature. And you get it right or you don’t get it right. You can hem and haw and say, ‘I think Earth is molten in the outer core,’ or something like that, and you can come up with indirect tests of these things. But your theories don’t get the kind of examination that quake prediction theories get. There’s the chance of being either spectacularly right or spectacularly wrong.

In the case of the theory known as the seismic gap, which has dominated seismological thought for the past two decades, Jackson’s work indicates that it has got earthquakes remarkably, spectacularly wrong.

The gap theory in its modern form took shape in the late 1960s and early 1970s. That’s when geologists agreed that the whole planet was on the move--that the crust was made up of different plates riding on top of conveyor belts of flowing rock in the underlying mantle and that the plates were grinding against one another at their borders. The notion that two plates steadily grind past each other, building up strain just as you build tension in an alarm clock spring when you wind it, has a logical corollary. When the alarm goes off, or the earthquake hits, all the strain is released. And it takes a while to build up that level of strain again. There’s a comforting conclusion to be drawn from this: a fault that’s just released a quake will have to behave for a while, to lie quiet until the steady tick-tick-tick of plate movements can store up enough strain for another explosive release. But fault areas where there have been no quakes for some time are much more likely to shake, rattle, and roll. These are the seismic gaps.

I think what motivates a lot of people to accept the gap theory now is understanding the conveyor belts that are moving the plates, Jackson says, back in his UCLA office. We know the plates move by earthquakes. So it’s very easy to go the next step and say, ‘Well, when there hasn’t been an earthquake for a hundred years and the plates have moved five or six yards, that’s enough for an earthquake. So it’s got to go.’

Jackson was in graduate school when plate tectonics swept through geology, and the gap theory seemed, even to him, like a perfectly logical extension. He didn’t actually think much about it until a chance remark a few years ago set off some alarm bells. I liked the gap theory, so I started with quite an open mind, he says. I was in some discussion with a colleague--I don’t remember who--and he said something to the effect that ‘everybody knows that earthquakes occur in gaps.’

Settling back in a padded chair, Jackson pauses a moment for effect. As soon as I hear the phrase ‘everybody knows,’ I start to wonder, he says, a wry smile beginning to form beneath his gray beard. I start asking, ‘Does everybody know this? And how do they know it?’

Well, there was a paper that was quite influential back in 1979. It documented some quakes in places people had identified as gaps. That paper went on to label many gaps around the world. As time went on, some quakes occurred in those gaps, and people started to say, ‘Yeah, the theory’s true. It’s happening.’ And it was true. The earthquakes were happening in gaps.

But it occurred to me and my colleague Yan Kagan here at UCLA that there were two sides to this story. You should consider not only the successes but also the quakes that aren’t successfully predicted. About three years ago we took a catalog of quakes and plotted where they occurred. We found there were quite a few quakes that occurred in the areas that were not gaps and were not supposed to have large earthquakes. At that time we decided it was worth doing a rigorous test.

The test Jackson and Kagan settled on was based on a map. The team of seismologists at Lamont-Doherty Geological Observatory at Columbia University, led by William McCann, that had published the 1979 paper had charted earthquake activity along the entire Pacific Rim. They drew a line along the border of the Pacific plate going from Australia, up the Asian coast, across the top near Alaska, and then down the west coast of the Americas all the way to Cape Horn. It was a cheerfully colored line, divided into segments of different hues. Different colors meant different quake risks. The Lamont-Doherty workers had noted the number of large earthquakes--magnitude 7 or higher--that had occurred in each segment during the previous 100 years. If a segment had been quake free during that time they colored it a dangerous red--it was a seismic gap, and due for a shaking within a decade or so. Segments that had had a large quake within the past 100 to 30 years were colored orange for an intermediate risk. And segments that had clearly had large quakes within the past 30 years were judged safe and colored green. The line looked like a coral snake winding around the Pacific Ocean.

