It’s been almost two centuries since a possible link between large earthquakes and nearby volcanic eruptions was first proposed by none other than Charles Darwin, who compiled accounts of increased activity at a number of Andean volcanoes in the wake of the 1835 Concepcion earthquake. In the decades since, however, firm evidence of earthquakes triggering volcanic eruptions has remained somewhat elusive; but a new paper by Sebastian Watt and colleagues at the University of Oxford takes a fresh look at Darwin’s old seismic stomping grounds, and claims that spikes in volcanic activity in the Andes can indeed be observed in the months following large earthquakes.
The study focusses on the southern Andean subduction zone off the coast of Chile. The earthquakes are from the eastward subduction of the Nazca plate beneath South America (the blue line marks the plate boundary); here, water driven from the downgoing slab triggers melting of the surrounding mantle and generates a volcanic arc (the orange blobs shows the locations of volcanos known to be active in the last 10,000 years, courtesy of the Smithsonian Global Volcanism Program).
Watt et al. have compiled records of seismic and volcanic activity in this are stretching as far back as the 1500s, which included 16 earthquakes with a magnitude of more than 7.5, and more than 250 eruptions from 25 volcanoes. However, the record prior to 1850 is more poorly documented and likely to be incomplete. The plot below charts the number of eruptions that initiated in each year since 1850, with the five large earthquakes that occurred in this period also plotted (click for a larger version).
This plot shows that 2 of these earthquakes – in 1906 and 1960 – appear to have been associated with spikes in volcanic activity, with six or seven volcanoes starting to erupt in the following 6-12 months. Even though the eruption record as a whole is fairly variable and ‘spiky’, these peaks still seem a little unusual, standing out from the normal variation between 0 and 4 (and, very occasionally, five) eruptive events a year.
1906 and 1960 stand out statistically too: in general, the records follows an exponential distribution, with there being many more years with 0 or 1 eruptions than there are years with 3 or 4 eruptions (the average rate is 1.32 eruptions per year). The best-fit exponential distribution for the whole record predicts that there should be only one year with more than 6 eruptive events every 500 years or so, whereas this record not only has two such years in the last 150, but these years also both immediately follow large subduction zone earthquakes – an association that should occur once every 2,500 years or so if there was no connection between the the earthquake and the period of high volcanic activity. So the association does seem to be more than a coincidence, especially since the less reliable record between 1550 and 1850 indicates at least one more year – 1751 – where 6 eruptions occurred in the year following a subduction zone earthquake.*
It’s important to note that this study potentially establishes a statistical, rather than a phenomenological, link: that is, you can’t point at any of the 7 volcanoes that erupted in the year after the 1960 earthquake, and say “that one was triggered by the earthquake, so was that one, but that one would have erupted anyway”. This is because all of the volcanoes involved were probably already on the verge of erupting, and the earthquake somehow just provided a little extra push over the threshold. This need for volcanoes to already be ‘primed’ to erupt before they can be triggered by seismic activity probably explains why not every great earthquake on the Chilean subduction zone is followed by excess volcanic activity (as is the case in 1928, 1939 and 1985). For there to be a measurable effect you need a reasonable number of volcanoes with full magma chambers, which depends on factors like the rate of magma generation and ascent in the subsurface, and (possibly) the time since the last big earthquake.
However, this study does provide at least one clue about the actual processes behind this linkage, because the effect seems to stretch at least 500 km from the epicentre of the triggering earthquake, far beyond the distance where permanent changes in stress associated with deformation at the plate boundary are going to have a measurable effect on a volcanic magma chamber. This suggests that it is the passage of seismic waves generated by the earthquake that is somehow pushing the volcanoes over the brink, even if it sometimes takes a few months for that effect to be felt at the surface. Right now no-one really knows exactly what mechanism is involved: Watt et al. discuss a number of possibilities, including seismic energy forcing gas bubbles out of the melt, or dislodging partially solidified, mushy chunks of melt from the walls of the magma chamber. As it stands though, whilst large regional studies like these are needed to firmly establish a correlation between earthquakes and volcanoes, only detailed investigations of individual volcanoes are going to provide insight into the physical processes that may cause earthquake triggering.
*Rather ironically, the 1835 earthquake which prompted Darwin’s initial musings does not show up as having a particularly significant effect in this study, but this may be because the observations that he reported – which were mainly second-hand accounts, and therefore difficult to verify – were deemed not to be reliable enough to be included in the eruption catalogue.
S WATT, D PYLE, T MATHER (2008). The influence of great earthquakes on volcanic eruption rate along the Chilean subduction zone Earth and Planetary Science Letters DOI: 10.1016/j.epsl.2008.11.005