Seismic triggering of volcanic eruptions: from Darwin to today

My last post on possible triggers for volcanic eruptions left a hanging question: why do seasonal variations in crustal loading seem to have an effect, when those resulting from tides do not? It’s not like the forces and strains involved are any larger – if anything they’re generally smaller. The key seems to be the timescale over which they are applied. Modern views of magma chambers are moving away from the idea that they are a large, homogenous blob of molten rock; instead the magma is “mushy”, with small, interconnected pockets of melt within a more solid matrix. This structure means that the molten parts are insulated from the effects of a short, sharp force, which is absorbed by elastic deformation in the solid parts; only a force consistently applied over a substantial length of time will cause the pockets of melt to be squeezed and moved around. See here for more discussion of these ideas, with reference to an Alaskan volcano.

Based on that, you’d expect the effect of earthquakes on the timing of volcanic eruptions to be negligible too – the passage of seismic waves through a magma chamber is a short, sharp shock rather than the longer term change which seems to be required. Evidence that this might not be the case came to my attention from an unusual source -from reading This Thing of Darkness by Harry Thompson, which is based around the voyages of the Beagle, including but not restricted to the one which Darwin sailed on. When I’m reading (or watching) fiction my scientific interest is usually only piqued when I notice a stupid error, or am confronted with a lame misunderstanding of an entire field (I am then forced to stop reading and engage in breathing exercises to control my blood pressure). In this case, however, I was fascinated by a passage in Chapter 22 (pages 366-67 of my edition) which describes how, as the Beagle sails up the west coast of Chile on the evening of the 20th February 1835, the crew observe the eruption of the Andean volcano Osorno. Further in the distance, two more volcanoes also start erupting, and later the Beagle is shaken by the passage of a tsunami, which we later find out was caused by a massive earthquake which flattened Concepcion. The clear implication that multiple volcanic eruptions along the Andean chain had been triggered by this earthquake

Now, I knew that Darwin had experienced a very large earthquake whilst he was in Chile in 1835, but I had never heard anything about this associated volcanic activity. Fortunately, most of Darwin’s eyewitness testimony is available online at repositories such as this, so we can check how close this fictional account is to the actual sequence of events. It turns out that although the author has compressed timelines, distances and observers for dramatic effect, there is a kernel of scientific truth in this passage. Here’s the actual sequence of events:

  • The eruption of Orsono described in the novel matches an eyewitness report by Darwin, but occurred a month before the Concepcion earthquake, on the night of the 19th/20th January. He later learnt that Aconcagua*, 480 miles further north, and Cosiguina, 2700 miles further north still, had also been active (these eruptions were far too distant for Darwin to observe directly).

  • Darwin and the Beagle then headed north to Valdivia, 250 miles north of Orsono and 200 miles south of Concepcion; they were still there on the 20th February when the big earthquake hit (the Beagle did not reach Concepcion until March 4th).

  • However, Darwin sent a request back south for accounts of the earthquake in southern Chile, and received much of interest in reply:

    At the moment of the shock, Osorno threw up a thick column of dark blue smoke, and directly that passed, a large crater was seen forming on the S.S.E. side of the mountain; it boiled up lava, and threw up burning stones to some height, but the smoke soon hid the mountain…

    …The action of Minchinmadom (a neighbouring volcano) was similar to that of Osorno: two curling pillars of white smoke had been observed all the morning; but during the shock, numerous small chimneys seemed to be smoking within the great crater, and lava was thrown out of a small one just above the lower verge of the snow.

    After the shock, the snow around craters on Corcovado and Yanteles, other nearby volcanoes, appeared to have melted, indicating minor activity.


  • The same source (a certain Mr. Douglas) also reported similar phenomena 10 months later:

    on the night of November 11th… Osorno and Corcovado both burst out in violent action, throwing up stones to a great height, and making much noise. He subsequently heard, that on the same day, Talcahuano, the port of Concepcion, little less than 400 miles distant, was shaken by a severe earthquake.

These observations certainly suggest some sort of connection between large earthquakes and volcanic eruptions. Perhaps surprisingly, although there’s the odd anecdote here and there, I found only recent two papers which really get to grips with this phenomenon. The first is a 1998 paper by Linde and Sacks [1] (which just to show how active this area of research has been, cites Darwin), in which they perform a global survey looking for correlations between the timing of volcanic eruptions and nearby large earthquakes. And find one they did: on days where there was an earthquake of magnitude 8 of larger, there was a significant peak in volcanic eruptions of VEI>2 within a 750 km radius of the epicentre. For earthquakes of magnitude 7-7.9, the increase was restricted to a smaller radius of 200 km. However, interesting as this study is, in terms of predicting eruptions and hazard assessment it’s not amazingly useful. It tells us that large earthquakes might tip nearby volcanoes already on the verge of erupting over the edge, but it doesn’t help us identify which volcanoes those might be.

