Ballistic fallout from major explosions and paroxysms at Stromboli (Italy): from the mapping of historically affected areas to probabilistic hazard assessment

Based on mathematical analyses of the extensive historical and recent records of its explosive activity, this study presents the first probability hazard maps of the areas potentially affected by ballistic fallout from major explosions and paroxysms at Stromboli (Italy). These are more energetic explosions that punctuate the persistent explosive activity of the volcano, where the paroxysms are the greatest category, able to reach inhabited areas.

Conditional probability maps have been produced by adopting a novel approach that develops and integrates three statistical models of ballistic fallout patterns and the associated uncertainties. Model 1 mirrors the areas observed to be affected in the past, whereas Models 2 and 3 address data under-sampling and morphologic variations of craters and/or shallow part of conduits, respectively assuming independency between the categorical data of ballistic distance and dispersal direction, or radial symmetry. This approach could be applied to similar hazard mapping problems, where it is possible to represent the data in terms of distances and directions with respect to a relatively constrained geographical center.

Notably, this study is also based on a new method to map the areas affected by ballistic fallout of a sufficient number of major explosions and paroxysms from historical and recent records. This mapping method adopts a simplified description of the affected areas by a circular proximal area and up to three circular sectors with variable radius and width, and associated uncertainties. The dataset of maps includes a total of 67 events over ≈150 years, based on an extensive review of historical, observational, and monitoring data. The new mapping approach is less detailed than free-hand isopach drawing, and includes some elements of expert judgement to manage non-homogeneous and decades-old information. Nevertheless, all its steps are transparent and replicable, from the original excerpts of contemporaneous sources to the geographical mapping of the areas affected by ballistic fallout.

In addition to conditional maps of major explosions and paroxysms, the study presents temporal probability assessments of these phenomena. A comprehensive probability distribution of maximum ballistic distances at Stromboli should merge the major explosions and the paroxysms, but their assessments are based on two different catalogs, and therefore the estimates of their occurrence ratio are significantly uncertain. In fact, major explosions before 1970 are affected by likely under-recording issues. Vice versa, only 5 paroxysms occurred after 1970, and their time series is irregular and characterized by temporal clusters and a 44-year gap between 1959 and 2003. Moreover, the probability rates of major explosions and paroxysms are not constant in time, but significantly increase in the weeks/months after one of these events has occurred. Combining the maps with the variable occurrence rates of the events, cumulative estimates of ballistic fallout probability in the next years are presented, as well as hourly probabilities as a function of the time passed after the previous major explosion or paroxysm.

These findings open the way to individual and societal risk assessments for this phenomenon at Stromboli, and represent useful approaches for studying ballistic hazard, especially on island volcanoes.