Hazard Assessment Part 4: A case study in misreporting

Vedde ash particles under an electron microscope. Source: Lowe & Turney (97), J. Geol. Soc v.154 p. 609

If this article is to be believed, “Another giant UK ash cloud ‘unlikely’ in our lifetimes”.  The original paper was recently published in Geology, and consists of a systematic study of peat cores to look at how many ash fall events have been recorded across Europe in the last 7000 years. The article is subscription only, but the BBC news piece does a good job of basically copying the abstract verbatim.  My belief is that science reporters exist largely to communicate the often dry but important academic findings of researchers to the public.  If a stuffy professor in a lab* somewhere has discovered when and where a tsunami is going to occur, I expect a science reporter to be able to parse that information and communicate it in a clear and accessible manner. Is this unreasonable?

What the BBC article does not do is look at any of the caveats presented in the published article.  For example, the final sentence in the conclusions “These figures should be considered as minimum estimates” at no time make an appearance in the BBC piece.  It strikes me as bizarre that the news article barely touches on the paper authors own admissions that the records are almost certainly incomplete.  Instead it gets caught up in how resilient humans have been to these events – something which the paper doesn’t actually deal with in any case.

One of the figures published in the journal article, and reproduced in the news piece is that there is a 16% probability of a large ash event every 10 years on average. Now, I find this very difficult to rectify with the title of the bbc article (the quote with which I started this piece), unless, contrary to my understanding, everyone in the country is over the age of 60.

The fact is that the median age in the UK is about 40. With an average life expectancy of about 80 (76 for us men), then we have about 40 years average life left.  So:

Probability of eruption in 10 years (p)= 0.16.  Probability of it not happening (q) = 1-p = 1-0.16 = 0.84. Probability of it happening at least once in 40 years = 1-q^4 = 1-0.84^4 = 0.502.

So almost exactly half.  Except the authors have already stated several times that their sampling method is almost certain to have underestimated the number of events. A friend of mine works in a team carrying out a somewhat larger study across the Mediterranean and Europe doing volcanic tephra correlations, and they keep turning up examples of people mis-correllating layers.  There is a huge temptation for people to look at an ash layer, not bother doing any geochemical analysis (which would give you some data for working out what the source was), and just match ashes by what the glass shards look like under a microscope. There’s quite an interesting review piece here.  The Vedde Ash in Europe is an excellent example – an Icelandic ash layer which has been found across Europe and into Russia, and associated with some really bad correlations.  This article – albeit in rather dry academic tones – underlines the problems.  Basically it’s got to the point where if anyone finds an ash layer anywhere near 10,000 years ago it gets thrown into the ‘Vedde’ pile, despite the fact it has been realised that many different prior and subsequent eruptions had similar particle make-up.

Anyway, back to my original point – the BBC article demonstrates either a willful denial of the facts in order to write the article they wanted to, or a complete misunderstanding of probability and scientific dataset collection.  Either of which is reprehensible for a science journalist when trying to communicate risk.

*When I say professor in a lab, clearly I mean professor in an office somewhere from which he can command his postgrads and postdocs, who practically live in the lab.  See here.