The Next Wave
It was an idyllic setting: a palm-fringed finger of sand on the northern coast of Papua New Guinea. To one side lay the tropical waters of the Bismarck Sea; to the other, the fish-laden shallows of Sissano Lagoon. Beyond the lagoon stood a verdant, steamy tropical forest. About 15,000 people lived here in the small villages of Sissano, Warapu, Arop, and Malol, many in simple dwellings on the spit itself.
The evening of Friday, July 17, 1998, marked the beginning of a four-day holiday. Children were playing touch rugby and preparing for the evening festivities. Villagers were relaxing, strolling along the beach, or visiting friends and relatives.
Shortly before sunset, the earth began to shake, and a thunderous boom shattered the air. Large cracks suddenly gaped open on the beach, and the village children, filled with curiosity, ran to look, encouraging the grown-ups to come and see, too. As a crowd gathered, the ocean began to recede from the shore, and, as the water drew farther out, a distant murmur grew to a rumbling.
Within a few minutes the rumbling had grown to a deafening roar, like the sound of an approaching jet squadron. The sea had returned as a wall of water, glowing red near the crest. As the watchers turned to flee, the wave broke offshore, surging through the villages and knocking people off their feet. But this was prologue. On the heels of the first, a second monstrous wave swept in. This one crashed directly over Warapu and Arop, engulfing the villagers as they raced inland, and drove landward for hundreds of meters. The dead and brutally injured were dumped into the lagoon and surrounding mangroves. When the wave retreated, it pulled many of the victims out to sea.
As darkness fell, yet another huge wave struck the spit. When the tsunami was over, half of Sissano and Malol were obliterated, and where the houses of Arop and Warapu had stood, nothing remained but cement slabs. The dark swirling waters, churning debris, and treacherous currents had claimed over 2,000 lives.
In the ensuing hours and days, the stunned villagers tried to make sense of what had happened. As the waves crashed down, they had been lit, as if on fire; and the bodies of many victims seemed burned. The earth was still shaking intermittently. Would more waves come? Were the people being punished?
Thousands of miles away, earthquake and tsunami scientists were also looking for answers. Within hours of the earthquake, three seismic groups had analyzed data from the worldwide seismometer network and posted the earthquake's 7.1 moment magnitude via the Internet to the scientific community. Now information about the tsunami was coming in. The waves were massive--reportedly over ten meters high in some locations. Yet at magnitude 7.1, the Papua New Guinea earthquake was only moderate. How could it have caused so large a tsunami?
Tsunamis are always a possibility in the Earth's largest ocean. While the coastal communities that circumscribe the Pacific have hearthside seats at the Ring of Fire, where Earth's greatest upheavals occur, some locations are especially susceptible. The Hawaiian Islands, in particular, lie directly in the path of tsunamis created on most of the Pacific Ocean subduction zones.
The term "tsunami," adopted for general international use in 1963, is a Japanese word, represented by the characters tsu, meaning "harbor" and nami, meaning "wave." Most tsunamis are caused by underwater earthquakes in which fault slippage moves huge segments of oceanic crust vertically, displacing the water column over vast areas. Picture two adjacent blocks of the Earth's crust with the fault separating them as a tilted plane: one block rests above the fault plane, the other below it. In reverse faults, the overlying block moves upwards. It is on gently dipping reverse faults--or thrust faults--that tsunamigenic earthquakes usually occur.
"The waveform and height of tsunami waves at their point of origin is related to the inclination of the fault plane and the amount of vertical displacement on the fault," says geophysicist Eric Geist, a tsunami scientist at the U.S. Geological Survey (USGS) in Menlo Park, California. "Off the coast in deep water, the initial height may be only a meter or two, even for major tsunamis. But as they travel from deep to shoaling waters, the waves slow down, their wavelength shortens, and their height increases dramatically."
As scientists examined bathymetric maps of the Bismarck Sea, they turned up another possible clue to the massive wave size of the Papua New Guinea (PNG) tsunami. Off the northern coast, the depth of the sea floor drops precipitously. An extreme deep-water epicenter would mean greater amplification of the waves as they traveled toward shore. On the other hand, if all of the fault slippage had occurred on land or in shallow water near shore, other factors, such as underwater landslides, were surely involved.
