Wednesday 5 February 2014

What's that coming over the hill...?

Understanding how pyroclastic density currents behave through time and space.
by Rebecca Williams (@volcanologist)

Pyroclastic density currents are flows of searing hot gas, ash and rocks that swoop down the sides of a volcano during the most violent eruptions. They can travel at speeds of up to 450 mph and can be as hot as 1000°C. Historically, they have been responsible for over 90,000 deaths, and so are the most deadly volcanic phenomenon.  In order to try and reduce the risk to people from these hazardous flows, we strive to understand how they behave. The trouble is though, is that they are very difficult to observe. Even small flows, such as those recently seen at Sinabung (Indonesia) and Tungurahua (Ecuador) volcanoes are hidden in a cloud of ash. The largest types of these currents however, have not been observed at all, and no instrument exists that can be deployed (and survive) to record them. So, we try to understand the processes in these currents by studying their deposits.
A typical ignimbrite as found on Tenerife. You can study the changes in size and quantity of pumices (yellowish, bright bits) and lithics (dark bits) to infer current dynamics. You can also interpret sedimentary features such as low-angle cross bedding (on a level with the scale) or if the deposit is massive (as it is towards the top of the photo). Photo credit: Rebecca Williams
 The deposit of a pyroclastic density current is called an ignimbrite. We can infer quite a lot about a current by studying an ignimbrite – such as which direction it was going, whether it was turbulent and dilute or whether the flow was dense and more full of blocks of rocks than it was gas and ash. At a single location, we can study the vertical section through a deposit and see how the ignimbrite changes from the bottom to the top. This tells us how the current was changing from its start (the bottom of the deposit) to its end (the top of the deposit), as the deposit builds up through time, recording the changing dynamics of the flow. What we need to know though, is how the current behaves through time everywhere it flows. Did it flow down that valley at the same time that it flowed down this one? Did it go over that big hill at the start of the eruption, or did it have to bury some of the topography to make it easier to flow over? Of particular interest are the very large flows that form a circular deposit around a volcano – are these formed by a radially expanding current that flows in all directions around a volcano all at once? This is much harder to do: there is often nothing in the deposit that tells us where we are in time in the current's duration. So, we can’t join up all those vertical sections to understand how the current behaved through time and space all around the volcano. That is till now, as published in the February issue of Geology.
A 'typical' section of the Green Tuff Ignimbrite - it's not typical! You can see it doesn't look like the Tenerife ignimbrite - that is because it is welded. The ash and pumice were so hot when they were deposited they welded together into glass. You can still see lots of primary sedimentary features though, including imbrication as seen here - the current was flowing from right to left. Photo credit: Rebecca Williams

The study here maps out the chemical composition of an ignimbrite which was deposited from a single, sustained pyroclastic density current and reveals how the current behaved through time and space. The ignimbrite used in the study, The Green Tuff Ignimbrite on Pantelleria, Italy, is a circular ignimbrite. It is chemically zoned, that is the chemical composition of the deposit is different at its top than it is at the base, and the composition changes gradually between the two extremes. This is because the magma chamber that the current was erupted from was zoned. At the top of the chamber the magma was more evolved than the magma at the bottom. So, when the volcano erupted it progressively withdrew magma of gradually changing composition and this was recorded in the volcanic deposits. This then, gives us a timeline through those deposits – we can say whether that chemistry was erupted at the beginning, the middle, or the end of the eruption.

The type section of the Green Tuff Formation and representative graphs of the changing composition. Copyright Geology. Full figure caption:
  What we found in the Green Tuff was that the chemical change was gradual, so we were able to divide the unit into 8 different compositions, and these were equivalent to 8 different time steps through the eruption. Once we’d determined this chemical stratigraphy, and defined it at a type locality (a place that records the entire compositional change) we then went and mapped the entire ignimbrite. I spent over 6 months spread over 3 years logging, mapping and sampling the Green Tuff Ignimbrite and analysed over 500 samples to determine their chemical composition. I then input all this data into ArcGIS so I could see what the spatial extent of each of the 8 compostional zones were. I produced maps of these different zones and these maps represent the footprint of the current that formed the Green Tuff at a snapshot in time during the eruption.
Footprint maps of the Green Tuff pyroclastic density current based on zironium (Zr) compositional zones. Copyright Geology. Full figure caption:
By analysing these maps we realised that, contrary to assumption, the current did not flow in all directions all at once. Instead, at the start of the current's duration it didn’t get very far at all, and only flowed in certain areas around the island. As the current continued it flowed further in more directions, but was often deflected or reflected around topographic barriers such as hills and old caldera walls. However, at the current's peak, it DID flow in all directions all at the same time. If there had been humans living here at the time, there would have been no escaping the devastating currents. The current continued, but the directions it travelled in decreased and it wasn’t able to go as far. It stopped being able to top over those hills and eventually decreased til it was able to flow only short distances from the vent before stopping altogether. We think that this all happened within about one and a half hours.

