16 April 2016 | Quake Insights
A Mw=6.1 earthquake struck on April 14, followed by vigorous aftershocks. The epicenter as determined by the Japan Meteorological Agency (JMA) lies approximately where the ‘right-lateral’ (San Andreas-like) Hinagu and Futagawa faults meet, on the eastern edge of Kumamoto, a city of 700,000 people on the island of Kyushu in southwest Japan. The quake was followed today by a Mw=7.0 quake that may have ruptured parts of both fault systems. The events were shallow (10-12 km or 6-8 mi deep), both were destructive, but the second quake was about twenty times larger than the first one. How large is a Mw=7.0 quake?
In a moment of modern madness, on 9 August 1945 an atomic bomb was dropped on Nagasaki, just 70 km (40 mi) west of Kumamoto, killing 60,000-80,000 citizens. The energy released by today’s M=7.0 earthquake was 400,000 times greater.
The Mw=7.0 event is about 40 km (25 miles) long, may have ruptured to the earth’s surface, and appears to have propagated largely to the northeast. The evidence for the propagation are the seismograms that Gavin Hayes at the USGS in Golden, CO, used to infer the distribution of fault slip, and also the very high shaking (‘peak ground accelerations’) seen as much as 50 km (30 mi) to the northeast of the mainshock. Typical wood frame homes in the US and Japan can begin to suffer structural damage at peak ground accelerations of about 0.40 g (40% of the acceleration of gravity), but values twice that high are seen both near the epicenter and 50 km to the northeast. If these are representative, there will be significant damage over a broad area.
Perhaps most important for the science of probabilistic seismic hazard assessment that is also used in the U.S., and which forms the foundation of the Temblor app, a high chance of shaking at the site of the quakes was forecast in the Japanese national seismic hazard model (warm colors in the map above). Further, both the strike-slip faults that likely ruptured today, and other inclined (dipping) sources along the shore, were known to the Japanese geologists and seismologists, and so were included in the model.
Broadly, these faults are part of the Oita-Kumamoto Tectonic Line (OKTL), which is itself the southern continuation of the Median Tectonic Line (MTL). This 800-km (500-mi) long fault zone is among Japan’s most dangerous inland faults, and plays a tectonic role very similar to the San Andreas fault in California (Mahony et al., 2011) at about half the San Andreas slip rate. The most recent large earthquake on the MTL system was the 1995 Kobe earthquake, about the same size as today’s quake.
The Kobe shock claimed 6,000 lives, whereas a quake of the same size and time in California, the 1989 Loma Prieta quake, took about 60. This humiliating loss utterly changed the practice of seismology in Japan. Almost overnight, seismic networks were transformed from feudal baronies into highly integrated and centralized data collection systems that were opened up to all the world’s scientists for instant sharing and collaboration. Today, just about every facet of the Japanese seismic monitoring systems surpasses all others, and seismologists are attracted to Japan like moths to a light.
Was the M=6.1 a foreshock of the M=7.0, or was the M=7.0 an aftershock of the M=6.1?
So far, the evidence suggests that both are true. Depending on how one counts (and how one counts matters), something like 2-10% of mainshocks are preceded by foreshocks. That’s the good news. The bad news is that no one—no one—can tell a foreshock from any other shock. Foreshocks lack any distinguishing features that mark them for future greatness. So, while tantalizing, this hindsight statistical association has ultimately proved a dead end. But increasingly, we have come to accept that each quake starts a game of roulette: There is a small chance the shock will trigger a big one, and a large chance that it won’t. Earthquakes beget earthquakes, yet most are small, and most triggered quakes are smaller than the triggering quake. But not always. And that brings us back to today’s sequence.
On April 14, Shinji Toda at Tohoku University calculated the Coulomb stress imparted by the M=6.1 quake to surrounding faults, using the Coulomb 3 software that Shinji, Jian Lin, Volkan Sevilgen and I have developed over the past decade. Shinji generously made the map shown here available to the Japanese press on April 14. His calculation assumes that surrounding faults are right-lateral and roughly parallel to the strike to the M=6.1 shock—reasonable but untested assumptions.
What one sees is startling: Both the Hinagu and Futagawa faults lie in the (red) stress trigger lobes, where we would expect more aftershocks and a heightened chance of a subsequent mainshocks. The extent of the trigger zones, about 35 km, is also roughly the length of the M=7.0 rupture.
Was this an earthquake prediction? Absolutely not. Rather, the calculation shows that the roulette odds got better in the trigger zones, and worse in the (blue) stress shadows. If one earthquake ratchets up the stress on a nearby fault, its more likely to fail. Relieve the stress, and it becomes less likely to fail. It’s not a prediction, but it is progress.
It will be fascinating to see if the 28 hours of small aftershocks of the M=6.1 event lit up the future fault rupture. Based on past published studies and experience by ourselves and others, the answer will probably be no, or at best, sort of. But that’s the state of the science, and the state of the art.
Ross Stein and Volkan Sevilgen, Temblor
Data from USGS National Earthquake Information Center, Japan Meteorological Agency (JMA), National Research Institute for Earth Science and Disaster Resilience (NIED); Shinji Toda (IRIDES, Tohoku University); S.H. Mahony, L.M. Wallace, M. Miyoshi, P. Villamor, R.S.J. Sparks and T. Hasenaka (2011), Volcano-tectonic interactions during rapid plate-boundary evolution in the Kyushu region, SW Japan, Geol. Soc. Amer. Bull., 123, doi: 10.1130/B30408.1