Does Fracking Cause Earthquakes?

Mother Jones reports on recent earthquakes in regions not accustomed to much seismic activity, Oklahoma, Arkansas, and Ohio. Much of their story consists of anecdotal evidence, particularly the strongest earthquake in Oklahoma history at magnitude 5.6 in November 2011, which happened along a fault which a Univ. of Oklahoma geophysics professor referred to as “a dead fault that nobody ever worried about.” Since this quake “injured two people, destroyed 14 homes, toppled headstones, closed schools, and was felt in 17 states,” people are starting to worry.

The putative culprit is “injection wells,” the destination of wastewater from hydraulic fracking (and in some cases from dewatering operations). We looked at earthquake data a while ago, because of a study from the U.S. Geological Survey which reported a dramatic increase in earthquakes of magnitude 3.0 or greater over a large area of the U.S., stating that the increase is “almost certainly man-made,” and attributing it to oil and gas production.

We saw that there was indeed a recent, and rather large, increase in the number of earthquakes of magnitude 3 or greater in their study region (longitude 108W to 85W, latitude 25N to 50N). This new article made me wonder, did the increase also occur in Ohio (just east of their study region)? Also, what about the rest of the USA?

So I retrieved earthquake data from the U.S. Geological Survey for the time span January 1973 to the present, for the entire continental USA (specifically, longitude 125W to 65W, latitude 25N to 50N). Let’s look at the data from 1973 through the end of 2012.

Here are the locations of earthquakes with magnitude 3 or greater (click the graph for a larger, clearer view):


Clearly the majority of earthquakes happened along the “Pacific rim,” where continental plates border each other. But how has the number of quakes changed over time, and how are the changes distributed geographically?

I split the entire region into 5×10-degree latitude-longitude grids. Then I counted the number of earthquakes of magnitude 3 or greater in each grid. I then “normalized” the numbers to compensate for the fact that different regions have vastly different numbers of quakes. I didn’t normalize the usual way — subtract the mean and divide by the standard deviation — I just divided by the mean value.

Here’s the result, with the time series of normalized quake count for each grid superimposed on the map of the USA (click the graph for a larger, clearer view):


The largest “spike” happened in 1980 in the Pacific northwest. I strongly suspect this is related to the eruption of the Mt. St. Helens volcano, which was preceded by much seismic activity.

The next largest spikes are recent. The biggest is in the grid covering the northern half of Oklahoma and the grid just to the east of that. There’s also a recent spike in the Pacific southwest, covering the southern California/Mexico border.

There’s no sign of enhanced activity in the grids which cover Ohio. But it turns out that there have been so few quakes of magnitude 3 or greater, the counts are too low to have much statistical power. If we count the number of quakes of magnitude 1 or greater, we get this:


Now the grid which includes northern Ohio shows some sign of recent activity, and that which covers southern Ohio shows a distinct spike.

I also looked specifically at Oklahoma (the bulk of it, longitude 100W to 95W, latitude 33N to 37N) and Ohio (about longitude 85W to 80W, latitude 38N to 42N). At magnitude 3 and above, Oklahoma shows a distinct recent spike:


It’s also there when including all quakes of magnitude 1 or greater:


For Ohio, the count of quakes with magnitude 3 and above is just too small to tell very much, not more than 3 in any year:


But when counting quakes of magnitude 1 and above, there’s an unmistakeable recent spike:


So far, the fossil-fuel industry has denied any connection between recent earthquake activity and oil/gas production. The U.S. Geological Survey disagrees. Who you gonna believe?

36 responses to “Does Fracking Cause Earthquakes?

  1. Opus the Poet

    While I agree that ther does seem to be strong correlation between fracking and a recent upsurge in minor earthquake activity I have to remind people that correlation does not equal causation. However as there is no other structurally active man-made or otherwise activity in the area the data is very suggestive.