Finding out the details wouldn't be easy. Many of the stations in the worldwide seismic net are in a so-called "shadow zone" for the PNG region. Reception of some earthquake waves was interrupted by the Earth's internal structure, explains David Oppenheimer, seismologist at the USGS. As a result, fewer than the usual number of seismic reports were available to determine the earthquake's location. The USGS placed the epicenter at the coast; teams from Harvard University and the Earthquake Research Institute in Japan calculated it at 50 and 75 kilometers offshore, respectively.
Though the initial data did not reveal the exact epicenter, the aftershock data could, explains Oppenheimer. Aftershocks are usually centered around the focus of the initial earthquake. It was essential to install local seismographs while aftershocks were still strong and numerous.
Such a seismic net might be able to solve another part of the puzzle as well. Determining the dip, or inclination, of the earthquake fault plane was crucial to understanding why the tsunami was so large. Initial analysis of earthquake seismic data offered scientists two possible solutions: one was a near-vertical reverse fault; the other an almost-horizontal thrust fault. Assuming an offshore fault, a vertical slip would create a higher initial wave. A near-horizontal slip, on the other hand, would argue against the earthquake as an efficient tsunami generator, suggesting that another mechanism was involved.
Also critical was documenting the physical evidence of the tsunami's character--the area inundated, wave heights, current velocities, sand deposits. Many of the clues, such as water marks and debris caught in treetops, wouldn't last long. The scientists also hoped to interview survivors.
A 13-member International Tsunami Survey Team (ITST), with scientists from Australia, Japan, New Zealand, and the United States, traveled to PNG two weeks after the earthquake, to join local scientists to assess the tsunami. The team found that the brunt of the tsunami extended along 40 kilometers of coastline, about the length of fault rupture expected for a 7.1 magnitude earthquake, according to team leader Costas Synolakis of the University of Southern California. In the central 20 kilometers of inundation, the waves peaked at 15 meters, the height of a four-story building, but elsewhere wave heights dropped rapidly. Residents only 15 kilometers from the area hit hardest scarcely felt any earthquake tremors.
The ITST also kept an eye out for evidence of uplift or subsidence onshore. They didn't find any, but they did find an earthquake-caused landslide, leading seismologist Emile Okal, a team member from Northwestern University, to postulate that a larger underwater version of the slide could have been responsible for the tsunami.
During a series of presentations explaining the nature of tsunamis to the local populace, the visiting scientists clarified the reason why the waves seemed to be on fire: they were full of dinoflagellates, bioluminescent microorganisms excited by the turbulence. And with the tsunami moving three times faster than the fastest flood waters, victims had been burned not by flaming waves but by friction as they were dragged across the sand.
Although several types of offshore earthquakes can cause tsunamis, the majority of tsunamigenic earthquakes occur on thrust faults in subduction zones. Consisting of solid, rigid rock, tectonic plates are blocks of the Earth's outermost layer called the lithosphere. Oceanic plates are transported outwards from mid-oceanic spreading zones, moving over layers of underlying softer rock; at oceanic margins they are pulled by gravity under continental plates at oblique angles, a process known as subduction. Beneath continental fringes and island arcs, buried oceanic crust melds with the Earth's mantle, venting on the surface as volcanoes that compose the Ring of Fire. Paralleling the coast, offshore trenches (in some cases, sediment-filled) trace the massive underwater thrust faults on which subduction occurs.
Slippage within the upper portions of these megathrust faults--where movement causes uplift and subsidence of large sections of the ocean floor--can generate gigantic tsunamis. Though a tsunami travels outwards in all directions from its source, the deformation along the fault trace on the ocean floor is linear; a tsunami, in essence, separates into a pair of waves. One descends rapidly on the nearby shore causing a "local tsunami" while the other speeds out to sea, arriving in far away locations as a "distant tsunami."
As each of the twin waves travels, it disperses into a series of component waves, just as a pebble thrown into a pond creates a succession of rings. In local tsunamis, the distance traveled is short, and the waves don't disperse much. However, when the tsunami crosses the Pacific, its component waves may arrive as much as an hour apart. These waves interfere with each other in complicated ways, sometimes combining to create even higher run-ups.
