Can fracking cause bigger, more frequent earthquakes?

Injecting fluids into the Earth, whether to recover natural gas or to obtain thermal energy from the planet, can cause earthquakes. New reports that look at American fracking, deep waste-water injection, and geothermal activities suggest there are big risks and thus a need to develop strong regulatory framework to deal with them.

The most striking indication of human-induced earthquakes is provided by the graph below, which shows the cumulative number of earthquakes in the central and eastern US that were greater than or equal to magnitude 3.0 on the Richter scale. The clear increase from 2005 coincides with the rapid increase of shale gas wells and associated increased, deep waste-water injection. Between 2005 and 2012, the shale gas industry in the US grew by 45 percent each year.

Three reports have been published this month in Science that add to our limited but growing data on the causal link between fluid injections and earthquakes.

Pumping fluids into the ground changes how ground water travels through the porous rock systems. This can affect the rock in two ways. First, these injected fluids can leak directly into faults, which are fractures in rocks, and change pore pressure along the fault, causing failure. Second, they can alter the mass or volume of the rocks that overlie a fault, which changes how much loading the layers underneath have to bear. In this case the fluids do not interact with the fault directly, but cause it to fail by changing its surroundings.

In the first report, a review article, William Ellsworth of the US Geological Survey points out that earthquakes are occurring in unusual locations in North America and Europe. He looks at activities where injecting fluids into the ground may cause earthquakes—such as mining for minerals and coal, oil and gas exploration/production, as well as the building of reservoirs and large waste-water disposal sites. Ellsworth examines three case studies of deep injection which are particularly convincing.

In 1961, fluid was being injected to a depth of 3.6 km at a Colorado chemical plant for disposing hazardous chemicals. By early 1962, nearby residents started reporting earthquakes. By 1966, 13 such earthquakes had been recorded in that area of magnitude 4.0 or more.

In 1969, the US Geological Survey also started injecting fluids at another site in Colorado. This time, their aim was to understand how fluid pressure could influence earthquakes. They noticed that whenever the fluid pressure went beyond a critical threshold, more earthquakes were observed. This indicated that earthquakes could potentially be controlled if the pressure at which fluids are injected is controlled properly.

The most remarkable example, however, comes from injections in Paradox Valley in Colorado (which are still ongoing). In that area, during the period of 1985 to 1996, only three tectonic (natural) earthquakes were recorded within 15km of the site. Between 1991 and 1995, when the injection tests were conducted, hundreds of induced earthquakes were detected within 1km of the site, while few were detected beyond 3km from the site; all were below magnitude 3.0. This situation, however, changed with continuous injection activities. In 2000, there was an earthquake of the magnitude 4.3 recorded 8km from the site, while earlier this year there was one of magnitude 3.9 recorded roughly the same distance away. This indicates that long term injection can lead to expansion of the seismically active area and trigger bigger earthquakes.

In the second report, Emily Brodsky and Lia Lajoie of the University of California at Santa Cruz looked at the Salton Sea Geothermal Field in California. They tracked the total volume of fluid injected and extracted to extract heat from the Earth’s core. The team found it correlates to the number and magnitude of earthquakes. So it’s not just injection, but also extraction that needs to be given attention.

The last report is by Nicholas van der Elst of Cornell University and his colleagues. This study tracked induced earthquakes that are triggered by much larger, natural earthquakes that occur far away. Injection of water in the deep ground elevates pore pressures and makes the faults and fracture networks in the rocks more vulnerable. With that, a distant event can push the system over the edge and cause earthquakes around the injection sites.

The fact that human activity alters the landscape in a way that can cause earthquakes is not surprising given the scale of activities. The mud volcano “Lusi” (Sidoarjo mud flow) in Indonesia is a striking example of what can happen when drilling interacts with the subsurface, and debate lingers as to the “human trigger” for the event. At Lusi, a mud volcano erupted shortly after drilling for gas, and as the disaster developed, it covered the surrounding houses, displacing some 13,000 families and closing 30 factories and hundreds of small businesses.

Humans are aware of the fact that earthquakes can be induced by fluid injection, and we are learning more about how injections cause them. So perhaps, as Ellsworth concludes, it's high time to intervene with a clear regulatory framework to mitigate the risk.

Science, 2013. DOI: 10.1126/science.1225942, 10.1126/science.1239213 and 10.1126/science.1238948 (About DOIs).

Dougal Jerram is Professor II at University of Oslo. This article was first published at The Conversation.

This story has been updated to clarify a paragraph.

Listing image by Flickr user: danielfoster437

View the original article here

Volcanic earthquakes produce a “seismic scream” just before eruption

The Redoubt Volcano, after a 1990 eruption melted most of a glacier off its southern face.

Volcanic activity is intimately associated with seismic activity. You simply can't force molten or semi-molten rock through a mountain without cracking a few faults in the process. If we were ever able to understand how to read the seismic activity correctly, it could provide valuable advanced warning about impending eruptions.

A 2009 eruption of Alaska's Redoubt Volcano may not get us much closer to an advanced warning, but it provides a detailed glimpse of the last moments before an explosive eruption. Shortly before the eruption, small faults within the volcano were breaking so frequently that they merged into what's being called a "seismic scream." Then, within a few minutes of the eruption, the scream got cut off as the last resistance gave way.

