Santorini unrest 2011–2012: an immediate Bayesian belief network analysis of eruption scenario probabilities for urgent decision support under uncertainty

Abstract

Unrest at the Greek volcanic island of Santorini in 2011-2012 was a cause for unease for some governments, concerned about risks to their nationals on this popular holiday island if an eruption took place. In support of urgent response planning undertaken by the UK government, we developed a rapid evaluation of different eruption scenario probabilities, using the Bayesian Belief Network (BBN) formulation for combining multiple strands of scientific and observational evidence. Here we present three alternative BBN models that were devised in early 2012 for assessing the situation: (1) a basic static net for evaluating probabilities at any one moment in time, utilising just four key unrest indicators; (2) a compound time-stepping net, extending the basic net to update probabilities through time as the indicators changed; and (3) a more comprehensive net, with multiple lines of other data and observations incorporated, reflecting diversity of modern multi-parameter monitoring techniques. A key conclusion is that, even with just three or four basic indicators, it is not feasible, or defensible, to attempt to judge mentally the implications of signs of unrest -a structured probabilistic procedure using Bayes' Rule is a rational approach for enumerating evidential strengths reliably. In the Santorini case, the unrest, and official anxiety, diminished quite quickly and our approach was not progressed to the point where detailed consideration was given to BBN parameters, analysis of data uncertainty or the elicitation of expert judgements for quantifying uncertainties to be used in the BBN. Had this been done, the resulting scenario probabilities could have been adopted to determine likelihoods of volcanic hazards and risks caused by possible eruptive activity, as identified in a concurrent assessment of the scale and intensities of potential volcanic impacts (Jenkins et. al., Assessment of ash and gas hazard for future eruptions at Santorini Volcano, Greece. forthcoming). Ideally, such hazard and risk assessments should be elaborated in detail and critiqued well before crisis-level unrest develops -not initiated and implemented within a few hours just when a situation looks ominous. In particular, careful analysis of all information is required to determine and represent parameter uncertainties comprehensively and dependably.

Rational volcanic hazard forecasts and the use of volcanic alert levels

“…Bayesian approaches are often used in conjunction with Event Trees (e.g., Newhall and Hoblitt 2002;Marzocchi et al 2004Marzocchi et al , 2008Newhall and Pallister 2015, and references therein), that represent the complex ramification of possible outcomes, each one quantified as a probability distribution which is allowed to evolve as long as new information is added (e.g., when new observations are available). To-date, Bayesian approaches have been employed in a large number of situations in volcanology, including forecasts of volcanic hazards over the shortterm (Aspinall et al 2003(Aspinall et al , 2006Marzocchi et al 2008; Lindsay et al 2010;Brancato et al 2011Brancato et al , 2012Bell and Kilburn 2012;Marzocchi and Bebbington 2012;Sandri et al 2009Sandri et al , 2012Selva et al 2012Selva et al , 2014Garcia-Aristizabal et al 2013;Anderson and Segall 2013;Rouwet et al 2014;Aspinall and Woo 2014;Sobradelo et al 2015;Boue et al 2015;Tonini et al 2016;Bartolini et al 2016) as well as over the long-term (Martin et al 2004;Baxter et al 2008;Neri et al 2008;Orsi et al 2009;Marzocchi et al 2008Marzocchi et al , 2010Sobradelo and Martì 2010;Passarelli et al 2010a, b;Selva et al 2012;Marzocchi and Bebbington 2012;Sandri et al 2012Sandri et al , 201...…”

“…As was the case with other fields of investigation, e.g., fluid geochemistry, satellite data analysis, and computational thermo-fluid dynamics, that have been increasingly contributing to volcanic hazard forecasts and scenario evaluation, there is no doubt that probabilistic analyses will increase in relevance for and be more present in volcano observatories, as well as at nearby cooperating Universities and research centers. To-date, computational tools and manuals can be freely downloaded from a number of sites, e.g., https://vhub.org/resources/betunrest for Bayesian Event Tree analysis for both magmatic and nonmagmatic unrest phases (Marzocchi et al, 2008;Rouwet et al, 2014;Tonini et al, 2016), or https://www.norsys.-com/download.html for Bayesian Belief Network analysis, the latter having been applied to urgent decision support during the unrest at Santorini in (Aspinall and Woo, 2014.…”

Volcano Crisis Communication: Challenges and Solutions in the 21st Century

“…Current practitioners of Bayesian Event Tree (BET) analysis use either the Cooke-Aspinall method (Cooke 1991;Aspinall 2006) or the INGV (National Institute of Geophysics and Volcanology) method (Marzocchi et al 2004(Marzocchi et al , 2008, although there are other implementations (e.g., Sobradelo et al 2014;Jolly et al 2014;Newhall and Pallister 2015). In addition, Bayesian Belief Networks (BBN), another graphic method that does not require the same type of linear time progression as in BET systems, may be used effectively in some situations (e.g., Lindsay et al 2010;Hincks et al 2014;Aspinall and Woo 2014). All of these methods integrate some form of elicitation of opinions from a team of experts to assign probabilities and uncertainties based on monitoring data, past eruptive behaviour and conceptual models.…”

Volcanic Unrest and Pre-eruptive Processes: A Hazard and Risk Perspective

“…Whilst ground deformation, seismicity, thermal flux or anomalous degassing are indicators of possible future activity these phenomena also pose significant immediate threats to population, infrastructure and other assets in affected areas during the unrest. From a scientific point of view, hazard assessment relating to eruptive activity has made considerable progress in recent years partly through the deployment of increasingly powerful computational models and simulation capabilities (e.g., Esposti Ongaro et al 2007;Manville et al 2013) as well as through advances in the development of probabilistic eruption forecasting tools (e.g., Marzocchi et al 2008;Aspinall 2006;Aspinall and Woo 2014) and improvements to fundamental understandings of the root drivers of changing activity (e.g., Cashman and Sparks 2013).…”