  2. How does the timing of the ’80 quakes line up with the eruption of Mount St. Helens?

  3. Mount Pinatubo? I think you might mean Mt St Helens?

  4. I’ll believe the USGS too, because they pin the blame on waste-water disposal wells, not fracking.

    And the fossil-fuel industry isn’t denying that water disposal causes earthquakes (though they might be denying this one–is there a reference where they are?). They’ve known and admitted for several decades that waste-water injection can cause earthquakes, though these tremors typically tend to be so small they can’t be felt or they’re just tiny shakes (the 5.7 earthquake is quite an exception to this). So, now that it’s known that bigger earthquakes can be triggered, things have to be done more carefully, though a big earthquake is still going to be pretty rare from a risk perspective.

    And, frankly, it’s easy to regulate this kind of risk. Better geophysical and stress mapping to identify potentially risky faults. Plus, big quakes just don’t happen without some smaller seismicity indicators first. So you establish seismometers in the area around an injection well and, when you start getting smaller tremors, you shut down the disposal well. If the problem with that well appears to be chronic, then maybe the well should be abandoned.

    It’s pretty simple. Problem is, in the U.S., industry has a knee-jerk reaction against any additional regulatory measures. Meanwhile, to people opposed to any oil and gas exploration, there’s no acceptable risk no matter how small.

    • Having just watched the films ‘Silkwood‘ and ‘Erin Brockovich‘ again recently, I’m reminded that where industry smells profit, there’s very little risk that’s considered ‘unacceptable’.

      • Random Thoughts

        Do you know the difference between fiction and non-fiction?

        [Response: Aren’t both those movies based on true stories?]

  5. You said:
    I strongly suspect this is related to the eruption of the Mt. Pinatubo volcano, which was preceded by much seismic activity.

    Did you mean Mt. St. Helens?

  6. One other thing to be aware of: ordinary mining activities, such as quarry explosions, routinely produce shocks which the US picks up as magnitude 1-2 earthquakes. I doubt that this makes a difference in your results, but its worth considering.

  7. Tamino, I suspect the 1980 spike, (hey. there’s that spike again…) was Mt. St. Helens.

  8. Nice analysis. For the sake of completeness, I think you mean St. Helens rather than Pinatubo.

    [Response: Doh!]

    Are there appreciable changes in seismometer density over this period? My guess would be no. But it might be worth checking in places like Oklahoma, where recent exploration has been matched by increases in geophysical observations of all kinds — and, hence, perhaps higher detection probability for smaller earthquakes.

    Finally, do they provide depth estimates? I would expect fracking-triggered earthquakes to be shallow more often, even if the direct trigger often cascades to non-local slip.

    • The US array network, from what I recall, was completed in the late nineties (not counting the temporary arrays to get a 3D map of US crust). Several states have also a regional network, but they are not well advertised so I do not know (except of course the California network)

      For the depth estimates, one word of advice : don’t believe it if it comes from USGS and it’s 10 km or less. Not the USGS fault, it’s simply that, with a regional surface array, you cannot get straigthforwardly a depth estimate for the shallow earthquakes. Only a local array would be able to get a depth estimate with an error bar below several kilometers.
      And lots of natural faults in the US are shallow ones.

  9. Re the 1980 spike – very definitely caused by Mount Saint Helens. I downloaded the data myself to have a look, and there were some 388 earthquakes in that grid box between March 15th and May 18th 1980. March 15th was the onset of volcanic activity, May 18th the date of the cataclysmic eruption. The smaller bump later on is due to seismic activity associated with the 2004-2008 activity at the volcano.

    If you do a 3D plot of latitude, longitude and depth of earthquakes in that grid cell you can clearly see the subduction zone that causes the seismicity, with lots of deep earthquakes in the west of the box and no deep quakes in the east.

    For the Oklahoma box, the 3D latitude-longitude-depth plot shows that almost all earthquakes there are very shallow. In fact they are almost all recorded as having a depth of 5.0km. I wonder if that’s a default depth to assume when the quake can’t be properly localised.

