UCL ENTERPRISE

Insurance companies take all manner of risk factors into account when deciding what premiums to set; similarly, reinsurance companies – the insurers’ insurers – need to know how likely it is that an untoward event will happen. Chris Kilburn from the Aon Benfield UCL Hazard Centre (ABUHC) in UCL’s Department of Earth Sciences is trying to predict when and where volcanic eruptions might occur, and the potential damage they could cause to the local area.

Predicting the Weather

Natural disasters can result in huge insurance liabilities, but it’s difficult to predict precisely when and where they will happen. The reinsurance company Aon Benfield is continuing a partnership that began in 1997 with scientists across UCL. A major success has been forecasting hurricanes off the east coast of the US and, more recently, European wind storms, thanks to innovations from UCL’s Mullard Space Science Laboratory. But now they are turning to events that are even more difficult to predict – geological hazards, such as volcanic eruptions and landslides, alongside Chris Kilburn and colleagues in the ABUHC.

The Eyjafjallajökull eruption in 2010 highlighted the potential value of volcano insurance cover – although it was only a small eruption, the impact on business of aeroplanes being grounded was huge, from missed meetings to produce destined for airfreight rotting in warehouses. But before they can set insurance premiums, the insurers and reinsurers need to get a better idea of the likelihood of an event happening, and the damage it might cause.

Risk assessment companies already look at earthquakes, for example, but there is a real gap in the market regarding volcanic eruptions and landslides, which are notoriously difficult to predict with any great certainty. Kilburn’s research includes taking field measurements at erupting volcanoes, and creating models that enable predictions of how far and in which direction lava is likely to flow. This helps the authorities decide whether and when to evacuate towns and villages.

Volcanic Activity

A further impact of the research is to give people living and working near a volcano a little more certainty by trying to predict when an eruption will occur. ‘As an eruption approaches, the number of earthquakes a day increases, but up to now deciding whether the earthquakes are getting so fast that an eruption is imminent has been fairly arbitrary,’ he says. ‘We are applying statistical techniques and combining that with the underlying physics to understand better what is going on.’

An important application is forecasting eruptions at volcanoes that have been quiet for generations, and thus are not routinely monitored. When an emergency begins, there’s a massive scramble to get data, and Kilburn believes they are becoming able to recognise points when the changes in precursory events, like local earthquakes, become so dramatic that an eruption is imminent. While predicting precise timings is not yet possible, as they depend on the complexities of how rocks deform and break, this research helps the authorities develop realistic strategies for responding to emergencies, even though there is a high chance of a false alarm.

A Market for 'minor perils'

‘Previously, eruptions and landslides were considered “minor perils”, and insurance companies weren’t really interested in them,’ he says. The market for hurricane insurance is quite saturated, so minor perils represent a potential new area of business. ‘Companies are particularly interested in developing insurance products in south-east Asia, which is one of the world’s most volcanically active areas, and they really need to take the possibility of eruptions into account.’