How to measure the size of a volcanic eruption

One of the most fascinating things about volcanic eruptions is the enormous variation in their scale and violence. On the one hand they can involve the tranquil effusion of sluggish red lava - spectacular but essentially harmless unless your house happens to be in the way. On the other they can take the form of explosions so staggeringly huge that ejected gas and dust blot out the sun for years, plunging the planet into bitter volcanic winter.

Devising a scale that can take account of such extremes, and everything in between, has always proved difficult, and until recently, volcanologists have been stuck with a sort of Richter Scale known rather cumbersomely as the Volcanic Explosivity Index or VEI.

On the scale, quiet, lava-producing eruptions score zero or one, points two, three and four are reserved for small to moderate explosive eruptions that occur somewhere on the planet every year. Explosive blasts meriting a five occur every decade or so, and include the 1980 eruption of Mount St Helens in Washington State.

A volcanic explosion great enough to score six on the index happens, on average, about every century - Pinatubo in the Philippines hosted the last in 1991, while a thousand years or more may separate the colossal eruptions deserving of a VEI7. The last of these enormous blasts tore apart the Tambora volcano in Indonesia in 1815 - the year of Waterloo. Top of the scale are the VEI8 super-eruptions, including two cataclysmic events at Yellowstone, Wyoming, in the last couple of million years and the second greatest explosion known, which gouged out the world's largest volcanic crater at Toba, Sumatra, around 73,000 years ago.

The only reason that all these blasts can be crammed into the index is because it is logarithmic. Every point on the scale relates to an eruption broadly ten times bigger than the one below. This makes the most explosive eruptions known a thousand times or more larger than that which ripped off the top 400 metres of Mount St Helens almost a quarter of a century ago.

While performing well in terms of squeezing a quart into a pint pot, volcanologists have never been entirely happy with the VEI, mainly because eruptions are scored primarily on the volume of debris they eject. The problem is that this takes no account of the density of the erupted material. The same amount of magma ejected at two different volcanos might produce an enormous volume of fluffy ash at one and a much smaller volume of dense volcanic rock at another. As a consequence the first might score eight on the index while the second has to make do with a seven. This is hardly fair and, more seriously; it means that the scale is not up to the job of objectively measuring and categorising the true size of volcanic eruptions.

Help is at hand. In a recent paper, published in Bulletin of Volcanology, Ben Mason, David Pyle and Clive Oppenheimer of Cambridge University apply a scaling system that groups eruptions according to magnitude. This new scheme depends upon converting the volumes of all eruptions to erupted mass (in kilograms) - at a stroke solving the ash versus rock problem. As well as providing a more accurate means of comparing eruptions, the new scale ensures that we can now be more certain of their true sizes, something particularly important when we are dealing with those top-end super-eruptions that have the potential to drastically impact upon the Earth's climate. Using the magnitude scale, Mason and his colleagues identify 47 cataclysmic eruptions over the last half a billion years, including 42 during the last 36m years alone. All are characterised by the ejection of enormous amounts of debris weighing at least 150 times more than that erupted during the earlier mentioned 1991 Pinatubo blast. Previous estimates of the frequency of VEI8 super-eruptions hovered around the 50,000 year mark. The newly designated magnitude eight eruptions appear to occur much less frequently - around every million years - with the last devastating the Taupo region of New Zealand's North Island about 26,000 years ago. It seems likely, however, that numerous magnitude eight eruptions remain to be unearthed in the recent geological record and it may well be that such an event may indeed occur every 50 millennia or so.

Enough of the past; given their potential for ruining our summers - to say the least - what about the future? Mason and friends looked in their crystal ball and have come up with some predictions. There is 1% chance of a magnitude eight event erupting onto the scene in the next 460-7,500 years, depending on how the figures are juggled. These odds may well seem far too small to lose sleep over, but they mean that everyone alive today is far more likely to experience a volcanic super-eruption than they are to win the lottery. How lucky are you feeling?

· Bill McGuire is director of the Benfield Hazard Research Centre at University College London. To buy A Guide to the End of the World: Everything you never wanted to know (OUP) for £8.99 plus free p&p; call Guardian Book Service on 0870 836 0875