Application of the probabilistic model BET_UNREST during a volcanic unrest simulation exercise in Dominica, Lesser Antilles

摘要: We report on the first ‘real-time' application of BET_UNREST (Bayesian Event Tree for Volcanic Unrest) probabilistic model, during a VUELCO Simulation Exercise carried out island Dominica, Lesser Antilles, in May 2015. Dominica has concentration nine potentially active volcanic centers and frequent earthquake swarms at shallow depths, intense geothermal activity recent phreatic explosions (1997) indicate region is still active. The exercise scenario was developed secret by team scientists from University West Indies (Trinidad Tobago) Auckland (New Zealand). simulated unrest provided to exercise's Scientific Team three ‘phases' through injects comprising processed monitoring data. applied newly created model its software implementation PyBetUnrest, estimate probabilities having i) ii) magmatic, hydrothermal or tectonic origin, which may not lead iii) an eruption. obtained each phase raised controversy deliberations among members scientific team. results were often considered be ‘too high', included any reports presented ODM (Office Disaster Management) revealing interesting crisis communication challenges. concluded that PyBetUnrest itself successful brought tool one step closer full implementation. However, as with proposed method it needs more testing, order able use future we make series recommendations applications. This article protected copyright. All rights reserved.