Oilquakes update (updated)
A few quick notes:
In contrast, the Greeley earthquakes died down while the injection company stopped injecting and plugged the part of the well that was directly connected to a fault. But as they resumed injecting, a M2.1 showed up, so this probably isn’t the end of the story…
And finally we get the odd news from USGS scientist Sue Hough (why does the WSJ deny her her doctorate?–she is quoted as “Ms. Hough”)) that induced earthquakes seem to cause less shaking for a given magnitude than tectonic earthquakes (and by the ways, if you see headlines including the phrase “fracking-induced earthquakes” then you know the headline writer is ignorant, and if it is in the article, then the writer is ignorant; these earthquakes are induced by wastewater injection). Something seems kind of fishy, but as the article doesn’t seem to be online yet (presumably will be posted later today); right now this seems most likely to GG to be because all the attenuation models are keyed for earthquakes in the basement rocks and not in the more attenuative sedimentary rocks, though there are hints in some of the online stories that Dr. Hough (we’ll give her her doctorate) proposes that the stress drop is lower because of fluid pressures, which in turn reduces the peak accelerations. Anyways, hard to be sure when working from an AP article; will update when the BSSA article goes online.
Update is in the full post.
Update: The paper is out. The hints were correct; Hough is thinking that the main cause of this discrepancy is a lower stress drop, which is hardly an unreasonable idea. This should be directly testable with some decent instrumental records (and we should have a few of those from Earthscope stations that were in the vicinity of some of these events). One could argue that this is revealing issues with the empirical intensity-distance drop off curves for shallow events, and that should be on the table. Because all of these events’ depths are poorly known (both tectonic and induced), we probably can’t do much more with this dataset. There is an argument in the paper against intrinsic attenuation, but no consideration was given to propagation effects. Since most of the Did You Feel It responses are thought to be from the higher frequencies in the wave train, presumably body waves are the main carrier (if all the waves were attenuated, the magnitudes would drop with the intensities). It is possible that the transmission coefficients going from the shallow events downward are low (and reflection coefficients high), which would mean more energy would be trapped near the events and less would extend out as refracted energy; this basic pattern is indeed observed.
So an alternate hypothesis would be that the discrepancy observed is due to event depth and not event cause. This would also resolve the puzzle posed by Hough about why there aren’t some tectonic earthquakes caused by the change in the stress field from the induced earthquakes: they could be there if depth is really the key factor in affecting ground shaking and not high fluid pressure. This is actually something important to determine because Hough suggests using this magnitude-intensity discrepancy as a means of discriminating between induced and tectonic earthquakes. A quick look at the moment tensor inversions from Saint Louis University reveals that if events are very shallow, the computed magnitudes are significantly smaller than if a bit deeper (this is because regional solutions are sensitive to surface waves, which are more easily generated the closer an event is to the surface); this could contribute about 0.3 magnitude units to the 0.7-1.3 magnitude unit discrepancy reported by Hough if events are systematically being located a bit too deep.
Although the news stories carry the “these earthquakes produce less shaking” message, it is important to note that close to the earthquakes, where there are far fewer entries in the Did You Feel It catalog, the shaking rises well above that predicted by the intensity models used (contrast Fig. 2 and 3 in the paper). As these are the places that will have the most damage, it is misleading to characterize the overall shaking as less and thus imply that damage from these events would be less (which is the spin of most of the news articles GG has read). Yeah, folks 100 km from a M5.5 might only wobble around as much as a tectonic 5.0, but those folks near this event would get clobbered just as much–if not more–by an induced M5.5 earthquake than a tectonic M5.5.
So how do we resolve this? Ideally it would be by getting good depths (within +/- 1 km) and ground shaking measurements (or, better, near-source high sample rate seismograms) on both tectonic and induced earthquakes. In many places, we have decent numbers on the depths of induced events (Paradox Valley in Colorado has pretty good depths; we should see some good depths from the Trinidad sequence. The small earthquakes near Greeley are being well located but are too small for this kind of analysis). It is possible that by incorporating the surface-wave based tomography from groups like the Ritzwoller group at CU that the uncertainty of regional waveform solutions (such as the SLU ones) could be reduced, though a lot of problems still exist in the very shallow subsurface, which is critical for this work. Or we might get lucky and have a really shallow tectonic event somewhere where there are enough instruments to know it is really shallow (good luck with that; for body wave depths you want a station within a focal depth–which basically means within a couple kilometers of a really shallow event–that would be quite the coincidence with the sparse array of seismometers in this region; for surface waves you need a high resolution velocity model and a good focal mechanism between source and receiver to remove ambiguity). So Grumpy here predicts a lengthy interval of back-and-forth on this particular idea…