Yesterdays post dealt with some of the problems inherent in earthquake hazard assessment, today I’m going to focus on something a little closer to my heart – the problems in assessing volcanic risk. I will say straight away that I’m not going to deal with predicting eruptions here (at least not today), but instead the longer term risk assessment associated with identifying which areas around a volcano are most at risk from volcanic products. Eruption prediction is as much of a black art as predicting earthquakes is, and suffice to say we’re a fair way off.
Volcanoes come in many shapes and sizes and, depending on type of magma which feeds them, erupt in a wide variety of styles. Basically you can divide them into those which spew lava, and those which chuck out a lot of ash and pyroclastic flows. Contrary to what many people believe lava is actually fairly low risk – it’s slow moving and generally pretty confined. While it will destroy anything it comes into contact with, you can at least evacuate areas pretty easily. Pyroclastic flows on the other hand are very fast moving (<200 km/hr), highly destructive, hot (<700 °C) flows which in some circumstances can even go uphill.
These explosive ash-generating eruptions have the additional risk of lahars. These ash-laden flows are formed when heavy rainfall (often seeded by the huge particulate output of the eruption itself) picks up the ash deposited by the eruption and travel down the river basin. The video below gives you an idea of what they look like, but essentially what you ened to envisage is a flow with cement-like properties. They are highly destructive due tot he shear mass of material they move down a river catchment, and have the potential to inundate vast areas with a huge amount of sediment. Crops, buildings, bridges – in fact anything in their way – will be left in ruin or simply buried.
In actual fact, the very biggest risk associated with large explosive volcanoes is that imposed by the widely dispersed ash; the tendancy for it to blanket vast areas gives it the potential to destroy crops and buildings – in fact it’s rather beautifully summarised here. As far as roof collapse goes, the important point can be taken from this page:
Description Density kg/m3
New Snow 50-70
Damp new snow 100-200
Settled Snow 200-300
Dry uncompacted ash 500-1300
Wet compacted ash 1000-2000
Basically, once you’ve had a bit of rain after an eruption, the ash which has collected on roofs is 20 times heavier than an identical thickness of new snow would represent. The picture to the right is a very famous picture from the USGS demonstrating the effect of snow loading on buildings at Clark Airforce Base after the Pinatubo eruption of 1991. Actually, maybe even more instructive is the image below.
Anyway, I’ve kind of gone off topic. There are many dangerous things associated with volcanoes, and there’s a nice case-by-case demonstration available here. As far as hazard assessment goes, the ash risk tends to be dealt with as a regional problem. What governments are more interested in is the risk posed by the aforementioned flows. Lahars, pyroclastic flows and lava flows all behave quite differently, and numerous groups around the world are expending a vast amount of effort into trying to understand how these things behave. None of them are simple Newtonian fluids, and for none of them do we have a comprehensive understanding of their physics. As a result, researchers are working on modelling their behaviour both using computers (numerical modelling) and experiments (analogue modelling), tied in with field scale observations. My own Ph.D. was targetted at investigating how pyroclastic flows deposit and interact with the substrate they flow over.
What governments really want – as in the Earthquake examples I discussed yesterday – is a map of exactly who will be effected by a given eruption type and scale. The most basic form of hazard map for a volcano looks like a bullseye – a very simple series of rings drawn around the volcano, saying that x km from the vent is high risk, y km is low risk, and z km is safe. However, these have fairly low confidence levels, and people tend not to trust these as much. As a perfect example there were many problems during the evacuations of Montserrat, where a broad zone was evacuated and people refused to move because they didn’t trust the hazard assessment and felt it was too generalised. Furthermore, in areas of larger populations governments want to know specifically which areas of the population should be prioritised for movement.
As a result of these kinds of problems volcanologists are under a lot of pressure to develop more specific hazard maps. However, if one thing has been made abundantly clear to me in the last few years – and specifically in a wide range of sessions at IUGG last week – we really have no clue how these flows operate.
Take, for example, pyroclastic flows. We don’t understand what the inside structure of these flows is. Lab modelling is restricted to using fairly simple particles – certainly nothing as complex as pressurised magma pieces constantly degassing dissolved volatiles into the flow as it progresses. Practically no work at all has investigated the erosion of these flows. Erosion may be very important as it entrains extra mass, which in turn can increase the momentum of the flows. More significantly we have absolutely no idea of what the physics are that govern friction within the flow. In fact, the way most numerical models work is that you start with some known inputs at the top of the volcano, and you keep running the experiment, arbitrarily changing the value for internal friction until the final deposit looks a bit like what you see int eh field. In fact most models have to change the value of internal friction during the flow to get it to work. No one knows why, there’s understanding of what the physical explanation for such a thing might be. We just do it in order to try and match the known deposit.
We do not understand these flows. The hazard assessments produced are little more than looking at what has gone before and drawing a line around them. That’s all very well, but there have been several spectacular examples of where a flow passing down a valley has filled up obstacles, creating a smooth chanel for subsequent flows which go on to travel much further than any previous flow has been able to (e.g. the October/November series of flows from Merapi in 2010*). We are so far from effectively modelling features such as deposition and erosion (let alone flow motion itself) accurately that I wonder how meaningful these hazard maps really are, or whether in fact they simply provide a false sense of security and understanding.
Scientists are being pushed to provide more detailed hazard maps, moving away from simple ‘bullseye’ type maps, but in reality the models are simply not up to the task. Once presented with these maps I wonder how many civil officials actually consider the validity of the data they’ve requested and been presented with.
Going to wrap this one up for now, but the next one will return to the topic of volcanic hazard assessment – instead looking at the aviation risk – particularly how well we can model how an ash plume will move around the globe and interfere with airspace in different areas.
*Excellent paper presented by G. Lube et al at IUGG last week. No doubt to be found in a journal near you soon.