Herb Dragert of the Pacific Geoscience Centre in Sidney, B.C., was checking the GPS records of tectonic strain accumulation in southern Vancouver Island when he caught what appeared to be an error in one of the measurements. Since 1992, the Pacific Geoscience Centre and the Pacific Northwest Geodetic Array (PANGA) GPS networks had been recording the slow accumulation of elastic strain in the North America Plate as the oceanic Juan de Fuca Plate drives northeastward beneath it. GPS stations on southern Vancouver Island and northwestern Washington had been showing a northeast-directed shortening of the geodetic baselines between them and a station at Penticton, B.C., in stable North America to the east. In his previous checks, the stations were doing just that, but this time, the baseline was going in the opposite direction, southwest rather than northeast, extending rather than contracting. Dragert determined that all the stations in southern Vancouver Island and northwest Washington were doing the same thing, lengthening their base lines. After he recognized that he was witnessing something real, the extension stopped, and the baseline to Penticton began contracting again. Dragert had observed a slow earthquake in early 1997, defined as an earthquake that doesn’t quake, in this case, a movement on the subduction zone not accompanied by strong shaking (Figure 4-24).
Meghan Miller, Tim Melbourne, and their colleagues at Central Washington University at Ellensburg began looking at the entire data set starting when the PANGA network first became operational in 1992. The network recorded a slow earthquake in mid-1992 and has been recording them ever since at intervals of about fourteen and a half months. If the strain had been released suddenly in an ordinary earthquake, it would have had a moment magnitude of 6.7, about the size of the devastating Northridge, California, earthquake of 1994. But because the strain was released slowly over a period of two to four weeks, nobody felt a thing. Michael Brudzinski of Miami University, Ohio, determined that the duration of these non-shaking earthquakes varies from place to place. In western Oregon, the duration is 19 months, and in southernmost Oregon and northern California, it is about ten months. The area with the longest time between tremor events overlies the thick slab of crustal basalt known as Siletzia.
At first, it was thought that the seismograph network didn’t detect anything, either. But Garry Rogers of the Pacific Geoscience Centre, working with Dragert, found that the slow earthquakes are accompanied by a strange seismic signal of lower frequency than ordinary earthquakes (Figure 4-18). These signals resemble volcanic tremor rather than slip on a fault, and Rogers suspects that they are related to fluids moving in the deeper part of the subduction zone. He was able to locate these seismic events on the subduction zone at depths of fifteen to twenty-eight miles. They are found in the same places that the GPS stations recorded the slow earthquakes, although the seismic signal may extend farther northwest into Vancouver Island, beyond the reach of detailed GPS coverage. They are now known as episodic tremor and slip (ETS), and they have been identified in other subduction zones around the world, including southwest Japan and southern Mexico. The comparison of seismic tremor with ordinary earthquakes is like comparing your stomach growling with breaking a stick or setting off a firecracker.
There have been enough of these ETS events to locate their “epicenters” on a map, shown here as Figure 4-25. The western boundary of the ETS events is relatively sharp, and this is now thought to correspond to the eastern boundary of the locked zone, independent of any estimates based on a downward projection of temperature. The eastern boundary of the locked zone is just east of the coastline in Oregon and includes a slightly wider part of the Washington coast.
Figure 4-26 is a cross-section of the Cascadia subduction zone subdivided according to its expected response to deformation. It is useful to compare this figure with Figure 2-1, showing the increasing strength of rock with increasing depth. The westernmost subdivision is closest to the “trench” and is weak, moving by sliding unaccompanied by earthquakes. The next region has the highest strength. It has no earthquakes on it at all at the present time and is expected to release its strain in gigantic subduction-zone earthquakes. The next zone to the east is characterized by ETS, with slow earthquakes probably influenced by high fluid pressures. Still farther east is the zone of stable sliding, beneath the brittle-ductile transition, which extends eastward beneath the volcanic arc.
Does this release of strain on the subduction zone reduce the threat of another M 9 subduction-zone earthquake? Dragert thinks that the opposite may be true: the release of strain in the transitional zone between brittle failure and ductile failure may increase the elastic strain on the shallower part of the subduction zone that is completely locked. It is possible that the next major subduction-zone earthquake might be preceded by a slow earthquake, The 1960 Chile earthquake, the greatest of the twentieth century, was preceded by a slow earthquake. For this reason, the time of an ETS event is a time of greater concern that one of these events might trigger the next great subduction-zone earthquake. Everybody gets nervous. At present, it is the only possible “precursor” we have, and it might not be a precursor at all.