    More interestingly, the latitude-longitude plot looks extremely different for years post-2000 compared to pre-2000. Before, earthquakes occurred mostly in the southern half of the box but with no obvious spatial concentrations. Post-2000, there are three very obvious spatial concentrations, at (-97, 35.5), (-102, 35.5) and (-105,37).

    I tried to find out if those latitudes and longitudes correspond to locations where fracking is taking place but couldn’t find the necessary information. Definitely no volcanoes there though.

  10. The Uk firm doing the first- um er 2! wells [and no gas yet] admitted the tiny quake we had was probably caused by them
    “the report, commissioned by energy firm Cuadrilla, also said Fracking ‘likely cause’ of quakesthe quakes were due to an “unusual combination of geology at the well site”.

    So no denial here- perhaps because the damage was non existent and no compensation required, but little earthquakes could crack the well lining allowing gas to enter surface strata. Is there a survey of people complaining about fracking related water contamination? And could this be corrilated to seismic activity?

  11. There have been many earthquakes in the last 3-4 years in Ohio, WV, KY (coal country now transitioning to natural gas) … nearly all of them have had a magnitude of around 2.5 … Fracking has been suspected in a recent Youngstown, OH earthquake, and a Braxton County, WV earthquake…. Other ones near Gallia county OH recently are so far inconclusive (one of the problems is that there are injection wells everywhere. Obviously correlating an increase of sub-3 magnitude earthquakes with the increased frequency/presence/use of the injection wells would end up speaking a pretty loud message.

    It’s a bummer that so much energy extraction these days (of many types) requires so much water.

  12. Rattus Norvegicus

    Note that the quakes in MO seem to cluster along the New Madrid fault. Well known and capable of causing very large quakes…

  13. I agree that there are plenty of knee jerk reactions on both sides from what I have seen. One question that has arisen is whether these shallow quakes can actually release stress on a more gradual scale. I believe this would be true in certain cases. What do you think, Tamino? By the way, nice blog.

    • David B. Benson

      Shallow earthquakes are most unlikely to significantly effect mid and deep earthquakes.

  14. Then there is the matter of sinkholes. Are they probable on areas with fracking? Would be a bit funny if fracking lead to a significant loss of equipment and not just to polluting the aquifers for 1000 years , so the payback would be immediate instead of gradual.

  15. Pretty clear the ways you’ve done it (counts of magnitude >1 and >3) but I wonder if it makes sense to multiply the magnitude and the frequency. A better way for me to ask my question is probably: What is the most statistically sensible way to combine magnitude and frequency for something like this? There are other questions where effects might be more easily detected by combining response variables (ice thickness & area is the only example currently coming to mind, but not a great example because thickness data aren’t easy to get), but I hope you know what I mean.

  16. How does the increase scale with magnitude? I don’t see any obvious difference between >1 and >3, but the increase might not be completely independent of magnitude. In any case, modelling the scaling should give you a decent estimate for how often, say magnitude 6+, quakes happen now. If there’s a decent chance of a bigger one within the next few decades, you’d probably have a good argument for banning fracking before that happens.

    • Opus the Poet

      The Richter scale is logarithmic, that is a double in magnitude means the quake is a squared higher energy than the next-lower number. So going from 1-3 means a 3 is a square of a square (4th power) more intense. A 1 is not much, but a 3 is a significant amount of energy released.

  17. Fracking works by blasting water, sand and chemicals into hydrocarbon-bearing strata, with the aim of shattering rock and releasing the hydrocarbons; of course it causes Earthquakes, if it didn’t it wouldn’t work. The marketing men employed by the industry might not know this, but the engineers & geologists certainly do.

    The questtion should be ‘Does Fracking Cause Earthqaukes Big Enough To Worry About?’

  18. Thanks for this Tamino, I’ve been looking for an analysis like this for some time and you’ve put it into a great, easy-to-visualize form.