Jackson and Kagan then took catalogs of earthquakes that occurred from 1979 through 1988 and plotted the locations of all the magnitude 7 or higher quakes on a large version of the coral-snake map. Jackson takes the map from a shelf behind his desk and unfolds it on his lap. We overplotted them on this very map, he says. We counted up how many lay in the red zones, and how many lay in the green and orange zones. And you’ve got the results right here. The snake is speckled with black dots representing quake epicenters. As much to our surprise as anyone else’s, we found that when we plotted out the quakes, a substantial fraction of them were occurring in the safe zones, the green zones, rather than in the red ones.

He moves his finger along a green band stretching over the Aleutian Islands near the top of the Pacific: This was the site of a great quake in 1964, measuring magnitude 8.4. According to the gap theory, he says, the quake should have cleared out the strain stored in that segment, which should then have been immune from further quakes for a number of years. Yet the green band has two dots--two earthquakes of magnitude 7 or higher since 1979.

For Jackson, however, the gap theory foundered on much more than the Aleutians. I think the contradiction comes not in any one event, Jackson says. You have to use the strength of numbers in this kind of test. The contradiction comes in the large numbers of events. There were 30-some, depending on which earthquake catalog you used, that occurred in the green zones. Yeah, there were some gap-filling quakes, but there were also some embarrassments.

All in all, Jackson and Kagan found, quakes rattled green zones five times more often than they hit red zones. They hit orange zones just as often as green zones. In terms of predicting earthquakes, the seismologists concluded, you could have done just as well by picking spots on the map at random. Actually, Jackson says, you’d have done better if you’d reversed the colors.

The two geophysicists published their assault on the gap last December, and immediately the rumbles were felt throughout the seismological community. Seth Stein, from Northwestern, wrote a glowing endorsement of the work in the journal Nature, and the ensuing months haven’t dimmed his enthusiasm. I think Dave has done a very clever experiment and raised some important questions, he says. This was the first attempt to test the gap theory in any rigorous way. The gap is a very simple, attractive model. At a deep level, in our heart of hearts, that’s how we’d like the world to work. But the first time anyone tested it, it didn’t work. A lot of people are disconcerted by this. To paraphrase Einstein, we don’t want to believe that God plays dice with fault systems.

One of the most disconcerted people is Stuart Nishenko. He was part of the team that drew up the 1979 map. Nishenko has spent the past 13 years refining the seismic gap theory, and right now Jackson is not one of his favorite scientists. We’re not very happy with what he and Kagan said, Nishenko says. It was a cheap shot, and it’s important to set the record straight. It was cheap, he continues, because the gap theory has changed since 1979. The theory has become much more specific--and more accurate, Nishenko says--about the kinds of quakes it predicts, and what Jackson tested was an outmoded, discarded version.

Much of the improved accuracy has to do with quake size. The 1979 criterion of magnitude 7 or higher was far too broad, Nishenko says. In the 1980s, seismologists studying seismic waves discovered that quakes under magnitude 7.5 often don’t shatter the entire depth of a fault segment--not all the way from the surface down to the point where tremendous heat and pressure make rock less rocklike and more akin to Silly Putty, able to bend without breaking. There’s still some stiff, strained, and unbroken rock left, so these quakes don’t release all the built-up strain in a segment and shouldn’t be used to declare an area a green, safe zone.

Furthermore, Nishenko says, magnitude 7 quakes don’t necessarily fill a gap. Researchers have learned a lot about the history of various fault segments in the last decade, and one thing they’ve found out is that segments tend to be ruptured at regular intervals by quakes of a similar magnitude. For instance, in central Chile, near Valparaíso, the historic record goes back to the 1500s. The area gets magnitude 8 quakes every 80 to 90 years, he says. There was one in 1822, then in 1906. I published a paper in 1985 forecasting there would be another one about that time. It hit that March, about a month before the paper came out. The next one should be about 2060. The gap theory suggests that Valparaíso should be safe from any magnitude 8 quake until that time.

