High-precision earthquake location, velocity determination, and event family identification at Augustine Volcano, Alaska, from 1993 through the 2005-2006 eruption

Volcano seismic networks typically have few stations and marginal coverage, providing challenges for earthquake location in a complex, three-dimensional setting. To improve location precision at Augustine Volcano, Alaska, we compute a three-dimensional P-wave velocity model using double-difference (DD) tomography combined with waveform cross-correlation (WCC) techniques. We also examine temporal changes in earthquake locations and waveform characteristics associated with the 2005-2006 eruption and pre-eruptive seismicity. The Alaska Volcano Observatory (AVO) has monitored Augustine using up to 9 stations since 1993, and the AVO hypocenter and waveform catalog from 1993-2006 serves as the initial dataset. Many of the catalog hypocenters locate above the summit, reflecting the limitations of applying standard location techniques in rugged and sparsely instrumented volcanic settings. WCC using bispectrum verification improves the pick accuracy of the catalog data and is used to identify similar earthquakes. Waveform similarity at Augustine is low compared to other Alaskan volcanoes such as Redoubt, and most event families contain less than 100 events. Earthquakes recorded during a period of increasing pre-eruptive seismicity in December 2005 form clusters of similar earthquakes over periods of days. Events prior to the 2005-2006 eruption can exhibit a high degree of similarity over multiple years. The DD tomography method provides significantly improved absolute and relative earthquake locations and source region velocity information. We use differential travel times from catalog and cross-correlation data to simultaneously invert for hypocenter location and P-wave velocity structure. Previous studies have shown a high degree of north-south trending variation in compressional wave velocity at Augustine. This is reflected in severe station correction-velocity-depth tradeoffs when performing standard 1D inversions to solve for a starting model. Using our combined DD tomography and WCC approach, we better constrain the 3D nature of velocity heterogeneity beneath the volcano.