Volcano-tectonic earthquakes: A new tool for estimating intrusive volumes and forecasting eruptions

Abstract

We present data on 136 high-frequency earthquakes and swarms, termed volcano-tectonic (VT) seismicity, which preceded 111 eruptions at 83 volcanoes, plus data on VT swarms that preceded intrusions at 21 other volcanoes. We find that VT seismicity is usually the earliest reported seismic precursor for eruptions at volcanoes that have been dormant for decades or more, and precedes eruptions of all magma types from basaltic to rhyolitic and all explosivities from VEI 0 to ultraplinian VEI 6 at such previously long-dormant volcanoes. Because large eruptions occur most commonly during resumption of activity at long-dormant volcanoes, VT seismicity is an important precursor for the Earth's most dangerous eruptions. VT seismicity precedes all explosive eruptions of VEI ≥ 5 and most if not all VEI 4 eruptions in our data set. Surprisingly we find that the VT seismicity originates at distal locations on tectonic fault structures at distances of one or two to tens of kilometers laterally from the site of the eventual eruption, and rarely if ever starts beneath the eruption site itself. The distal VT swarms generally occur at depths almost equal to the horizontal distance of the swarm from the summit out to about 15 km distance, beyond which hypocenter depths level out. We summarize several important characteristics of this distal VT seismicity including: swarm-like nature, onset days to years prior to the beginning of magmatic eruptions, peaking of activity at the time of the initial eruption whether phreatic or magmatic, and large non-double couple component to focal mechanisms. Most importantly we show that the intruded magma volume can be simply estimated from the cumulative seismic moment of the VT seismicity from: Log10 V = 0.77 Log ΣMoment − 5.32, with volume, V, in cubic meters and seismic moment in Newton meters. Because the cumulative seismic moment can be approximated from the size of just the few largest events, and is quite insensitive to precise locations, the intruded magma volume can be quickly and easily estimated with few short-period seismic stations. Notable cases in which distal VT events preceded eruptions at long-dormant volcanoes include: Nevado del Ruiz (1984–1985), Pinatubo (1991), Unzen (1989–1995), Soufriere Hills (1995), Shishaldin (1989–1999), Tacana' (1985–1986), Pacaya (1980–1984), Rabaul (1994), and Cotopaxi (2001). Additional cases are recognized at frequently active volcanoes including Popocateptl (2001–2003) and Mauna Loa (1984). We present four case studies (Pinatubo, Soufriere Hills, Unzen, and Tacana') in which we demonstrate the above mentioned VT characteristics prior to eruptions. Using regional data recorded by NEIC, we recognized in near-real time that a huge distal VT swarm was occurring, deduced that a proportionately huge magmatic intrusion was taking place beneath the long dormant Sulu Range, New Britain Island, Papua New Guinea, that it was likely to lead to eruptive activity, and warned Rabaul Volcano Observatory days before a phreatic eruption occurred. This confirms the value of this technique for eruption forecasting. We also present a counter-example where we deduced that a VT swarm at Volcan Cosiguina, Nicaragua, indicated a small intrusion, insufficient to reach the surface and erupt. Finally, we discuss limitations of the method and propose a mechanism by which this distal VT seismicity is triggered by magmatic intrusion.