Magnitude 8 is what Nishenko calls the characteristic quake in this part of Chile. The segment may be rattled by magnitude 7 shocks, but they are 30 times less powerful than a magnitude 8, and they don’t rupture the plate boundary. So, say adherents of the gap theory, in Chile these quakes simply don’t count, and Jackson shouldn’t have included them in his analysis. David is reaching a very strong conclusion, but it’s meaningless, says Allan Lindh, chief of the seismology branch of the U.S. Geological Survey in Menlo Park, California, and a longtime gap proponent. Magnitude 7’s and 8’s are completely different events. It’s like comparing apples and oranges. In fact, it’s more like comparing grapes and watermelons.

After a little thought Lindh moves away from fruited metaphors and seizes on a sharper analogy. Have you ever driven a wedge into a block of wood? he asks. Hear those cracks and pops as the wedge goes in? Those are wood fibers breaking. The wedge is like one plate driving into another, and those pops are the 7’s. When the wood splits, that’s the big quake.

Perhaps it’s unfair to point out that one of those pops was the 1989 Loma Prieta quake, magnitude 7.1. Lindh certainly doesn’t mean to trivialize an event that leveled portions of Santa Cruz and San Francisco, fractured the Bay Bridge, and knocked down the upper deck of Interstate 880, killing 62 people. Another magnitude 7 pop hit Cape Mendocino and destroyed part of the town of Ferndale three weeks before Lindh spoke. But the fact is, the vast majority of the quakes that rattle our planet are magnitude 7 or less; great, magnitude 8 quakes are rare. The seismic gap idea was originally intended to explain how large earthquakes work. Theorizers have, in essence, narrowed the gap’s focus to improve its predictive value. But if the theory, in its refined, modern version, cannot explain most of the large earthquake activity on Earth, how useful a theory is it?

It’s a truism in science that the more parameters you add to your model, the less it proves when it fits something well, Stein says. That something just gets smaller and smaller, and it ends up being a very tiny part of what you first wanted to explain. With most things that have turned out to be right in geology, the first pass at them turned out to be basically correct, he continues. But this thing, the gap, worked so badly on the first cut that it’s hard to see how tweaking the model is going to fix it.

Jackson readily admits that the gap theory has evolved since 1979. But these refinements haven’t been clearly stated, he says, and he couldn’t test a moving target. This is sort of like a card game where the other guy puts his cards down on the table, then sees what you’ve got and says, ‘Well, maybe I’d like another card after all.’ It’s the recognition that he’d like another card that brings progress in his thinking. But my gut feeling is these improvements are not going to help.

The problem, Jackson says, is that the gap theory has a basic disagreement with reality. Quakes don’t reduce stress to the point where there aren’t more quakes. The reason I believe this, he says, is there are instances where you have two or more big earthquakes in more or less the same place within a few years. There were magnitude 8 quakes in 1898 and 1899 near Alaska, two great quakes in 1811 and 1812 near New Madrid, Missouri, and about eight other pairs that Jackson can point to. Clearly Earth didn’t have to rest and save up another 15 feet of motion in order to have the next earthquake. It was still in the fault after the first one, he says.

The reason for this, he suggests, is simply that quakes don’t release stress on a segment completely; they only redistribute it. Sure, a crack releases stress, he says. But it also increases stress at the top and bottom of the crack. It’s like a crack in your car windshield. There’s a tremendous amount of stress at the tips of the crack, and that’s why you have to get the whole windshield replaced, even though it’s just a small crack. I think Earth works like that. We know from simple earthquake models that the ruptured surface has its stress reduced, but the area just around that--the tip of the crack--has its stress increased, actually by a lot.

Basically, a quake moves stress to the ends of a segment, increasing the strain on neighboring segments. One of those segments could then rupture, moving stress around again and reloading the original offending segment. That segment then cracks again, years ahead of schedule and much to the consternation of people who believe in a seismic gap.

Where the gap theory views each fault segment as isolated, affected chiefly by its own history, Jackson’s view is of a more interconnected, erratic world that is continually rearranging itself to barely avoid a massive failure. It’s a world that behaves something like a sandpile. If the pile’s slope gets too steep, it doesn’t collapse to a low slope. It simply sheds a few grains of sand on the one steep part. And other parts get steeper because the sand has been moved around, so they collapse just enough to stay stable, and the cycle starts again.