Local tsunamis, traveling directly to land, arrive within an hour, sometimes in as little as three minutes, and they may be huge, in rare cases over 30 meters. They often arrive too quickly for formal warning systems to be effective. The only notice might be the preceding earthquake. In the other direction, tsunamis can travel vast distances, losing little of their energy in route. Arriving on far shores, they may be as high as 15 meters. Prior to modern warning systems, distant tsunamis arrived totally out of the blue.
This century's largest earthquake, moment magnitude 9.6, occurred in 1960 on the subduction zone offshore of Chile near Concepcion. Vertical displacement occurred over an area approximately 920 kilometers long and 300 kilometers wide, with a maximum slippage of 54 meters. Here is one account of the local tsunami coming ashore:
Then, suddenly, they noted that the sea was beginning to retreat from the shores, exposing the ocean floor to distances well beyond the lowest tides. When this happened, the fire alarms were sounded, and firemen and carabineros systematically went through the streets warning everyone of the impending danger. The people fled afoot and on horseback to the hills and waited. Those on horseback made repeated trips to save the old and infirm. After 15 to 30 minutes, the sea returned, advancing upon the shore in a wave that was in places over 20 feet high. The wave rushed over the land, covering and carrying away the houses, killing the animals that could not be evacuated, and carrying off some of the people who, for one reason or another, had not left their homes...In several villages along the southern coast, such as Carelmapu...the mariscadores, or shellfish collectors, took advantage of the recession of the sea to wander over the exposed sea floor collecting shellfish in their baskets. When they had collected more than the usual quantity of mussels and locos, they returned to the shore, climbed up on the hills, and waited for the water to return. The waves continued all afternoon...A group of mapuchis or Araucanian Indians sacrificed a seven-year-old boy to the gods of the sea to calm the remorseless surf.
--P. Saint-Amand, 1961
Los Terremotos De Mayo-Chile
About 15 hours and 6,800 kilometers later, the outbound portion of the tsunami arrived in Hilo, Hawaii, where a ten-meter wave stopped clocks five minutes after one o'clock in the morning. Sixty-one Hilo residents lost their lives. The following story describes the aftermath:
At dawn my Grandpa, Dad, and Uncle Harold went by car to a spot about where Burger King is today. You could not drive any closer to the Bay because there was a 20-foot pile of rubbish all along Hilo Bay. They climbed up the pile and looked over and saw only a vast open space where formerly the houses and businesses of downtown Hilo had been...
They walked to the location of the family business and all that was left was a flat cement slab that looked freshly poured. The wave had sandblasted off every speck of oil and grease. There was not any sign of the buildings or their contents. Everything had been destroyed. They found the store's 2,000-pound safe on the bayfront by where the Ironworks building is today. Uncle Harold helped my Great Grandpa clean out the safe. Dad says Uncle Harold remembers his job was to iron the money to dry and save it...
--from "My Dad and the 1960 Tsunami: The story of Tom Goya" as interviewed by Issac Goya, an essay submitted to the Pacific Tsunami Museum 1998 essay contest.
And 23 hours and 17,000 kilometers later, the great waves, still over six meters high, reached Japan's Sanriku coast, half way around the world from where they were generated. Work by Kenji Satake of the Geological Survey of Japan has shown that variations in oceanic depths, which refracted the propagating wavefronts, focused the tsunami's fury on Japan. Thousands of structures were wrecked or washed away; about 200 people lost their lives. This same coast has been struck numerous times by tsunamis.
While warning centers now exist to detect distant tsunamis generated in the Pacific, current scientific understanding of tsunamis and available technology permit only limited prediction. Warning centers can determine whether an earthquake is capable of producing a tsunami and when such a tsunami will arrive, but they can't say whether it will be a six-centimeter bulge or a ten-meter cataclysm. A multitude of factors, such as bathymetry, absence or presence of coral reefs, the shape of the shoreline, steepness of the beach, and distance from the origin, play a role in the final presentation.
A tsunami disguises itself so well in deep water that it is imperceptible, both from airplanes and to ships at sea. Unlike wind-generated waves, the pulse of displaced water in a tsunami wave extends all the way to the bottom of the ocean. As tsunami waves travel across the ocean basin at speeds about 750 kilometers per hour, the wavelength, from crest to crest may spread up to 400 kilometers, while the amplitude from trough to crest is usually under two meters. Only as the sea becomes shallower towards shore, does the wavelength shorten and the tsunami grow destructive. In his 1975 article on tsunamis in Oceans, Michael J. Mooney describes the experience of fishermen out at sea when the great 1896 Meiji Sanriku tsunami struck Japan, causing 27,000 deaths, injuring 9,000, and sweeping away over 10,000 houses:
Earlier that evening, local fishermen put out to sea for their nightly harvest. Twenty miles offshore, the incoming waves passed unnoticed beneath their keels owing to the insulating depths of the sea. On returning home the following morning, they came upon heaving carpets of debris and corpses floating a short distance offshore. On land they found desolation and destruction.
