The danger of a very large earthquake striking the coast between northern California and British Columbia proves much greater than suspected
Few people question the possibility of a devastating earthquake once again hitting Los Angeles or San Francisco. The state of Alaska has also suffered some serious shaking, including, in 1964, one of the world's largest earthquakes. Until recently, however, many residents believed that the intervening territory from northernmost California to southern British Columbia (an area sometimes referred to as Cascadia) was a safer place to live. Seismologists had recognized that Vancouver and Seattle were not exactly sheltered--sizable earthquakes buffeted the region in 1946, 1949 and 1965--but no truly disastrous events had ever damaged these cities.
Yet views have changed drastically. Ten years ago Thomas H. Heaton of the U.S. Geological Survey and Garry C. Rogers of the Geological Survey of Canada began warning that giant earthquakes could indeed strike this seemingly quieter stretch of coast. Initially, many scientists questioned the seriousness of the threat, but most doubters now realize that such earthquakes have happened in the past and will do so again. How could perceptions have shifted so quickly?
To understand the change in thinking requires some knowledge of the way seismologists estimate how and where powerful but infrequent earthquakes occur. For most active fault zones, the rate at which earthquakes take place decreases with increasing size in a systematic way, as was shown in the 1930s by Beno Gutenberg and Charles F. Richter. This regular pattern applies up to some maximum earthquake size--one that corresponds to a break of the entire fault zone from end to end. Using the Gutenberg-Richter relation, seismologists can gauge how often large earthquakes strike a given place even if no such events have ever been recorded. Engineers can then design buildings, dams and other structures accordingly.
ALASKA suffered widespread devastation in 1964, when a giant thrust earthquake struck on Good Friday. Much of Anchorage (upper left) was damaged by the ground motion, and coastal sites around Sewar(upper right) were inundated by water and mud. The southern coast of Alaska still shows the effects of this great earthquake in the many trees killed when the ground subsided (bottom), allowing saltwater to drown their roots.ALASKA suffered widespread devastation in 1964, when a giant thrust earthquake struck on Good Friday. Much of Anchorage (upper left) was damaged by the ground motion, and coastal sites around Sewar(upper right) were inundated by water and mud. The southern coast of Alaska still shows the effects of this great earthquake in the many trees killed when the ground subsided (bottom), allowing saltwater to drown their roots.
In a few areas, this strategy fails. Sizable earthquakes can hit without small ones, presenting seismologists with a vexing problem: How can the danger from large earthquakes be reasonably defined? This difficulty applies to Cascadia, where one of the tectonic plates underlying the Pacific Ocean thrusts underneath the coast of western North America in a process termed subduction. Although regional seismic activity can be quite intense in some areas inland of this coast, no earthquakes of any size have been detected where most of the motion is seemingly focused--on the main thrust fault that separates the Juan de Fuca plate from the North American continent.
In global perspective, the lack of such thrust earthquakes is surprising. Most subduction zones have experienced great thrust events (defined as those having a Richter magnitude higher than 8) at some time. These earthquakes are especially concentrated around the rim of the Pacific Ocean in an vast hand called the ring of fire--a name that comes from the lines of active volcanoes that lie landward of where the oceanic crust dives into the earth's mantle.
No Large Earthquakes?
There are several possible explanations for the absence of major subduction earthquakes. Although the Cascadia part of western North America has many of the characteristics of a subduction zone, the Juan de Fuca plate may have stopped moving toward North America in geologically recent times. Twenty years ago, when geologists first debated this question, my colleague at the Pacific Geoscience Center Robin P. Riddihough and I wrote an article making the case that convergence and underthrusting are indeed continuing. The work of many researchers has since confirmed that the Juan de Fuca plate has not made a sudden stop. Persuasive evidence for continued motion comes from the study of sediments lying underwater at the base of the continental slope. These muds and sands were laid down in the deep sea as flat layers, but even the most recent deposits are found to he highly contorted. The North American continent, acting as a giant bulldozer blade, has scraped them off the oceanic crust and left them as crumpled evidence of continuing subduction.
Perhaps the most dramatic evidence for ongoing subduction came in 1980 with the volcanic eruption of Mount St. Helens in southwest Washington State. Scientists have recognized for many years that such volcanoes are a consequence of subduction. Some geologists had thought the Cascade volcanoes were dormat. But this volcanic cataclysm left little doubt that the Cascadia coast is indeed an active part of the ring of fire.
