Earth’s Upper Crust Seismically Excited by Infrasound from the 2022 Hunga Tonga–Hunga Ha’apai Eruption, Tonga

Abstract

AbstractRecords of pressure variations on seismographs were historically considered unwanted noise; however, increased deployments of collocated seismic and acoustic instrumentation have driven recent efforts to use this effect induced by both wind and anthropogenic explosions to invert for near-surface Earth structure. These studies have been limited to shallow structure because the pressure signals have relatively short wavelengths (<∼300 m). However, the 2022 eruption of Hunga Tonga–Hunga Ha’apai (also called “Hunga”) volcano in Tonga generated rare, globally observed, high-amplitude infrasound signals with acoustic wavelengths of tens of kilometers. In this study, we examine the acoustic-to-seismic coupling generated by the Hunga eruption across 82 Global Seismographic Network (GSN) stations and show that ground motion amplitudes are related to upper (0 to ∼5 km) crust material properties. We find high (>0.8) correlations between pressure and vertical component ground motion at 83% of the stations, but only 30% of stations show this on the radial component, likely due to complex tilt effects. We use average elastic properties in the upper 5.2 km from the CRUST1.0 model to estimate vertical seismic/acoustic coupling coefficients (SV/A) across the GSN network and compare these to recorded observations. We exclude many island stations from these comparisons because the 1° resolution of the CRUST1.0 model places a water layer below these stations. Our simple modeling can predict observed SV/A within a factor of 2 for 94% of the 51 non-island GSN stations with high correlations between pressure and ground motion. These results indicate that analysis of acoustic-to-seismic coupling from the eruption could be used to place additional constraints on crustal structure models at stations with collocated seismic and pressure sensors. Ultimately, this could improve tomographic imaging models, which rely on methods that are sensitive to local structure.