Walter Dudley, author of the book Tsunami!, teaches students about tsunamis at the University of Hawaii, Hilo. Interviewed in the National Geographic film "Killer Waves," he observes that the picture people have of tsunamis is "one giant wave that always seems to leap up in the middle of the night like surf over a sleeping village." Perhaps what they imagine is the big, curling, tendril-bearing wave in the famous nineteenth-century print by Japanese artist Katsushika Hokusai (although the artist never intended it to represent a tsunami).
On the contrary, tsunamis are actually trains of up to ten or more waves. In contrast to wind-generated waves, in which water is momentarily displaced vertically, tsunami waves actually transport water forwards and backwards. And, although the wave is shortened from its deep-water wavelength, as it approaches shore, it still extends several kilometers crest to crest, says wave mechanics expert Synolakis. Photographs of tsunamis coming in sometimes look like an approaching giant stairstep that stretches as far as the eye can see. Eyewitnesses often speak of water that just kept on coming.
While the characteristic tsunami wave appears as an extreme and rapid tide, other wave forms can occur, like the breaking second wave in the PNG tsunami. Such waves create considerably more turbulence and damage. Particularly fearsome is the bore, a vertical wave of tumbling water.
When early waves arrive gently, residents are often lulled into believing that the tsunami is nothing to worry about. They may even venture to the shore to watch subsequent waves come in, losing their lives to later waves that are as much as ten times larger than the initial ones. The first two tsunami waves from the 1964 Good Friday earthquake in Alaska reached Crescent City, California, as small surges causing minor flooding. After returning with friends to retrieve his cash, a tavern owner dallied too long over drinks. They were caught by a seven-meter-high monster.
A wild card in the barrage of waves is a phenomenon known as "trapped edge waves." "A wave coming into shore can give rise to component waves," explains George F. Carrier, professor in the Department of Engineering and Applied Sciences at Harvard University.
As these propagate along the shore, they bounce off and refract around headlands and seastacks. Sometimes edge waves add together and become larger than the direct waves. A study by Frank Gonzolez, of NOAA's Pacific Marine Environmental Laboratory in Seattle, and colleagues shows that the largest observed waves arriving at Crescent City after the April 25, 1992 Cape Mendocino earthquake were edge waves that arrived almost three hours after the onset of the tsunami.
Though tsunamis are usually generated by underwater earthquakes, they may be caused by any massive disturbance of the water column, either from above or below. (Although older accounts often refer to tsunamis as "tidal waves," one natural phenomenon that doesn't cause tsunamis is the tides.) They may, for instance, be caused by large volcanic eruptions, like the infamous August 27, 1883, eruption of Krakatau in the East Indies. The sudden collapse of the 2,000-meter high volcano triggered a powerful tsunami that killed over 36,000, and left no trace of 165 coastal villages. In rare instances, the impact of celestial objects such as meteors can cause tsunamis.
Two types of landslides can generate tsunamis that, although localized, can be immense. Subaerial slides begin on land, but their debris can impact the water. Submarine slides begin and end completely under water. The largest tsunami-like wave on record occurred on July 10, 1958, in Lituya Bay, Alaska, following a local earthquake, which loosened 90 million tons of rock at one end of the Bay. The landslide caused a surge that rose to heights of 500 meters on the opposite shore.
Submarine landslides not associated with earthquakes can be particularly insidious, generating tsunamis that arrive with no warning. Such slides could occur, for example, in the steep-walled submarine canyons common off the California coast.