To reconcile the plates' convergence with the absence of thrust events, some scientists have supposed that the downward push below the coast involves a smooth, stable slide, not the jerky "stick-slip" behavior that generates earthquakes. The alternative explanation is that the fault between them is truly locked (the friction being large enough to hold the two plates firmly together), so that there is not enough movement to generate even small earthquakes. If the fault is freely sliding, the chance of large thrust earthquakes is slim. But if the fault is locked, the plate convergence must be accommodated by the silent hut deadly buildup of strain in the rocks around the fault, the makings of a significant earthquake.
The lack of substantial earthquakes in the historical record might at first seem to favor the idea that the fault is slipping quietly. That interpretation, however, neglects the brevity of the record along this coast. Only a little more than 200 years ago did the explorers Juan Perez and James Cook first visit the region. The limited written history contrasts markedly with the span of Japanese records describing many large subduction earthquakes and the so-called harbor waves ("tsunamis") that usually resulted from them. That detailed archive extends hack to the seventh century.
BURIED PEAT (brown layer at right) Iying below coastal marshes attests to a past earthquake. Such deposits formed when the surface of the land dropped suddenly, and tsunamis washed into the subsided region, burying the intertidal vegetation in sand. Mud then filled the remaining depression before plants once again established themselves on the new surface. A series of peat layers lie below.
When Was the Last One?
BURIED PEAT (brown layer at right) Iying below coastal marshes attests to a past earthquake. Such deposits formed when the surface of the land dropped suddenly, and tsunamis washed into the subsided region, burying the intertidal vegetation in sand. Mud then filled the remaining depression before plants once again established themselves on the new surface. A series of peat layers lie below.
To probe those times before Europeans arrived on these shores, researchers have sought traces of past earthquakes in the geologic record. They found some telltale evidence in sheltered inlets where salt marshes form between high and low tide. Excavations of these coastal marshes uncovered a remarkable record. Brian F. At water of the U.S. Geological Survey was first to show that distinct layers below the present marsh (spaced at successive depths of about a meter) contain peat that is made of the remains of vegetation identical to the flora now living in the intertidal zone. He concluded that each peat deposit constitutes a former marsh that was buried when the ground abruptly dropped with the release of strain in a sizable earthquake.
What makes his interpretation even more convincing is that many of the buried peat layers are covered by sand washed in by the huge tsunamis that rushed onto the subsided coast. Theoretical modeling as well as preserved geologic effects on the shoreline indicates that these waves attained heights of 10 meters on the open coast and much higher skill in some confined inlets.
After the tsunamis dissipated, mud slowly filled the subsided region, and the marsh vegetation returned. Thus, the repeated sequences of peat, sand and mud clearly demonstrate that large earthquakes have plagued the region in the past. But how long ago were these prehistoric upheavals? The ages of the peat layers are difficult to determine precisely, but coastal fir trees have been found that were drowned by the ocean after the land abruptly subsided. By examining growth rings and measuring radiocarbon in these trees, researchers have estimated that they died in the last great earthquake, which hit the area about 300 years ago. Before that, similar events struck at irregular intervals of about 500 years.
This conclusion is also supported by unusual deposits found far out on the floor of the ocean. Scientists at the University of Oregon have sampled seafloor sediments in long core tubes and found fine-grained muds alternating with sandier layers. Mud is typical of the deep sea bottom; it accumulates from the slow, continuous rain of fine sediment settling from the ocean above. The sandier sediments, however, are strange to find far from shore. John Adams of the Geological Survey of Canada provided an explanation: energetic earthquakes could have triggered huge submarine landslides that carried coastal sediments down the continental slope and out onto the deep ocean floor.
The timing of the events is hard to judge from the sediments, but a peculiar deposit found near the base of some of the cores gives an important clue. This layer contains volcanic ash from the eruption of the former Mount Mazama in Oregon (now known as Crater Lake). That colossal explosion--similar to the recent Mount St. Helens blast--happened 7,700 years ago. Assuming that the rain of mud onto the seafloor was steady, the chronology for these earthquakes proves similar to the results from coastal peat deposits. The most recent event happened about 300 years ago, and the 12 previous submarine landslides were separated by 300 to 900 years.
(left) TSUNAMIS generated from giant earthquakes in western North America would be large enough to travel across the Pacific and strike Japan. The wave that washed over the island of Honshu in January 1700 may have had such a distant origin. (right) NOOTKA of Vancouver Island and other Native American tribes of the region were susceptible to tsunami disasters.
(left) TSUNAMIS generated from giant earthquakes in western North America would be large enough to travel across the Pacific and strike Japan. The wave that washed over the island of Honshu in January 1700 may have had such a distant origin. (right) NOOTKA of Vancouver Island and other Native American tribes of the region were susceptible to tsunami disasters.