    The first point is one directly pertaining to your analysis. I notice that the OK/MO earthquake spike is just that — a “spike”. That is, it seems to have dropped off rather precipitously. Is that an artifact of the way that you organized the data (e.g. does it not have all of 2012?), or is it real?

    [Response: 2012 is complete. Note that while its value is less than 2011, it’s still much higher than the background level.]

    A few potential contrarian talking points (i.e. why I’m not fully convinced yet frakking causes earthquakes):
    1.) Earthquakes don’t follow frakking. Why are there Earthquakes in Oklahoma and Missouri, but not in Pennsylvania, Kansas, and the Dakotas, where there is also large scale fracking?

    2.) Physical mechanism. How does frakking cause earthquakes? One thing that bothers me is that the earthquakes are frequently deeper than the depth of fracking (e.g. 15 km deep vs. deepest fracking wells of 5 km). The wells are also usually not on the fault itself. Compare the map I linked to the first figure in Tamino’s post. The earthquakes aren’t happening at the well locations — there’s a substantial offset.
    (3. Spikiness of the data (if real). Why is it not a continual increase, or plateauing? Frakking is ongoing and increasing, why are the earthquakes a discrete event? Futhermore, in relation to 1 and 2, they seem to be following a linear feature aka fault. What’s to say it’s not just a fault that’s becoming active again after many years, since you can see spikes elsewhere with natural causes (e.g. Mt. St. Helens))

    I haven’t thought of a convincing response to any of these points yet….

    • It’s related almost entirely to waste disposal, where brines produced during oil and gas production are injected into a deep formation that’s pretty conductive to water, which allows it to flow fairly far from the disposal well (a mile or more). So, if there’s a fault within a mile radius, there’s a chance an earthquake can be triggered (a chance, depending on stresses and how locked that fault is). Of course, earthquakes can’t be bigger than what natural stresses will permit.

      Earthquakes directly related to fracking have occurred in Wales (two small earthquakes) and in the Horn River Basin of northeast British Columbia (a series of very small earthquakes, hundreds if you count the tremors below 2 on the Richter scale, which can’t be felt at all on the surface), where it appears that fluid was directly injected into faults.

      So the lesson is don’t frack into a fault (which operators want to avoid anyways: if a fault sucks up their frack fluid, then they’ve wasted a frack stage and a few hundred thousand dollars). Regulations in British Columbia have changed as a result too: if an earthquake of 4.0 or more on the Richter scale is triggered during a fracking operation, the operation must be stopped. Earthquakes of 2.0 or more must be reported.

      Finally, to get to the point, the difference between waste injection and fracking is distance, pressure, and time.

      The fluids injected during fracking aren’t going to travel beyond where the fracture system is created (only a few hundred metres from the well), so, if there’s no nearby fault, then there won’t be a problem.

      Pressure wise, the pressures used during fracturing are much higher than for waste injection, so if you do intersect a fault, you could trip it just by changing those pressures so dramatically (limited by how long the fault is exposed to that pressure).

      However, time wise, the higher pressures created during fracking are only sustained over a few days and are drained once production begins. Meanwhile, waste-injection wells increase the pressure in a formation for many years.

      So, like everything, it’s complicated.

  19. FWIW, a quarry about 1 mile from my place once got a bit happy with LOX/carbon (20Klbs??) which produced a claimed USGS 3+ quake and as I recall they pegged it at > 3 miles of depth (it was nearly a surface blast). It broke a tall window in my house, which shook much more like a 5+ than a 3+. What was interesting at the time was the delay from 1st sharp ground jolt to the air blast which was also significant. That plus the enormous billowing dust cloud rushing down the canyon from the quarry made the source of problems pretty obvious.