The second paper I found tries to answer the potentially more useful, but much more complicated question of whether the longer-term risk of a volcanic eruption can be affected by nearby large earthquakes. Warner Marzocchi and his group certainly think so: after comparing the timings of the 8 20th century eruptions with VEI>5 with those of all large (Ms>7) earthquakes over the same period, Marzocchi [2] claimed that these eruptions could be linked to large earthquakes less than 1000 km away occurring either 0-5 or 30-35 years earlier. I have to say I’m less convinced by these results, simply because when you’re comparing a mere 8 volcanic eruptions to more than 700 earthquakes, the possibility of spurious correlations can’t be easily ruled out, especially in a system where numerous other factors, not considered in this study, could also be affecting the volcanic system. What I’d like to see is evidence that the precursors to these large eruptions – such as increased local seismicity or subsidence/uplift patterns suggesting magma movement – could be linked to a nearby earthquake. Given its temporal and spatial proximity to the Boxing Day 2004 earthquake, I wonder if anyone has done that for Mount Mayon?

Whatever the case, we are left with a mystery. If the forces exerted by the tidal cycle are too rapidly varying to really affect the stress state within a magma chamber, how is it that the much more rapidly varying seismic waves apparently do? The answer might lie in the fact that many magma chambers, especially those belonging to volcanoes above subduction zones, contain a third phase in addition to solid and melted rock: gaseous volatiles, such as carbon dioxide and water, in the form of bubbles within the melt. The pressure changes associated with the passage of seismic waves could be enough to shrink or expand these bubbles as volatiles are forced in and out of solution with the melt, which could potentially alter, or trigger, the movement of magma.

To a certain extent, though, almost 180 years after Darwin first made his observations, we’re still scratching our heads.

*I can’t find a volcano with this name in the Smithsonian database, so either they’ve renamed it or this information proved to be false.
[1] Nature 395, p888-890, 1998.
[2] Journal of Geophysical Research 107, 2002 [doi].
NB As for This Thing of Darkness, it was a fascinating read, but because (I think) it is narrated mainly from the perspective of Robert FitzRoy (the Beagle’s captain, about whose life and work I was glad to gain a greater appreciation), the picture it paints of Darwin himself is somewhat unflattering.

Categories: earthquakes, geohazards, geology, paper reviews, volcanoes

Comments (5)

  1. Neat post. I look forward to hearing more of your volcano adventures in New Zealand and elsewhere.

  2. yami says:

    AFAIK there were no volcanic responses triggered by the Boxing Day earthquake – nor the 1964 Great Alaska. There was a paper in GRL a short while ago, looking at two Javanese volcanoes responding to regional earthquakes (ah, here it is). Satellite monitoring of overall eruption temperature/volume is a big improvement over the motley assortment of eruption start times we’ve got currently.
    Also, I don’t know if you saw Mang & Brodsky’s review paper or not, but they cite a few more analyses, mostly of individual volcanoes responding to local tectonic events.

  3. John Wilkins says:

    That was a very nice post, Chris. Now can you address the claim made (I saw in on television so I can’t source it) that there were a chain of volcanic eruptions correlated with seismic activity in the period from c1400BCE-900BCE along the northern edge of Asia minor and in the Mediterranean? What I want to know is whether they have the same distal cause – tectonic stresses- whether they directly cause one another.

  4. Matthew says:

    Aconcagua is the tallest mountain in South America, it is not a volcano (though it might be a remnant of a very old, extinct volcano). Darwin mistakenly thought it was an active volcano.

  5. Chris Rowan says:

    John – the Med is quite active tectonically, because it’s a very complicated plate boundary zone between Africa and Europe, which are still slowly colliding with each other. For example, there’s a subduction zone south of Crete which is responsible for generating most of the Greek Islands, including Santorini. Unless you’re talking about some specific correlation? I’m not sure how you’d establish such a thing…
    Matthew – yes, that’s the only Aconcagua I could find, but as well as not being an active volcano it seemed to be too far north to match Darwin’s description.