The 1964 moment magnitude 9.2 Good Friday earthquake in Prince William Sound was so intimidating that few Alaskans focused on the impending danger gathering offshore. Uplift and subsidence had occurred over a swath of almost 200,000 square kilometers, much of it ocean floor, giving birth to a train of high-energy tsunami waves just off the Alaska Coast. Yet before the train arrived, earthquake-triggered submarine landslides caused additional tsunamis, demolishing the Valdez waterfront within three minutes of the earthquake's onset. At nearby Whittier, landslides produced a 30-meter tsunami, destroying sawmills and the railroad depot, while at Seward, a landslide-generated ten-meter wave was topped with burning oil from ruptured petroleum tanks. The major wave train followed soon afterward, surging ashore, and spreading the fire.
In 1992, when a ten-meter tsunami struck El Transito, Nicaragua, residents were caught completely unaware. Many had felt only feeble shaking; others did not notice the earthquake at all--even though they were less than 100 kilometers from its epicenter. This illustrates another complicating factor in tsunami prediction: so-called "tsunami earthquakes," first identified by preeminent seismologist Hiroo Kanamori of the California Institute of Technology. While any earthquake that causes a tsunami is a "tsunamigenic" earthquake, a "tsunami earthquake," according to Kanamori, is one that generates a tsunami many times larger than would be expected from the amount of shaking.
In some tsunami earthquakes, the fault ruptures relatively slowly. Unlike typical earthquakes, in "slow" earthquakes, there is a large differential between the surface wave magnitude (Ms), and the moment magnitude (Mw), (see box page 32). Slow tsunami earthquakes tend to be shallow, beginning less than 15 kilometers below the seafloor. Typically, the earthquakes are in very deep water at or near the bottom of a subduction zone trench.
One of the most infamous slow tsunami earthquakes occurred in the Aleutian Trench on April Fool's Day, 1946, at 1:30 a.m. Less than an hour later, a 30-meter wave engulfed the five-story reinforced-concrete Scotch Cap Lighthouse on Unimak Island, killing the five men inside. Meanwhile, the distant tsunami was fanning out across the Pacific. Traveling at the speed of a jet liner, it reached Hawaii in a little less than five hours, inundating shores on all of the islands with waves as high as 15 meters.
Could a slow earthquake have caused the anomalous size of the PNG tsunami? Work by Emile Okal and members of his research group showed that although the PNG earthquake had some slow features, it did not completely fit the slow earthquake profile.
By late fall of 1998, scientists from Australia and New Guinea had recovered temporary seismic instruments and determined that the epicenter of the PNG earthquake was approximately 25 kilometers directly offshore of Sissano, where the highest waves had been recorded. Aftershock data failed to clearly determine the fault plane, but data from tide gauges in Japan showed sea level variation consistent with the waveform of a tsunami generated by near-vertical faulting.
"The data so far suggest three contributing factors to the size of the tsunami," explains Geist. "Even though the origin was only 25 kilometers offshore, the tsunami was generated in very deep water--about 2,500 meters--and was greatly amplified as it traveled to coastal waters. Second, the origin was close to shore, so the wave lost little energy on the way in. The third factor is the probable near-vertical faulting, which results in a shore-bound wave with a steep leading face. The steepness of the wave has a big effect on the run-up height."
"However," Geist notes, "a triggered submarine landslide may well have amplified the earthquake-induced tsunami." In a recent issue of Science News, Okal explained that the sediments piled on the steeply plunging shelf could have easily cascaded downslope after the shaking, and a recent research cruise has revealed evidence of an underwater slide.
In the meantime, a second ITST mission found evidence of a prior tsunami. "At approximately 120 centimeters below the surface, a thin layer of coarse silt and fine sand was observed," says Bruce Jaffe, an oceanographer at the USGS. Along four transects, team members examined newly laid deposits, recording their character and thickness, along with land elevation, and water depth and flow direction.
"Refining the ability to read tsunami deposits to infer the height, power, and extent of a tsunami is not only valuable for understanding the July 17 event, but also for understanding prehistoric tsunamis worldwide," explains Jaffe.
Tsunami sand deposits are a major reason why scientists have been prompted to reassess tsunami risk in the Pacific Northwest.
Historically, there is no record of megathrust earthquakes in the 1,000-kilometer long Cascadia subduction zone, which begins north of Cape Mendocino and continues offshore up to northern Vancouver Island. This inactivity deceived scientists into thinking there was little danger from subduction-related thrust earthquakes and related local tsunamis.
