Gargling Eruptions

By Greg Valentine

Many explosive volcanic eruptions involve sustained jets of hot gases and magma fragments (pyroclasts) that exit a vent at supersonic speeds. Once these jets enter the atmosphere their dynamics can be quite complicated.  If the gas-particle jet mixes effectively with surrounding air, it can transition to a buoyant plume that rises to altitudes of 20 kilometers or more and disperses the pyroclasts by fallout over a huge area. On the other hand, if mixing is not efficient, the jet will collapse once its initial kinetic energy is spent, so that gas and pyroclasts fountain back to the ground and speed along it as devastating pyroclastic flows. Although this description sounds simple, after decades of research the volcanology community continues to explore and understand more and more complexities in these processes - much of that research was actually led by our own (now retired) Professor Bursik!

The largest explosive eruptions on Earth emit hundreds to thousands of cubic kilometers of pyroclastic material during a single eruption. Yellowstone is a famous example of this. Because so much magma is evacuated from its underground reservoir during one of these eruptions, the ground over the reservoir subsides and forms a depression, or caldera. The calderas have diameters from about 10 to 60 km or more. PhD student Meredith Cole compiled data from calderas around the world and showed that it is typical for about half of the volume of pyroclastic flow deposits to be preserved within the calderas, forming tuff deposits that can be more than a kilometer thick. The other half is spread out over the ground outside the caldera and is usually much thinner. A major conclusion is that the calderas subsided during - not after - eruption, and because pyroclastic flows pond within topographic lows, much of the erupted material gets trapped within its own developing caldera.

This, in turn, means that as a super-eruption proceeds, the gas-pyroclast jets that are coming out of the ground must penetrate their own ever-deepening deposits (see Figure 1). In other words, the eruptions .... gargle.

We used computational fluid dynamics to try to understand what the effects of gargling might be, compared to eruptions that discharge straight into the air. The computer code solves the equations of fluid dynamics - conservation of mass, momentum, and energy - for both gas and particles. We modeled a "half-space" in two dimensions with gas-particle jets erupting (in-rupting?) through a 50-100 m thick layer of particles.

Although there are many details to be learned from this modeling, the upshot is that an eruption through its own earlier deposits completely changes the dynamics.  Figure 2 illustrates this by comparing an eruption into clean air (A) with eruption through a 50 m thick layer of particles (B,C).  An eruption into clean air may produce a buoyant plume (no pyroclastic flows) as in A, while the same eruption through a relatively modest thickness of deposits (50 m) does result in pyroclastic flows.  Although we won't go into it here, the eruption behaves just as you would expect for a good gargle - it fluctuates or pulses in height and in turn this would affect the pyroclastic flows leaving the caldera.  Our numerical simulations are consistent with published stratigraphic data from detailed studies of pyroclastic deposits around calderas as well as the 1912 eruption of Novarupta in Alaska, which erupted through a 2.5 km-wide vent that was mostly filled with loose debris. Always nice when theory is consistent with ground truth in the form of deposits and eruption observations!

This work will soon appear in an article by myself and Meredith Cole in the journal Geology. Doing this work relied on the wonderful computing resources available to UB faculty and students through the Center for Computational Research.  

Much remains to be explored on this topic! Meredith, in particular, is conducting a study of a vent structure exposed in a now-eroded caldera in Colorado. The structure is filled with pyroclastic material that is similar to, and likely from the same eruption as, the thick caldera deposits that it cuts through. She will be quantifying the nature of the contact between the vent and surrounding deposits, potential mixing between them, and the geometry of the whole system. We will use the computational fluid dynamics approach to test various hyphotheses that might arise from her fieldwork. More to come....

Next article: UB Paleoclimate Dynamics Team