This has changed how we think about pyroclastic density currents and will impact on the way we model them for hazard assessments. We should think about these currents as dynamic, and able to change rapidly – each of the 8 zones represents only around 11 minutes. A current that was initially quite slow, restricted to a small area and unable to flow up hills, changed and was able to flow in all directions and over 500 m hills in less than an hour. 

The Green Tuff ignimbrite drapes the La Vecchia caldera wall - both the inward dipping caldera scarp and the seaward-dipping older stratigraphy - at Scauri, Pantelleria. Photo credit: Rebecca Williams
On the other hand, without knowing that the current changed through time, we might have overestimated how big it was. You can use the maximum distance that the current travelled and the volume of the deposit to estimate how much material was coming out of the volcano with time (mass flux). If we had used the maximum values and assumed that was what the current was like for its entire duration, we would have massively overestimated the eruption's mass flux.

Zoned ignimbrites such as the Green Tuff are not rare, therefore we hope that this technique can be applied to other deposits so we can continue to advance our understanding of pyroclastic density currents. 

This blog is a summary of this open-access paper:
Williams, R., Branney, M.J., Barry, T.L., 2014. Temporal and spatial evolution of a waxing then waning catastrophic density current revealed by chemical mapping. Geology. 42, 2, 107-110.


  1. Really enjoyed your talk at Cheltenham Science Festival this afternoon. I've been fascinated by volcanoes for a long time. A quick question - I have heard of pyroclastic flows before, but never pyroclastic density currents. Are they the same thing? If not, what is the difference?
    Thanks - Sue

  2. Hi Sue.
    Thanks for coming to see us in Cheltenham! I really enjoyed giving the talk and chatting to people afterwards. I'm glad you enjoyed it.
    Pyroclastic density currents are essentially the same thing as pyroclastic flows. The change in name came about because of a profound change in our understanding of them. The old nomenclature was too wrapped up in the old understanding, so volcanologists decided at the time, that a change in name was the way forward. In hindsight, I'm not so sure this was a good idea, as it causes a lot of problems! If you're interested in the whys of the name change and the ongoingdebate, I wrote about it here:

  3. Thanks for the explanation, and the link to the fascinating discussion about terminology. I suppose, as a member of the general public whose limited knowledge comes from books and the occasional TV programme, I am guilty of assuming that these terms are fixed. Good to have a better insight.
    Going back to your talk at the Science Festival; can you explain what the demonstration was intended to show. I confess that, although it was all very dramatic and exciting, I didn't 'get it'.

  4. Oops, thats our bad explanation. I think we were taken a bit by surprise by the second demonstration.
    The point was that we made some lava. Lava on its own contains some gas, which as we explained with reference to Laki, can be damaging to life and the atmosphere. Scale this up to a Large Igneous Province (LIP) and you have the potential for a mass extinction. But, not all LIPs are related to mass extinctions, why not?
    The first experiment we erupted our 'lava' over granite - an igneous rock. And whilst, pretty, not much happened. The second experiment, we erupted our lava over coal - and you saw the dramatic effects. Lots of carbon released as smoke.
    If you add the carbon released by burning coal to the carbon dioxide released during an LIP eruption, you may well have enough gas to trigger a mass extinction. So maybe its the rocks that the lava is erupted through and onto that's important. Ineed, the Permian-Triassic mass extinction has now been correlated with layers containing fly ash - and we certainly produced a lot of that in our second experiment!

    I've put a link to videos of the practise runs of the demo in the latest blog I've written, where I also explain what they are showing.

  5. Thanks for taking the time to explain, and to put up the videos - much clearer now. Glad to hear that you enjoyed Cheltenham. Good luck with your future research, and hope to see you at Cheltenham Science Festival again sometime.