However, concrete evidence of a potential hazard appeared on April 25, 1992 in the form of a moment magnitude 7.1 earthquake and accompanying tsunami. Subsequently, scientists, including David Oppenheimer of the USGS and Eddie Bernard of NOAA's Pacific Marine Environmental Laboratory, surmised that a shallow thrust earthquake on the entire Cascadia subduction zone could generate a tsunami that, minutes after the main shock, would inundate many coastal communities in the Pacific Northwest, and could persist for as long as eight hours.
A few years earlier, Brian Atwater of the USGS and others had discovered the remains of large coastal forests in Washington, Oregon, and northern California. The remains were reminiscent of forests on the Chilean and Alaskan coasts that had subsided following large earthquakes and flooded with saltwater. Sand-filled cracks, a feature of liquefaction, also implied that the sudden coastal subsidence happened during a strong earthquake. And scientists found key evidence of past tsunamis: sand layers on buried marsh and forest soils at estuaries between British Columbia and California.
Meanwhile, tsunami researcher Kenji Satake not only found evidence that a major earthquake had occurred in Cascadia, but was also able to determine the date, time, and approximate magnitude. Satake searched Japanese historical records and discovered that a tsunami of unknown origin had hit Honshu Island on January 27, 1700. Based on the size and arrival time of the Japanese tsunami, Satake estimated that a Cascadia earthquake with an approximate moment magnitude of 9.0 had occurred at 9 p.m.
Some of the most recent work by Atwater, dendrochronologist David Yamaguchi at the University of Washington, and their colleagues used rings in the roots of a dozen cedar stumps from the Copalis River south to the Columbia. The rings revealed that the trees had died after the autumn of 1699 and before the start of the 1700 growing season. Atwater says the latest dates clinch the argument that massive earthquakes occur on the Cascadia Fault. The dates concur exactly with those found by Satake.
Painstaking new work by micropaleontologist Eileen Hemphill-Haley of the USGS has pushed the record of Cascadia tsunamis further back. In freshwater Bradley Lake, near Cape Blanco, Oregon, she has found "a catch-basin for tsunami records." Diatoms in core samples going back 5,500 years show repeated saltwater intrusions and suggest 14 tsunami inundations.
The oral tradition of Pacific Northwest indigenous peoples also contains stories of tsunamis. This account about the Pachena Bay people on Vancouver Island was told to Eugene Arima of Parks Canada by Louis Clamhouse in 1964.
This story is about the first Anaql'a or "Pachena Bay" people. It is said that they were a big band at the time of him whose name was Hayoqwis'is, "Ten-On-Head-On-Beach."... I think they numbered over a hundred persons.... There is now no one left alive due to what this land does at times. They had no way or time to try to save themselves. I think it was at nighttime that the land shook.... They simply had no time to get hold of canoes, no time to get awake. They sank at once, were all drowned; not one survived.... I think a big wave smashed into the beach.
From the diverse lines of evidence, earth scientists generally agree that a major tsunamigenic earthquake is a potential hazard in the Pacific Northwest. And a further possibility is the unannounced arrival, a number of hours after the primary tsunami waves subside, of even larger edge waves.
How would a tsunami from a Cascadia earthquake compare with the recent tragedy in Papua New Guinea? Planning for tsunamis is based on a worst-case scenario; in Cascadia, this would be a rupture comparable in length to the one off Chile in 1960. According to Lori Dengler, professor of geology at Humboldt State University and director of the Earthquake Education Center, such a tsunami in Cascadia would inundate a coastline hundreds of kilometers long, and would continue well beyond the one-hour event in PNG. The earthquake would also cause significantly more shaking, so the tsunami would be advancing on a coast already severely damaged. Moreover, the distant tsunami would strike shores throughout the Pacific.
The five to ten minutes between the PNG earthquake and the onslaught of the first waves gave experienced elders the chance to escape the inundation. In the northern part of Cascadia the interlude could be similar, but near Cape Mendocino, where the megathrust fault comes ashore, the earthquake and the tsunami would occur almost simultaneously.
"A Cascadia event is capable of producing wave heights at least as robust as Papua New Guinea, and in some locations it could be significantly worse," says Dengler. "If you feel an earthquake, get off the beach and find out about it later!"
Anne M. Rosenthal is a freelance science writer based in the San Francisco Bay Area. She profiled scientist Barbara Block in the Winter 1998 issue of California Wild.