A clever strategy may pinpoint the time of the most recent earthquake even more precisely. Tsunamis generated by Cascadia earthquakes with magnitudes near 9 should be large enough to be noticed in Japan even after traveling across the Pacific Ocean. Recognizing this fact, Kenji Satake and his colleagues at the Geological Survey of Japan think they have found the written record: a two-meter-high tsunami that washed onto the coast of Honshu nearly 300 years ago. After correcting for the time the wave would have taken to travel to Japan (and the time zone change), Satake determined that the earthquake occurred along the North American coast on January 26, 1700, at about 9 P.M.
Remarkably, that detective work agrees with reports of a disaster preserved in the oral history of the original residents of British Columbia. My colleague Rogers found what may be a description of this event in the provincial archives in Victoria. Native tradition records that an earthquake struck Pachena Bay on the west coast of Vancouver Island one winter night; in the morning the village at the head of the bay was gone. Gary A. Carver of Humboldt State University uncovered a similar account m the unwritten lore of northernmost California. Thus, native stories, Japanese writings and sedimentary deposits all point to the inevitable conclusion that giant earthquakes do in fact haunt the Cascadia coast.
FLEXING like a board bent over the edge of a table, the North American plate develops an inland bulge as its western margin is pulled downward by the oceanic slab (top). After an earthquake releases the stress, the bulge collapses (middle), forcing much of the coastal region to subside and fill with sediment (bottom).
The Great Earthquake Cycle
FLEXING like a board bent over the edge of a table, the North American plate develops an inland bulge as its western margin is pulled downward by the oceanic slab (top). After an earthquake releases the stress, the bulge collapses (middle), forcing much of the coastal region to subside and fill with sediment (bottom).
Like all earthquakes, large subduction zone events prove complex when considered in detail. The basic process, however, follows the simple "elastic rebound" theory first developed for the notorious San Andreas fault in California. According to this concept, ongoing movement between two plates compresses and bends the crust as stress accumulates. Contrary to the illusion of the earth being made of rigid and solid rock, the contraction is nearly elastic. If not squeezed too much, the earth acts like a gigantic piece of rubber. Eventually, however, the tectonic forces become so extreme that they exceed the hold of friction along the fault. The surface slips abruptly, and the elastic energy that was stored over many years radiates outward as ground-shaking earthquake waves. The fault then locks once more, and the cycle of tectonic stress buildup and release resumes.
Along the Cascadia subduction zone, the oceanic Juan de Fuca plate encroaches on North America by about 40 millimeters a year. This progress may seem slow, but it represents a considerable shortening--about 20 meters in a typical 500-year-long stretch between giant earthquakes. The motion is taken up by elastic shortening distributed across a swath several hundred kilometers wide. But the tectonic stresses cause more than just horizontal contraction--the ground moves vertically, too. As the oceanic plate dives under the coast, it drags the seaward nose of the continent downward and causes parts of the North American plate further inland to flex upward; this process mimics the bending of a long board over the edge of a table--as the front is forced down, a bulge forms behind. When a large earthquake breaks the locked fault, the seaward part of the continent springs back, and the bulge collapses. The abrupt rise of the outer continental shelf generates tsunamis, and the sudden fall of the "flexural bulge" centered near the coast causes the drop that buries intertidal salt marshes.
The positron of the locked zone proves especially important because this surface becomes the source of seismic-wave energy when the fault eventually gives way in an earthquake. The landward limit of the earthquake source zone affects how closely the earthquake will impinge on the larger population centers; the seaward edge controls where tsunamis will develop. The total width of the source zone influences the seismic hazard because it sets the maximum size of the earthquake.
Scientists can determine the extent of the locked zone from the form of crustal deformation. If the locked zone is narrow, extending only a short distance down the inclined fault, the region of elastic bending will also be quite restricted. Conversely, if the locked zone runs appreciably farther down, the bending deformation will reach a long distance inland. Land surveying can thus help to map out the earthquake hazard. The rates of deformation are only a few millimeters a year, but they can be resolved with modem surveying techniques if the measurements are applied with exceptional care.
DEFORMING CRUST changes shape extremely slowly, but careful measurements can track the subtle movements of the ground. Using a laser range finder between mountaintops (left), scientists were able to measure horizontal contraction in the region more than a decade ago. But modern instruments that use the Global Positioning System of satellites (right) have eased the task of conducting precision surveys.
Watching the Strain Build Up
DEFORMING CRUST changes shape extremely slowly, but careful measurements can track the subtle movements of the ground. Using a laser range finder between mountaintops (left), scientists were able to measure horizontal contraction in the region more than a decade ago. But modern instruments that use the Global Positioning System of satellites (right) have eased the task of conducting precision surveys.
