Jim Lehane at The Geology P.A.G.E. has set this month’s accretionary wedge task of “What geological concept or idea did you hear about that you had no notion of before (and likely surprised you in some way)”.
I’m going to go back nearly thirty years for this one, to when I was looking for a PhD topic. In the UK in those days geology departments produced lists of potential topics which were submitted to the Natural Environment Research Council (NERC) for approval. NERC would then provide a list of approved topics and a number of grants that they were willing to support, which was fewer than the number of topics approved. As a undergraduate student, one would apply for a particular topic, not knowing if it was actually going to be funded.
I studied for a degree in geology at Imperial College specialising in geophysics in my final year. In the list of approved PhD topics for University College, Cardiff (now Cardiff University) there was a project on “Geophysical Investigation of Small Scale Geological Problems”. This was ideal for me and I applied, only to receive a reply stating that this was one of the projects that they would probably not fund, but would I be interested in applying for a project on “Deep Earthquakes” instead. To be honest, I wasn’t that keen as I was more interested in doing practical, hands on, field geophysics but in order to keep my options open I replied saying yes. In retrospect, I should have gone for Cardiff’s other funded project which involved collecting and analysing palaeomagnetic samples from New Zealand, a place I’d still love to visit. However, after being rejected by Cambridge and Durham for other potential projects I travelled down to Cardiff for interview and to my surprise was awarded the PhD grant. I now had to find out about how deep earthquakes happened, something I had not heard about, certainly surprised me, and still baffles me somewhat today.
Shallow earthquakes are relatively straight forward. Stress builds up in a block of rock containing a fracture, whose two sides are held together by friction. Eventually the stress overcomes the friction and the two sides either side of the planar fracture move past each other, releasing the energy that had been stored previously as elastic deformation in the rock mass neighbouring the fault. The system of forces acting at the source is well known and described as a “double couple”. This gives (amongst other things) the classic quadrifoliate pattern of compressional and dilatational P-waves which can be used to work out the orientation of the fault causing the earthquake.
The model is fine for shallow situations where the rocks are brittle. However, the temperature increases by about 30° C for every kilometre you go down. In areas like California where heat flow is moderately high, by the time you get to about 15 kilometres down the rocks are too soft to deform in a brittle fashion and instead flow plastically. In intraplate areas like the UK where the heat flow is less, the brittle-ductile transition is just over 20 km.
In the shallow part of subduction zones, cold lithosphere is subducted and so cold, brittle material can be found at larger depths and earthquakes within the upper parts of the Wadati-Benioff are still explainable. Earthquakes in subduction zones tend to tail off around 200km down but in some start to reappear around 400 km and continue to the bottom of the upper mantle at 660km. It is these deep earthquakes that are particularly problematic as even in the heart of the subducted slab temperature will be well above the brittle-ductile transition. Brittle shear earthquakes simply can’t occur at these depths.
The depths at which the earthquakes switch off, on and off again (220, 410, and 660km) appear to correspond to changes in mineral structure due to the increased temperature and pressure with depth so this would strongly implicate mineral phase changes in deep earthquake sources. Down to 220 km the mantle is olivine + pyroxene [enstatite], from 220 – 410 km olivine + garnet [pyrope]. At 410 km olivine transforms to ringwoodite [the spinel form of olivine]. At the lower-upper mantle boundary at 660 km there is a further phase change to silicate perovskite (Mg,Fe,Al)SiO3. Since the lower mantle comprises 55% of the Earth by volume this makes silicate perovskite the Earth’s most abundant mineral.
The task of my PhD was to investigate the source mechanics of deep earthquake to see if they had the same double-couple mechanism as the shallow earthquakes or whether they had a non-double-couple source and, due to mantle phase changes, to see if there was a volumetric change component to the source.
Most earthquake source mechanism determination was in those days, and still is today, undertaken using inversion methods which produce a best fit between the observed seismograms and synthetic ones calculated by computer. Some of the inversion solutions for large deep earthquakes had shown a significant non-double-couple component to their source, suggesting that the earthquakes may not be due to planar shear.
The technique that I was to use involved forward, rather than inverse modelling. It had been devised by my PhD supervisor for discriminating between natural earthquakes and underground nuclear explosions. By making measurements of the amplitudes of certain phases (P, S and the near-surface reflections pP and sP) with error bounds, all the possible source types and orientations of those sources that are compatible with those observations are saved. By combining the compatible sources for earthquake records from several recording stations you end up (hopefully) with only a very small range of possible sources and orientations.
The image in the header shows P, pP and sP phases together with a rare S-to-P conversion from the lower-upper mantle boundary S660P
It should be remembered at this time that digital waveform data was in its infancy and my earthquake seismograms were taken from analogue records. These were kept on microfiche in the basement of the British Geological Survey in Edinburgh (next to the core store so it was always cold, even in summer). The correct microfiche had to be found for the right day, at the right time (decoding the time pips by eye – these are the white blobs next to the seismogram) at the right station, for the right component. The record was then blown up on a fiche reader and printed on a wet photocopier (so one finished up the day reeking of petroleum spirt). The earthquake data was also a minimum of two years old as it took that long to collect the records from the recording stations, copy them to microfiche and then distribute those copies. The contrast to today when I can watch earthquake seismograms from around the world on my computer desktop, pretty much in real time, could not be more stark.
Anyway, the upshot of the study was that I never found a deep earthquake that could not be explained with a non-double-couple or any significant volume change. They were all double couple, exactly like the shallow ones. Those earthquakes that had inversion mechanisms suggesting non-double-couple sources on closer examination of the actual seismograms revealed that they were actually multiple earthquakes with significant pulses of energy several seconds apart. I could analyse these sub-events and found that they were double couple but with slightly different orientations. This multiple shearing with different orientations was confusing the inversion solutions which at the time only used long period waveforms and could not distinguish the sub-events giving apparent non-double-couple solutions.
I could also use a similar method using the phase pulse lengths instead of the amplitudes to estimate the speed of the rupture and shape of the rupture surface. The up-going phases are much higher frequency than the down-going ones, showing a strong doppler shift. This means that the ruptures are travelling very fast, certainly faster than the speed of shear waves through the material. This is something different to shallow earthquakes, where rupture velocities supersonic to S-wave velocity are impossible.
So we have a geological conundrum. How is a material that should flow plastically accumulating enough stress to generate a magnitude 8.3 earthquake such as the one that occurred on June 9, 1994 636km beneath Bolivia and generates a shear mechanism indistinguishable from a shallow earthquake (other than perhaps by rupture velocity)?
There has to be some processes (probably involving mineral phase changes) that can cause some shear instability runaway condition that generates a supershear, rupturing at fast velocities generating deep earthquakes in a plastic material. What that process is uncertain, and something we may never know.
This work is over 20 years old and I’ve not been keeping up with the literature, so apologies if some/all of this is out of date. This was never written up other than in the PhD Thesis which maybe still available from the University of Wales