Redoubt is a stratovolcano, built from material that melted as the Pacific plate subducted beneath Alaska. Like some more famous examples, such as Mount St. Helens, it alternates between slow eruptions of extremely viscous rock and sudden, explosive ones. The 2009 eruption was accompanied by a number of small explosions (small at least in the sense that the mountain was still there afterward); the researchers focused on the seismic activity that led up to these explosions.

Most of the earthquakes associated with the eruption were small (between magnitude 0.5 and 1.5) and centered a few kilometers below the volcanic vent. There was plenty of activity of this sort seen during the eruption, but something unusual happened before the largest explosion: "These small earthquakes occurred in such rapid succession—up to 30 events per second—that distinct seismic wave arrivals blurred into continuous, high-frequency tremor." This continuous tremor is what is being called the "seismic scream."

The earthquakes themselves might be enough to make you nervous, but something even more unnerving happened after a few minutes of screaming: things suddenly went quiet. For somewhere between 30 seconds to a minute, the low magnitude quakes stopped, although sometimes larger ones would happen. And then, the explosion hit.

The authors extrapolated the scales of known earthquakes down to something of this magnitude and came up with the rupture of small faults, about 20m long, that are only sliding a millimeter with each event. To understand the forces involved, the authors built a model of the internal faulting at the volcano and got it to reproduce the behavior seen on the seismograph.

As they gradually increased the pressure on the faults, the authors' model responded with more frequent earthquakes, with the frequency slowly ramping up through frequent earthquakes before reaching the seismic scream phase at stress rates of about five MegaPascals a second. At that point, the faults move steadily but alternate with sticking and sudden slips. As the stress rate reaches 20MP/s, the sticking stops, and the fault goes into a smooth glide.

The authors consider that value, 20MP/s, unexpectedly high and don't seem to want to go into what might require that much force to shift. "It is beyond the scope of this study to rigorously evaluate potential [magma] conduit processes responsible for such extreme loading conditions." But then they go ahead and do so anyway, suggesting that the magma is forcing an obstructing piece of rock against the walls of the conduit to the surface. All told, the obstructions seem to move only about five meters, after which the way is clear for the material to explode to the surface.

Nature Geoscience, 2013. DOI: 10.1038/NGEO1879  (About DOIs).

View the original article here

Volcanic earthquakes produce a “seismic scream” just before eruption

The Redoubt Volcano, after a 1990 eruption melted most of a glacier off its southern face.

Volcanic activity is intimately associated with seismic activity. You simply can't force molten or semi-molten rock through a mountain without cracking a few faults in the process. If we were ever able to understand how to read the seismic activity correctly, it could provide valuable advanced warning about impending eruptions.

A 2009 eruption of Alaska's Redoubt Volcano may not get us much closer to an advanced warning, but it provides a detailed glimpse of the last moments before an explosive eruption. Shortly before the eruption, small faults within the volcano were breaking so frequently that they merged into what's being called a "seismic scream." Then, within a few minutes of the eruption, the scream got cut off as the last resistance gave way.

Redoubt is a stratovolcano, built from material that melted as the Pacific plate subducted beneath Alaska. Like some more famous examples, such as Mount St. Helens, it alternates between slow eruptions of extremely viscous rock and sudden, explosive ones. The 2009 eruption was accompanied by a number of small explosions (small at least in the sense that the mountain was still there afterward); the researchers focused on the seismic activity that led up to these explosions.

Most of the earthquakes associated with the eruption were small (between magnitude 0.5 and 1.5) and centered a few kilometers below the volcanic vent. There was plenty of activity of this sort seen during the eruption, but something unusual happened before the largest explosion: "These small earthquakes occurred in such rapid succession—up to 30 events per second—that distinct seismic wave arrivals blurred into continuous, high-frequency tremor." This continuous tremor is what is being called the "seismic scream."

The earthquakes themselves might be enough to make you nervous, but something even more unnerving happened after a few minutes of screaming: things suddenly went quiet. For somewhere between 30 seconds to a minute, the low magnitude quakes stopped, although sometimes larger ones would happen. And then, the explosion hit.

The authors extrapolated the scales of known earthquakes down to something of this magnitude and came up with the rupture of small faults, about 20m long, that are only sliding a millimeter with each event. To understand the forces involved, the authors built a model of the internal faulting at the volcano and got it to reproduce the behavior seen on the seismograph.

As they gradually increased the pressure on the faults, the authors' model responded with more frequent earthquakes, with the frequency slowly ramping up through frequent earthquakes before reaching the seismic scream phase at stress rates of about five MegaPascals a second. At that point, the faults move steadily but alternate with sticking and sudden slips. As the stress rate reaches 20MP/s, the sticking stops, and the fault goes into a smooth glide.

The authors consider that value, 20MP/s, unexpectedly high and don't seem to want to go into what might require that much force to shift. "It is beyond the scope of this study to rigorously evaluate potential [magma] conduit processes responsible for such extreme loading conditions." But then they go ahead and do so anyway, suggesting that the magma is forcing an obstructing piece of rock against the walls of the conduit to the surface. All told, the obstructions seem to move only about five meters, after which the way is clear for the material to explode to the surface.

Nature Geoscience, 2013. DOI: 10.1038/NGEO1879  (About DOIs).

View the original article here