Several different kinds of observations repeated over time define how the Cascadia margin is currently deforming. Geophysicists can follow the horizontal shortening of the coastal region by measuring, for example, the distance between surveyors' benchmarks on mountaintops using a laser ranging device. This feat requires a good deal of care and plenty of clear sky (not common in the rainy West Coast mountains). Using this technique, James C. Savage and his colleagues at the U.S. Geological Survey first reported in 1981 that the crust near Seattle was shortening perpendicular to the coast. They concluded that strain was building toward an appreciable earthquake.
COASTAL UPLIFT in Cascadia is documented by repeated leveling surveys (dots). The data match theoretical predictions (solid line) for the bending of the North American plate and serve to locate the locked part of the subduction fault.
Some survey methods are sufficiently sensitive to vertical motion. The most simple of these, known as leveling, employs the same technique one sees being used along highways. Surveyors take sightings on calibrated rods to measure the difference in elevation between two places. By combining the measured offsets, surveyors can determine relative heights throughout a network of connected points spread over large distances. Repeated surveys after several years yield the uplift or subsidence of one position with respect to another. The Geodetic Survey Canada has, for instance, carried out several surveys of exceptional accuracy specifically to study earthquake-related uplift. One of these field experiments tracked back and forth across the width of Vancouver Island (about 100 kilometers each way) in a series of sightings, each of 100 meters or so. The complete circuit had a total vertical error limited to only one centimeter.
Another method makes use of tide gauges that track the level of the sea relative to coastal bedrock. The primary purpose of these devices is to monitor the ocean, but surprisingly, with gauges that have been recording for 20 years or more, it is possible to use the average level of the sea surface as a reference and to trace subtle vertical shifts of the land. The record must, of course, be long enough to smooth over tides and other oceanographic variations, such as El Nino, that can endure for years. One also needs to account for the steady global rise in sea level (of about two millimeters per year) and to correct for postglacial rebound, the slow but continuing rise of crust initially pushed downward under the weight of glaciers from the last ice age.
Yet a third way to detect vertical Motion is by using gravity, a force that varies with the square of the distance from the center of the earth. Although it is impossible for people to sense the slight shifts in their weight when they change altitude, sensitive instruments can register these variations. By repeatedly measuring gravity at one spot every few years, geophysicists have been able to estimate the rate of coastal uplift.
During the past few years, the satellite-based Global Positioning System (GPS) has permitted scientists to measure distances and vertical offset between sites spaced hundreds of kilometers apart. Herb Dragert and Michael Schmidt of the Geological Survey of Canada have used GPS to show that every year coastal Victoria shifts nearly a centimeter closer to Penticton (a locale some 300 kilometers inland). GPS is accurate and inexpensive and in the future may prove the most effective technique for keeping track of the subtle bending and squeezing of the earth's crust that leads to earthquakes.
All these methods give similar results: the Cascadia margin currently rises by one to four millimeters a year, and it also contracts horizontally by several centimeters every year. This deformation--direct evidence that the crust is being squeezed between converging plates--registers the slow but relentless accumulation of strain that is building toward the next catastrophic release.
CONVERGING PLATES are locked together over a confined region of the thrust fault under the coast. The extent of this locked zone is limited on its western side because clays deposited on the surface of the downgoing oceanic crust help to lubricate the fault. To the east, the locked zone gradually fades to the point where the deeply buried fault slides freely because of elevated temperatures.
A Troublemaker Locked Up
CONVERGING PLATES are locked together over a confined region of the thrust fault under the coast. The extent of this locked zone is limited on its western side because clays deposited on the surface of the downgoing oceanic crust help to lubricate the fault. To the east, the locked zone gradually fades to the point where the deeply buried fault slides freely because of elevated temperatures.
Because knowing the position of the locked part of a fault is so critical to defining earthquake risk, my colleagues at the Pacific Geoscience Center and I have tried to determine the extent of the locked zone by comparing survey measurements with mathematical models of the deformation. Fitting the observations to theory allowed us to map the width of the locked zone deep below the earth's surface. The actual situation is somewhat more complex than this simple conceptualization: at the deeper, landward boundary of the locked zone, there is a gradual transition between areas that are rigidly locked and those that are completely free-sliding.
The comparison between our data and models shows that for most of the Cascadia coast the locked zone is restricted to a swath 50 to 100 kilometers across that runs underneath the continental shelf. (It widens considerably only near the coast of northern Washington.) This surface represents a huge fault area with potential for enormous earthquakes. Yet, curiously, it is unusually narrow compared with other subduction zones.
