Three huge waves crashed over tens of kilometers of Papua New Guinea's shoreline
on July 17, 1998. More than 2,000 people died, swept away by the 7- to 10-meter
waves. Two villages disappeared.
Initially, scientists thought the tsunami had been created by an offshore earthquake, a magnitude-7.0 event, but the lagtime between the shaking and the waves made geologists take another look. They found the waves were too large to fit the earthquake that seemed to have produced them. Instead, they hypothesized, a slumping landmass could have generated the higher and more focused waves, which hit a discrete section of the Papua New Guinea shore. Underwater surveys showed a large slumping bowl tens of kilometers offshore a landslide triggered by the earthquake, perfectly positioned to send waves caroming inland.
The ensuing controversy over whether an earthquake or an underwater landslide generated the Papua New Guinea tsunami rekindled interest in such events; it also drew geologists into a field that had been dominated by modelers and seismologists.
"Tsunami science used to be considered part of seismology," says Philip Watts, an independent consultant who chaired one of several sessions on tsunamis at the American Geophysical Union (AGU) annual meeting last December, and who was one of the first to champion the submarine landslide's culpability. After Papua New Guinea, that's no longer true, he says.
Typically, seismologists and wave modelers attributed tsunamis to the push of a thrust fault underwater: Once an earthquake heaved a landmass up, the water would follow. The size and acceleration of a wave would depend on the mass displaced and its speed (slow earthquakes can still create large waves). Now, models are incorporating what landslides over the past century, from Alaska to Nicaragua, have shown: that a sliding landmass underwater, though it may not send a wave across the Pacific as large earthquakes do, might create local tsunamis.
The commercial heart of Kodiak, Alaska, was devastated following the arrival of several large seismically generated waves, an hour after the start of the 1964 earthquake. Before the offshore waves hit at almost 20 feet above tide level about 160 fishing boats occupied the harbor pictured here. However, landslide-generated waves and associated nearshore landslides caused most of the casualties and damage. Courtesy of George Plafker.
Modeling actual tsunami waves has taken a huge step forward recently, says Jose Borrero, a coastal engineer who works with Costas Synolakis at the University of Southern California. Nevertheless, he says, "exactly the mechanics of how water waves are generated, whether it be tsunamis generated from landslides or earthquakes are still under debate." Current models using a variety of methods have been effective at "hindcasting" tsunamis, taking into account fieldwork and other records. Borrero and others would like to see a network of submarine "tsunometers," much like the seismic network that monitors earthquake regions in California and elsewhere.
So far, the one tsunami warning system under development, the DART system built by the National Oceanic and Atmospheric Administration (NOAA), relies on past events for modeling current ones. Hydrophones, tide gauges and seismic reports provide the meat for models that can forecast tsunami waves within half an hour of their propagation. "In a nutshell, it's get the measurements as quickly as you can and run the model to match the measurements," says Frank Gonzales, with the tsunami research program at NOAA's Pacific Marine Environmental Laboratory in Seattle. Still, Gonzalez says, the system can listen only for large events that propagate ocean-wide waves.
Nevertheless, the nascent program, deployed along the Pacific Coast of the United States, in order to monitor and protect Hawaii, has already had some success. As scientists reported at the AGU meeting, the system modeled the wave action from an Alaskan earthquake last November in real time within one centimeter of what tide gauges registered. An Alaskan earthquake in the magnitude-7 or -8 range can generate a tsunami that will take coastline, for example, and can be forecast within 20 minutes or so. But if it's headed for a community two or three minutes away, Gonzales says, "no amount of technology in the world is going to save your life." Nevertheless, "a minute can make a difference," he says. "The farther you run away from the water, the weaker the tsunami will be when it catches you, regardless of the country, regardless of the situation." Gonzales and other researchers emphasize that education for coastal populations is the key to survival, and eventually, he says, "an early warning can save lives."
Paleotsunami research feeds into detection and warning systems, showing what kinds of events a region may have in its future. Characterizing older events is much like determining historical seismicity or flood frequency for a region. Each of the five western U.S. coastal states has a team creating tsunami hazard maps for their coastal communities, as part of NOAA's National Tsunami Hazard Mitigation Program, determining what regions are most at risk and for what magnitude events.
"The geologic record is powerful," says Bruce Jaffe, a geologist with the U.S. Geological Survey in Santa Cruz, Calif., providing records of events where people were not present. For example, in the coastal region off the Cascadia subduction zone, estuaries contain mud or sunken peat layers that subsided during an earthquake and were then covered with sand brought in by a tsunami. (Radiocarbon dating and tree-ring chronology from those estuaries also confirmed tsunami reports in Japan, from the occurrence of a huge Cascadia earthquake in 1700, modeled last November in Journal of Geophysical Research.)
Recent tsunami deposits on land are just as important, Jaffe says, particularly for recognizing the older geologic evidence. He traveled to Papua New Guinea several months after the 1998 tsunami to examine the new deposits there, as well as to Peru after a tsunami followed a magnitude-8.4 earthquake on June 23, 2001. In general, Jaffe says, tsunami waves deposit a layer of very fine, well-sorted sand brought in from offshore. Currently he is refining deposition models, attempting to determine tsunami flow velocities from sediment transport characteristics.
Researchers in the field may squabble over the details of the modeling, but most now seem convinced that the problem is complex: a dangerous cocktail of landslide dynamics, seismology, geology and other factors. That awareness took a while to become mainstream, says George Plafker, a geologist emeritus at USGS in Menlo Park, Calif. Plafker started working on tsunamis in 1964 after an immense earthquake struck Alaska, bringing both seismic- and landslide-induced tsunamis ashore. "There was never any question about what was going on," he says of the one-two punch. The event included immediate waves generated by local submarine landslides, as high as 42 meters in nearby fjords, followed by a large seismic tsunami that followed within an hour of the earthquake, over 10 meters high.
Since the Papua New Guinea event, models "are finally becoming realistic," Plafker says. "All these factors have to be taken into account before you talk about any particular tsunami."
The shift has brought new urgency to mapping the seafloor for regions with potential for underwater landslides, which share some characteristics of onland slope failures. After detailed mapping of such slumps offshore Santa Barbara, Gary Greene of the Monterey Bay Aquarium Research Institute and coworkers developed a set of criteria for recognizing potential tsunami-producing submarine landslides, including seismic setting, the circulation of groundwater and water inputs from onshore rainfall, and the presence of hydrocarbon reservoirs or other porous flow regimes. "Based on what we know today, we can predict areas that could generate tsunamis," Greene says, "and that's where we should go out and look."
Back to top