In a way, this approach is more straightforward than the gap. If you get a magnitude 7 quake, it tells you that whole area is lit up, and you may get another quake, Jackson says. You don’t have to worry about what is a plate-rupturing quake and what isn’t. And by my way of thinking you don’t have to worry about particular structures. Somewhere within that neighborhood, there’s another structure that’s ready to go. So from a practical point of view, you can’t walk away from an area like Loma Prieta and say, ‘Okay, this one is broken, it’s now safe, it’s not going to go again.’ It could have another quake, possibly stimulated by something that didn’t break in the first one, or by redistribution of stress.

This statement gets something approaching a guffaw from Nishenko, who has estimated the probability of the Loma Prieta segment’s breaking again within the next decade as just slightly greater than zero. What time frame are you talking about here, Dave? he asks. Ten years? Twenty years? Why don’t you make a map? Then we’ll come back ten years later and evaluate it.

Nishenko is right to point out that such predictions are too vague to be helpful. But quake prediction right now is a murky business no matter what theory you subscribe to. We’re in a sort of real-time experiment, he says. We don’t have the luxury of running tests with lab rats. But there are great societal demands to make these predictions. There’s demand for a fast turnaround time, so we can’t wait. But we’re light-years from being able to say, ‘Next week at 3 P.M. there will be a quake in the southern Santa Cruz mountains.’

Or in Landers and Big Bear. Neither Nishenko’s nor Jackson’s ideas can offer much in the way of reassurance or warning to the townspeople watching the geology discussion in Yucca Valley after the quake. There’s not enough history known about the frequency of ruptures along the Landers fault for Nishenko to apply the gap theory. Though he believes the fault segment has released its share of strain, he doesn’t know how long it takes to build up to another large quake. Jackson thinks the whole area is unsettled and likely to loose another quake, either on the Landers fault or near it. But tomorrow, or next week, or next decade? He doesn’t know.

Lurking behind these questions is a bigger uncertainty, and the biggest worry on everyone’s mind in the desert tonight. Landers and Big Bear are terribly close to the granddaddy of all California faults, the San Andreas. It passes within ten miles of the Big Bear quake zone on its journey north. Although no one knows how frequently or how hard this section of the San Andreas shakes, the section immediately to the south of it last loosed a big one in 1680, when it uncorked something around a magnitude 7.5. The California desert wasn’t a popular place to live in the seventeenth century, but today such a quake would rock cities such as Palm Springs, San Bernardino, and Riverside, as well as the small desert towns. And Los Angeles is less than 100 miles away. Have the Landers and Big Bear quakes done something to set the stage for devastation?

Again, the gap looms large in the answer to this question, but here it’s the gap between society’s needs and seismologists’ knowledge. Jackson, who a week earlier had been saying that he didn’t think the southern San Andreas was as ripe for a quake as other scientists did, because the area had been relatively quiet, has reversed himself. On the principle that shaky things come in bunches, he now thinks something is due on the San Andreas, but once again he has no timetable. He does, however, have a mechanism: the GPS sensors that his troops carried into the desert, as well as other mapping tools, indicate that the ground between the Landers and Big Bear faults shifted northward during the quake. This block of crust had been sitting just to the north of the San Andreas, pressing down on it like a clamp. With the shift this pressure has eased off. The big fault, slashing under the desert ground for miles and miles, has one less thing holding it together.

Up on the surface, where houses and communities have sprouted up among the Joshua trees, the seismologists and the townspeople can only watch and wonder. And wait.

1 free article left
Want More? Get unlimited access for as low as $1.99/month

Already a subscriber?

Register or Log In

1 free articleSubscribe
Discover Magazine Logo
Want more?

Keep reading for as low as $1.99!


Already a subscriber?

Register or Log In

More From Discover
Recommendations From Our Store
Shop Now
Stay Curious
Our List

Sign up for our weekly science updates.

To The Magazine

Save up to 70% off the cover price when you subscribe to Discover magazine.

Copyright © 2022 Kalmbach Media Co.