Explosive eruptions at Stromboli volcano (Italy): a comprehensive geochemical view on magma sources and intensity range

Abstract

A comprehensive understanding of the processes that occur during magmatic storage and pre-eruptive ascent—and of their associated timescales—is critical to identifying potential precursory signals, and to developing robust volcano early-warning systems. Stromboli’s persistent activity comprises continuous degassing and explosive activity that ranges from hourly, low-intensity “normal” activity to occasional, more violent, paroxysmal activity. While the magma source processes that drive normal and paroxysmal activity are reasonably constrained, eruptive activity intermediate in magnitude and intensity (i.e., major explosions) remains elusive in terms of classification, source region, and pre-eruptive timescales. Here, we investigate the 19 July 2020 major explosion that geophysical parameters place at the upper limit of the major explosions field, close to small-scale paroxysms such as the 2003 and 2007 events. The geochemical signatures of matrix glass, olivine, melt inclusions, and embayments—integrated with gas measurements—highlight important differences in eruption source, ascent behaviour, and pre-eruptive timescales of the studied event when compared to paroxysms. Melt inclusion volatile contents identify that magma rise begins from a slightly shallower source (~9.5 km below sea level, b.s.l.) than for paroxysms (11.4 km b.s.l.), with the activation of a shallower ponding zone at 5–6 km b.s.l.. This, in combination with intermediate matrix glass compositions, suggests complex ascent behaviour, characterised by CO2 buffering in the deep ponding region and magma self-mixing in the shallower zone. Fe–Mg-diffusion modelling in olivine indicates a system perturbation starting ~20–25 days before eruption onset, in agreement with the timescale of volcanic gas CO2/SO2 ratio changes observed in the plume, and significantly shorter than that observed prior to paroxysms (~4 months). The geochemical dataset provides insights into the processes controlling the steady-state conditions and the broad spectra in eruption magnitude and intensity at Stromboli and bears important implications for eruption forecasting.

Magma Differentiation in Dynamic Mush Domains From the Perspective of Multivariate Statistics: Open‐ Versus Closed‐System Evolution

“…Notably, historical data and temporal models suggest that the average frequency for paroxysmal eruptions was of one event every 4 years, during the last 140 years (Bevilacqua et al, 2020). The ubiquitous eruption of mafic lp magma during paroxysmal events indicates that 4 years actually represent the lower limit for the frequency of mafic recharge episodes to the hp reservoir since lp magma is also involved in some major eruptions and some recharge episodes may not lead to violent explosive scenarios (e.g., Landi et al, 2022;Petrone et al, 2022;Voloschina et al, 2023). Therefore, comparison of independent constraints on average residence times and recharge frequency indicates Geochemistry, Geophysics, Geosystems 10.1029/2023GC011396 that timescales of lp recharges are on the same order of magnitude or slightly shorter than those of magma rejuvenation in the hp reservoir.…”

CO2 Flushing Triggers Paroxysmal Eruptions at Open Conduit Basaltic Volcanoes

“…We first introduce the model that was inspired by the measurements collected at Stromboli volcano and present its results. We finally apply the model to the major explosion of the 19th of July 2020 because it is one of the strongest of the last 20 years of instrumental record (Andronico et al., 2021), we have data providing a continuous record of CO 2 , H 2 O, and SO 2 emissions and the petrological characterization of the eruption has been completed (Voloschina et al., 2023).…”

Statistical Insights on the Eruptive Activity at Stromboli Volcano (Italy) Recorded from 1879 to 2023

“…These are very powerful explosive events involving more than one vent and forming eruptive columns more than 1 km above the craters and occurring at a frequency of one every few decades [23,[39][40][41][42][43]. The last five episodes occurred on 5 April 2003, 15 March 2007, 3 July and 28 August 2019, and 19 July 2020 [29,30,42,[44][45][46][47][48][49][50][51]. This latest event of 19 July 2020 was on the boundary between ME and P because it was characterized by several discrete pulses involving more than one crater zone, and on the basis of the seismic trace, VLP size, area involved by the fallout, height of the eruptive column, thermal signature and magma source depth [32,34,51,52].…”

“…The last five episodes occurred on 5 April 2003, 15 March 2007, 3 July and 28 August 2019, and 19 July 2020 [29,30,42,[44][45][46][47][48][49][50][51]. This latest event of 19 July 2020 was on the boundary between ME and P because it was characterized by several discrete pulses involving more than one crater zone, and on the basis of the seismic trace, VLP size, area involved by the fallout, height of the eruptive column, thermal signature and magma source depth [32,34,51,52]. The collapse of the eruptive column accompanying the paroxysms may cause pyroclastic density currents (PDC) spreading along the SdF barren slope (Figure 1b) and possibly on the sea surface, up to 2 km away from the coastline [32,33,53].…”