  20. Curious — how much fossil-fuel CO2 needs to be sequestered, and how does that compare to the volume of methane being removed by fracking?
    And, does anyone know if fracking is a one-time release? I know that some (not all) aquifers collapse when water is removed from them and will not refill (Kettleman Hills in California is the example often used — where the laws said only those who kept pumping could retain their water rights, so everyone kept pumping til the aquifer collapsed, where if they’d all backed off and not pumped beyond the recharge rate the aquifer would have remained useful).

    If a site is fractured and the methane removed, does that area collapse so CO2 can’t later be injected into it? (If so, it’d make sense to get the CO2 sequestration running in tandem with methane removal to keep the pore/crack spaces open while exchanging the contents).

    Foresight: expensive, but consider the alternative.

    • The reservoir doesn’t collapse. It’s a shale, meaning it’s not karsted (which is reserved for limestones and dolostones).

      Shales, while everywhere, aren’t great candidates for CO2 injection, however. The natural permeability is very low, which is why they have to be fracked so hard in the first place to get the oil and gas out (of which they’ll only produce about 20 to 25% of the gas and far less of the oil).

  21. Ask NAM (Shell/Exxon) in the Netherlands what gas production causes over near 50 years and what they’ve been paying for since decades. A special damage assessment team is employed 24/365. Schalie is big discussion, doubtlessly causing for strata to compact, slowly or acutely. How acutely… you tell me. There’s the Troll gas field in the Norwegian sector, that had it’s platform lifted by quite a few meter due to seafloor subsidence [anticipated before hand but not by a factor as large as it developed]. On land it means dramatic change of water flow at the subsidence is not regular. Considering the Netherlands for a large portion is below sea level, guess what needs done to the dikes and levies.

    • The Troll Field produces from a different kind of reservoir, with very high porosities in a sandstone. You remove most of the gas from that and then you can also remove a lot of the pore pressures supporting the sand grains. If the sand grains in the sandstone aren’t cemented together very well, then the reservoir can compact and then subsidence over it can occur.

      Shales are a different beast, typically with low porosities while gas recoveries will be lower. Further, shales being exploited for oil or gas are built more like glass than are composed of individual grains of sediment (the individual grains of sediment originally deposited in the shale were later recrystallized into something more amorphous and solid). Compaction with shale production is unlikely to occur.

  22. “The Richter scale is logarithmic, that is a double in magnitude means the quake is a squared higher energy than the next-lower number.”

    The rihter scale is logarithmic base 10. So a magnitude 3 is 10 times the amplitude of a magnitude 2.


  23. Typo – Richter scale.


    • Horatio Algeranon

      It’s “righter” too, than the “Wronger Scale”, which is blogarythmic (and used at WUWT)


  24. Lifetime Ohio resident.

    Either your data is incomplete or your reporting of it is intentionally lacking data conflicting your premise.

    In other words, why do you not have any dots in Central Ohio? There have been many reported and recorded 3.5+ earthquakes here in the last 40 years.

    [Response: If you think there’s a problem with the data I suggest you notify those who supply it at the U.S. Geological Survey. Your suggestion that I have intentionally omitted data conflicting my premise is wrong, and actually despicable. If you have any sense of shame, you’ll apologize.]

  25. I was living in Ohio when on 7-27-80 at 9:48 am EST I experienced a 5.1 magnitude earthquake that was centered south of the Ohio boarder near Cincinnati. I kept the article that reported it. The recording station was at John Carroll University (University Hts/ Cleveland Ohio)- I wonder what they would have to say about earthquakes frequency since they were the data gathers for earthquakes in the Ohio region during the late 70’s and 80’s (I don’t know if they still collect the data or not.)

  26. A Prof Richard Davies of Durham Uni is appearing on UK media saying there have only been 3 fracking earthquakes of any significant size -one each in the US, the UK and Canada. The biggest at Horn River Basin in Canada in 2011 had a magnitude of 3.8.
    It all sounds rather too definite to me but he must know what he’s talking about because he used to work for ExxonMobil.