Cascadia's Fault - Jerry Thompson [73]
It was the largest seismic event in Alaska since the Good Friday disaster of 1964. The ground shook and rolled for almost two minutes. Even though the event would be classified as a “great” earthquake (magnitude 8 or higher), it caused only moderate local damage—cracked masonry and concrete walls, collapsed ceilings and partition walls, spalling on concrete beams and piers—all of which was described by the Anchorage Daily News as “one of many temblors that rattle the chain every year.” Two things, however, made this one different.
The first was that it had been forecast a year earlier by a team of scientists at the University of Colorado. The researchers were looking for precursors, things that change in the earth just before large main shocks. This being one of the most seismically active regions on the planet and the same big subduction zone that had caused the disaster of 1964, a whole slew of new instruments had been installed by 1974—the Central Aleutians Seismic Network—providing a flow of data with enough details to spot even subtle changes.
One of those changes was a sudden drop-off of seismic activity. When all the normal rumble and grind along a big subduction zone mysteriously goes quiet, watch out—something’s bound to happen. Or at least that was the theory at the time. When this quiescence was noticed in a segment of the plate boundary near Adak Island, the researchers in Colorado decided to go out on a limb. They said a major earthquake would occur near Adak before the end of October 1985.
When October came and went and the only big rupture was down in Mexico City, the scientists gracefully admitted their mistake and chalked it up to experience as a “failed prediction.” Six months later, on May 7, 1986, a quake roughly a hundred miles (160 km) southeast of Adak Island did occur at precisely the location they had thought it would. By then a new computer system had been installed at the Alaska Tsunami Warning Center in Palmer, just north of Anchorage. It was designed to record and locate the focal point of earthquakes quickly, analyze the data, and predict whether or not a tsunami had been generated at the same time.
This was the second thing that made the May 7 shockwave different. It was the first test of the new software. The alarm system was tripped and a team of geophysicists on duty at Palmer had to decide whether to believe the computer. They knew this segment of the Pacific plate had failed before in a magnitude 8.6 temblor in 1957, generating a tsunami that did extensive damage in Hawaii. They confirmed very quickly that the current rupture was definitely located along the sea floor and was therefore certainly capable of generating waves. Whether it actually had or not was another question.
Sometimes a slab of sea floor will move more horizontally than it does vertically, so these kinds of jolts don’t always lift a wall of water that becomes a tsunami. In the end it was a judgment call. With the odds apparently in favor, the team at Palmer did the cautious thing and notified emergency officials, who sounded the alarm all around the Pacific Rim that a wave had probably been triggered.
From the moment of rupture until the alarm went out only eight minutes elapsed, less than half the time it used to take when the work was done by hand with calculators, rulers, and maps. The new system developed by Palmer station chief Thomas Sokolowski made this the fastest tsunami warning ever issued.
The trouble was that the computer system was still based primarily on seismic data—instrument readings of the earthquake’s ground motion—with no quick way of directly measuring waves in the ocean. This was back in the days before deep-ocean detection buoys that could register a change in sea level and provide accurate data about how big the waves were and what to expect in places like Hawaii, Japan, Vancouver Island, or Willapa Bay—where a tsunami might be headed. Without the buoys, the new warning system was still an educated guess based on tide gauges and