Such differences prompted us to examine which attributes of the geology influence the width of the locked zone. Many factors may contribute, but temperature plays a dominant role. For example, the clay-rich sediments that blanket the oceanic plate may act to lubricate the seaward, "up-dip" edge of the fault; as the ooze becomes more deeply buried, however, the clays chemically alter into stronger minerals that prevent the fault from sliding. This change happens at a depth of 10 kilometers or so, where the temperature reaches about 150 degrees Celsius.
Temperature also appears to control the landward, "down-dip" limit of the locked zone. At moderate temperatures the rocks show normal frictional behavior. That is, the large initial resistance to motion drops to a lesser level once the fault begins to slip. So, once sliding starts, runaway release of the stored elastic energy--an earthquake--ensues. But lower in the crust, where the temperature exceeds about 350 degrees C, the rocks lining the fault surface should behave more like a viscous fluid--faster motion meets with increasing resistance. Hence, the deeper, hotter parts of the fault are apt to creep along slowly without generating any seismic waves.
How well do these temperature limits correspond to the actual boundaries of the locked zone? My colleagues and I have tried to provide the answer by using our computer models to calculate temperatures on the subduction thrust fault. The results of that work confirmed what we had surmised. The depth on the fault where the rocks reach 3 50 degrees C agrees well with the down-dip limit of the locked zone as determined from the measurements of deformation.
One important question remains: Does the locked zone that we have calculated truly correspond to the source area for large subduction earthquakes? We believe it does because the vertical drop we predict for a rupture of the locked zone matches what has been observed in the buried coastal marshes. Further support comes from our efforts to apply the same techniques to other subducting margins. My colleague Kelin Wang and I, along with Makoto Yamano of the University of Tokyo, have shown that the width of the present locked zone on the Nankai margin of southwest Japan corresponds well to the rupture areas of the magnitude 8 earthquakes that struck there in the 1940s. So we can be confident that our models for Cascadia are telling us about the kind of devastating earthquakes that will eventually strike along the North American coast.
When the Big One Hits
How intensely would the ground shake at the major West Coast cities during a giant subduction earthquake? The answer rests on the exact earthquake magnitude as well as on the position of the seismic source zone. The maximum magnitude that a Cascadia earthquake could achieve depends on just how far along the coast the fault releases. Simultaneous rupture of the entire stretch from British Columbia to California would be surprising because such extended breaks have been rare anywhere in the world. Yet some evidence points to a failure of this size having occurred during the Cascadia earthquake of 1700. If the total locked zone (an area of nearly 100,000 square kilometers) releases at once, a giant earthquake of magnitude 9 could result--much larger, for example, than the catastrophic San Francisco earthquake of 1906. There have been only two events of this size ever recorded: an earthquake along the coast of Chile in 1960 and the other in southern Alaska in 1964.
Seismologists can estimate the amount of ground motion that might come from such Cascadia earthquakes in two ways. One approach is to compare the situation along North America's western coast with earthquakes that have occurred elsewhere. The other method is to use rather complicated theoretical models of the seismic rupture area and slip displacement. Either way the conclusions are similar. The next great earthquake in Cascadia will generate extremely large seismic waves lasting for as long as several minutes. After the shaking ceases, most coastal sites will be one to two meters lower and five to 10 meters seaward of where they started.
Fortunately, the locked part of the fault that would generate such earthquakes lies primarily under the continental shelf and extends little, if at all, below the coast. Hence, Vancouver, Seattle and Portland (which sit 100 to 200 kilometers inland) are subject to less severe shaking than sites near the outer western coast. Nevertheless, the seismic energy from such violent earthquakes radiates for a considerable distance, so the danger to those cities is still substantial. U.S. and Canadian residents of this Pacific coastal region who had imagined they lived on quiet ground will indeed have to learn to accept the threat of a giant earthquake upheaval happening at any moment.
Historical Precedents
The native Yurok people who occupied the coastal region personified natural powers such as earthquakes and thunder in their lore. This excerpt from an interview recorded by A. L. Kroeber in Yurok Myths describes what perhaps were relatively recent earthquakes:
And from there [Earthquake and Thunder] went south .... They went south first and sank the ground.... Every little while there would be an earthquake, then another earthquake, and another earthquake.... And then the water would fill those [depressed] places.... "That is what human beings will thrive on," said Earthquake. "For they would have no subsistence if there were nothing for the creatures [of the sea] to live in. For that is where they will obtain what they will subsist on, when this prairie has become water, this stretch that was prairie: there will be ocean there."..."Yes, that is true. That is true. That is how they will subsist," said Thunder. "Now go north." Then they went north together and did the same: they kept sinking the ground. The earth would quake and quake and quake again. And the water was flowing all over.
No comments