A Possible Warning Sign for Human-Caused Earthquakes

I always start my day by checking the U.S. Geological Survey earthquake map to see what’s been happening in California. Almost every day there are earthquakes of magnitude 3 here; they feel almost routine. But east of the Rockies, M 3 quakes are rare events that get in the newspaper. They’re big enough to alarm people though they seldom cause any damage. And in the last few years they’ve been getting more common, especially in association with oil and gas operations.

These human-induced earthquakes are not related to hydraulic fracturing, but to very deep disposal wells that inject waste fluids—formation brine, excess fracking fluid and the like. Over 100,000 injection wells of this type are operating in the United States. Almost all of them make no problems: they do their thing far away from water supplies and are monitored closely. But a handful of wells, maybe a dozen, have gone on to produce unexpected earthquakes.

In today’s issue of Science, a team of researchers reports that injection fields approaching an earthquake-ready state may give us a telltale sign: seismic waves sweeping through from huge distant shocks set off tiny local shakers in the process called dynamic triggering.

Dynamic triggering was first documented after the Landers quake in June 1992, when the M 7.3 rupture in Southern California set off tiny earthquakes over 1000 kilometers away in places like Yellowstone. Today many examples of dynamic triggering have been documented, and it’s no longer a scientific controversy. Since that time we’ve learned that earthquake faults are generally stressed quite near to their rupture point, and from that it would seem that the closer they get to rupture the more likely the slightest disturbance will set them off. I say “it would seem” because ordinary natural disturbances, like tides, magnetic fluctuations and weather fronts, don’t trigger earthquakes. Today’s Science paper fits into that background state of knowledge.

The United States was shaken from a distance by three recent great quakes: Chile on 27 February 2010 (M 8.8), Japan on 11 March 2011 (M 9.0) and Sumatra on 12 April 2012 (M 8.6). The seismic waves were very slow and moved the ground by mere millimeters; only seismometers could feel them. They left the midcontinent unperturbed, except near three injection-well sites in the southern Plains states that later produced earthquakes of their own. The map below shows those sites and the small events (red dots) that were triggered by the distant quakes.

Trinidad, Colorado, is the site of fluid injections related to coalbed methane production. The area had a quake swarm in August 2011 that included a M 5.3 event. The 2010 Chile quake triggered four shocks there within a day and many more nearby over the next few weeks. The 2011 Japan quake had no effect and the 2012 Sumatra quake triggered only a few shocks.

Snyder, Texas, hosts injection wells related to oil production. It had a quake swarm in September 2011 that included a M 4.5 event. The 2010 Chile and 2012 Sumatra quakes had no effect there, but the 2011 Japan quake set off eight events.

Prague, Oklahoma, is also an oilfield area with waste injection wells. It had a quake swarm in November 2011 that culminated in three magnitude-5 quakes. The 2010 Chile quake triggered dozens of events while the 2011 Japan quake did nothing and the 2012 Sumatra quake set off just a few events.

What these cases all have in common is that from 6 to 20 months before their earthquake swarms, there was a sensitive period when distant shaking set off small quakes. The picture they suggest is that the injected fluids were changing the balance of forces on a fault, forcing it toward a later earthquake swarm, and the triggered events were the fault’s response to what we might call a stress test. If this sensitive period can be identified in other injection-well fields we might be able to avoid impending quake swarms, perhaps by easing off on injections or injecting elsewhere.

This possible pattern is not universal (the authors didn’t find it at three other quake-producing injection sites), but it is suggestive. The data fits the scenario of fluid injection gradually raising underground pore pressures to a critical level on previously undetected faults. High pore pressure pushes the walls of the fault apart and allows it to slip more easily. Injection-well operators know how to deal with known faults, and it appears that temporary sensitivity to distant shaking is a clue about a fault that needs careful handling. Dynamic triggering may fit easily into the well operator’s current “traffic-light” system, in which high microquake levels warn them to slow down or stop injections and give the ground a rest.

Progress in this field relies on good data. The study couldn’t look farther back in the past because earthquakes east of the Rockies haven’t been closely monitored. The work reported in the Science paper was made possible by accurate earthquake records from USArray, the long-running federally funded research project in which a giant grid of hundreds of seismometers was temporarily placed, like a doctor’s stethoscope, in a wide belt from the Dakotas to Texas. Today the USArray grid is farther east, but it left some extra instruments behind to help out in the southern Plains states.