A UK aid review recently launched by ex-Liberal Democrat leader Lord Ashdown, says more lives could be saved by faster UK response times to natural disasters. Estimating an increase in the number of people hit by disasters over the next four years from 250m a year to 350m a year, Ashdown is keen to improve local resilience in at-risk countries, and help make the leadership of international responses better and more integrated. The review said doing more to help nations become more resilient to disasters should be a core part of the government’s work.
The University of Bristol’s Environmental Risk Research Centre (BRISK) is already working closely with government scientists in Guatemala to achieve just this goal. The scientists at The Instituto Nacional de Sismologia, Vulcanologia, Meteorologia e Hidrologia (INSIVUMEH) have been able to monitor activity at one of the most dangerous volcanoes in the region, the ‘Volcan de Fuego’, however, despite their gallant efforts, due to lack of funding, no formalised risk assessments were being carried out. Thousands of people living and working in the area surrounding the volcano are at risk of eruptions, but recent work there by BRISK, using Palisade’s decision making under uncertainty software DecisionTools Suite, has proved invaluable in mitigating the associated risks.
Conventional risk assessments have attempted to model the probability of a hazard and combined that with the vulnerability of the population to create societal risk curves or values of Individual Risk per Annum (IRPA). However, using The DecisionTools Suite’s Monte Carlo technique sampling, the BRISK group have gone a step further and developed a unique way of modeling the likelihood of a successful evacuation.
The new approach has modeled the likelihood that the people in the region will need to be evacuated if the volcano’s activity increases. It uses risk software @RISK and decision tree software PrecisionTree to input several variables, including the time between an eruption starting and a hazard hitting a location, along with the warning time that may come from the authorities and the time the population needs for evacuation. All of these are difficult to predict with any accuracy and need to be represented by appropriate uncertainty distributions by the experts. By a special elicitation process, these expert views are weighted and then pooled together. The variables and their uncertainties are combined together in a logic tree within PrecisionTree, with the end result being the probability of successful evacuation, or not. When fed back in the @RISK model, the beneficial effects of evacuation on risk, and the costs of failure, are made very clear.
This recent work clearly goes to show the value of a global community, and how pooling resources can help those in need. The new approach to modeling as has also been so compelling to the news world, that several news outlets have been writing about its significance, including Scientific Computing World who recently ran this piece on BRISK’s work.