Probably the last of these I’ll do for a little while, but perhaps the one with the broadest current public interest.
One of the interesting things about the shutting of airspace due to volcanic ash is that – for most people – it was a completely alien concept prior to the Eyjafjall eruption last spring. Since then there have been several other high profile airspace closures due to ash – notably the Grimsvötn eruption earlier this year, and the Puyehue-Cordón eruption which is currently ongoing. The first shut down European airspace, while the second cycled itself around the southern hemisphere making things exciting for South America, Australia, New Zealand, Chile and Argentina.
In actual fact airspace closures due to volcanic activity are nothing new – following events such as the 1982 Speedbird 9 incident over Indonesia (where all four engines on a British Airways 747 failed while flying through an ashcloud), and the 1989 Mt Redoubt KLM incident (another 747 with all 4 engines shut down) the Volcanic Ash Advisory Centres (VAAC) were set up and it became standard practise to shut airspace around active eruptions. The reason we were so suddenly made aware of it only last year is that the Eyjafjall eruption was the first to really smother routes which could not simply be bypassed. Some of the busiest airspace in the world was shut down. To be honest, the only reason it came as a surprise to us this late in the game is that Iceland’s volcanoes have been unusually quiet over the last 30 years.
One of the big debates regarding the hazard assessment of these events is at what level ash concentrations become dangerous to aircraft. That’s all very well as an academic debate, and certainly there is evidence that the various aviation authorities have the current thresholds very much on the safe side of the line. However, what interests me more is how the maps of ash concentration are derived.
A number of numerical models are used to model ash dispersal, but what it comes down to is that an artificial column of ash is injected into a simulated atmosphere, and then weather transport modelling is used to forecast where the ash goes. This information is then fed to civil agencies and governments to use in declaring areas safe or unsafe. A lovely example of the kind of graphics is shown below, showing ash dispersal after 10 days of the Puyehue-Cordón eruption in Chile, courtesy of the Canadian Met Office.
That’s lovely, and businesses and governments love these things, and they look pretty on news sites. Reading the rabid rantings of Daily Mail readers, deploring the nanny-state shutdown of airspace when there’s so little ash in such-and-such an area you might be forgiven for thinking we’ve nailed the problem and just need to revise our sensitivity to flying through this stuff. However, there are fundamental weaknesses with these models which simply are not communicated to the public.
Firstly, and most importantly – all the dispersion data is processed using weather modelling algorithms. When was the last time you trusted a weather forecast to get things correct?
Secondly, many of these models don’t even include realistic ash loads – they simply inject specific grain sizes into the atmosphere at pre-set altitudes and hope that what they see is realistic dispersal. It may be, it may not be. We have little data to refute or support.
Thirdly, in order to correctly measure and forecast ash dispersal, we have to know precisely what the source is doing. In other words – what is the eruption plume doing? For example:
- How tall is it
- How wide is it
- How is height varying with time
- How is width varying with time
- What velocity are particles being ejected at
- What is the temperature of the material coming out
- What is the size distribution of the material coming out
The vast majority of eruption plumes are first identified by satellite observation, which gives us no information on altitude, and certainly nothing on what the actual plume width is (this is hidden by the spreading mushroom-cloud-like blanket at the top of the plume). Data on variation in plume height and width are only really possible by on-site observation (e.g. webcams, assuming the field of view is appropriate to actually observe these things). The rest of the variables are basically out of the question even at the most highly monitored volcanoes. This will give you an idea of the kind of update you might get from an extraordinarily well monitored volcano.
The simulations we have at the moment are first-order-accurate at best, and pretty damn hand-wavy about the input parameters. The message to take away is that while we can model what happens when ash plumes are injected into the atmosphere, do not think for a moment we are accurately modelling what an actual ash plume is doing as it is injected into the atmosphere. The very fact that test flights went up and were not measuring the same concentrations as had been predicted tells you something about how poorly understood the system is, and until someone can tell you where that extra ash actually is, I would be extremely cautious about believing any of those nice pretty dispersal maps.