Detecting explosive volcanism using global long-range infrasound data

Volcanoes have the potential to erupt abruptly, ejecting ash into the atmosphere and posing risks, among others, to civil aviation. Such explosive eruptions often occur unnoticed, complicating the air traffic management efforts of Volcanic Ash Advisory Centres (VAACs) of the International Civil Aviation Organization. Explosive eruptions are efficient sources of infrasound that, thanks to the low attenuation in the atmosphere, allow volcano monitoring at long (up to >1000 km) distances. The Volcanic Information System (VIS) is a recent framework for detecting eruptive events in near real-time (considering the slow propagation of infrasound), based on infrasound observations from one or several arrays of infrasound sensors. Several studies have already proved the efficiency of long-range infrasound monitoring and VIS in particular, but a systematic and global validation has not yet been done. In the current study, we investigate the reliability of VIS based on 10 years (2010-2019) of data provided by 16 infrasound arrays of the International Monitoring System operated by the Comprehensive Nuclear-Test-Ban Treaty Organization. The dataset contains eruptions with a Volcanic Explosivity Index (VEI) of 2 (mild explosions) to 4 (very energetic explosions with a high ash column). In order to estimate the rate of false positives, VIS records are compared with reports from the Global Volcanism Program. Our results show that VIS is well designed for large (VEI > 3) and long-lasting (Sub-Plinian/Plinian) events, as well as for highly repeated Vulcanian/Strombolian explosions. On the other hand, it typically misses single explosive, transient (few seconds) eruptive events. Good results on true detected events are obtained within 1000 km and up to 2000 km range under downwind propagation conditions. Unresolved ambiguity often remains for cases with small angular separation between volcanoes with respect to the infrasound array. We tackle the issue of azimuth resolution by considering volcanic sectors rather than single edifices. This approach can still provide critical information to the VAACs for triggering independent and extended analysis on ongoing volcanic eruptions.