Bayesian event tree for long‐term volcanic hazard assessment: Application to Teide‐Pico Viejo stratovolcanoes, Tenerife, Canary Islands

Abstract

In modern volcanology one of the most important goals is to perform hazard and risk assessment of volcanoes near urbanized areas. Previous work has been done to assess volcanic hazard in the form of event tree structures containing possible eruptive scenarios. Probability methods have been applied to these structures to estimate the long term probability for each scenario. However, most of these event tree models show restrictions in the eruptive scenarios they consider and/or on the possibility of having volcanic unrest triggered by other forces than magmatic. In this paper, we present a Bayesian event tree structure which accounts for external triggers (geothermal, seismic) as a source of volcanic unrest and looks at the hazard from different types of magma composition and different vent locations (as opposite to a central vent only). We apply the model to the particular case of Teide‐Pico Viejo stratovolcanoes, two alkaline composite volcanoes that have erupted 1.8–3 km3 of mafic and felsic magmas from different vent sites during the last 35 ka, situated on a densely populated island, one of the biggest tourist destinations of Europe, and for which limited geological and no historical data exist. Hence, the importance of volcanic hazard assessment for risk‐based decision‐making in land use planning and emergency management. A previous attempt to estimate the volcanic hazard for Teide‐Pico Viejo has been done using an event tree structure based on Elicitation of Expert Judgment. The new method overcomes some limitations of the previous method, including human decision bias, epistemic and aleatoric uncertainties, restrictions on the segmentation complexity of the event tree structure, and automatically updating. The main steps are the following: (1) Design an extensive tree‐shaped Bayesian network with possible eruptive scenarios following the case of Teide‐Pico Viejo volcanic complex. (2) Build a Bayesian model to estimate the long term volcanic hazard for each scenario. (3) Apply the model to Teide‐Pico Viejo stratovolcanoes. Finally, we compare the results with those from the Elicitation method applied before, as well as previous Bayesian event tree structures developed for other volcanoes.