Investigating the erection of a lava spine at Mt. Unzen volcano, Japan

It's time to dust the cobwebs off this blog! This post is a guest post from my friend and colleague Adrian, a fellow volcanology postgraduate here in the University of Liverpool. He is writing about his research on Unzen volcano in Japan, which was published last week. Enjoy! 

My first paper, created in collaboration with Oliver and a host of co-authors from the University of Liverpool, Ludwig-Maximilians University in Munich and research institutions in Kochi, Nagasaki and Tsukuba, Japan, looks at the mechanics of a spine growth episode at Mt. Unzen volcano, Japan.

Left: The 'Tower of Pelée' spine at Mt. Pelée, Martinique, 1902. Photo credit: unknown.
Middle: The 'whaleback' spine at Mount St Helens in 2005. Photo credit: USGS, taken Feb 22 2005. 
Right: The remnant of the spine at Mt. Unzen. Photo credit: Yan Lavallée. 

Lava spines are a spectacular feature of some lava dome volcanoes. They appear as massive, cohesive blocks that pierce through the dome surface, exposing the dense magma that plugs the volcano at shallow depths. Spines can take various forms, from the curving ‘whaleback’ spines observed during the 2004-08 eruption at Mount St Helens, USA, to soaring monoliths, such as the famous ‘Tower of Pelée’, a 300 m tall spine extruded following the devastating eruption of Mt. Pelée, Martinique in 1902. Spines often form prior to periods of strong activity or during the waning phase of an eruption, as in the case of Mt. Unzen, where a 30 x 60 m spine grew during the final months of a four-year eruption, from October 1994 to January 1995.

Rising magma becomes increasingly rigid and viscous at shallow depths as crystals and bubbles form and gas is liberated (imagine forming honeycomb from syrup). Growth of a lava spine takes place when this dense, rigid magma forming the plug is pushed out through the dome surface by pressure from below, sliding against the surrounding rock (wall rock) along faults. These faults cause detectable volcanic earthquakes, and rapid movement along the steep, shallow faults changes the local rock properties, providing a feedback for further extrusion of the spine.

Left: The experimental rotary friction apparatus in the University of Liverpool.
Top right: Frictional melt forming during an experiment in Kochi Japan.
Bottom right: An illustration of the sample set up for the experiments.  

We investigated the spine growth at Mt. Unzen by simulating these faults in the laboratory using experimental rotary friction apparatus in Liverpool – pictured above – and in Kochi, Japan. Faults are replicated between cores of dome rock, with pressure applied between two facing cores, before rotating one core against the other (see the video below). Our high-velocity friction experiments showed that faulting of Mt. Unzen dome rocks generates substantial heat. In fact, at slip rates greater than 0.4 m/s, enough heat was generated to melt the samples at the slip surface, forming a frictional melt that caused dramatic changes in the measured slip dynamics. Frictional melts have been described as far back as 1916, where the slip zones of ancient crustal earthquakes have been exposed, however their presence and significance in volcanic environments has only recently been demonstrated.

When a frictional melt forms between the sliding samples in our experiments, it induces a rapid increase in friction along the fault. This effect is particularly strong at experimental conditions relevant for shallow depths. The increased friction may ‘brake’ and stop the sliding rocks, and prevent all the faulting stress from dissipating, encouraging repeating ‘stick-slip’ sliding behaviour. In the context of growth of a spine at Mt. Unzen, this would cause many repeating earthquakes during the period of spine growth.

A typical few hours of the earthquake record during the extrusion of the spine. 

Analysis of the earthquake record, led by Oliver and Silvio De Angelis at the University of Liverpool, revealed just these patterns during spine erection, with volcanic earthquakes occurring with great regularity over ~40 hour cycles. Grouping these signals by the similarity of their waveforms revealed two particularly large groups of very similar volcanic earthquakes. This was a crucial as it allowed us to describe the spine extrusion as individual slip events through the characteristics of these two earthquake groups. Using these waveforms, we were able to estimate how far, and how fast each slip event was moving. In short, they were moving fast enough to generate frictional melt.

These high-velocity experiments were done at room temperature, but temperatures inside the dome can be up to 1000 °C. Therefore, to apply our results to the spine at Mt. Unzen we modeled what happens at these temperatures. What we find is that the heating caused by each slip event would be enough to make the frictional melt at more than 500 m depth.  Our findings show that frictional melting in volcanic settings may be much more common than previously considered, and its effect on fault dynamics and temperatures are dramatic, therefore the formation of volcanic frictional melts must be considered within future studies of lava dome activity.

Post-experimental image of a sample, showing the light grey band of frictional melt sandwiched between the dome rock samples.

I am running out of time and space, so I’ll save the description of texture and viscosity of the frictional melts for those interested enough to read the paper – the image above is just a teaser! These observations, and more, are presented in the paper ‘Spine growth and seismogenic friction at Mt. Unzen, Japan’ published online in the Journal of Geophysical Research. The paper presents a more holistic approach to geomechanical investigations through incorporation of experimental and seismic data analysis together with rheological and geophysical modeling of fault processes. If you would like to know more please contact Oliver or me and we will be happy to send a copy of the paper! 

Note from Oliver: The seismic dataset has proven to be highly revealing, and a new study led by me will soon (hopefully!) publish in-depth analysis beyond the scope of this paper – watch this space!