Investigation of volcanic processes using seismology and geodesy at Okmok Volcano, Alaska

Okmok Volcano, Alaska is a frequently active system with eruptions in 1997 and 2008 that differed in style and vent location. We conduct various seismic and geodetic studies of Okmok, focusing on better characterizing the volcano's subsurface structure and changes leading up to the 2008 eruption. In the first study, we perform ambient noise interferometry using cross-correlation of noise between station pairs to investigate changes in Okmok's seismic properties preceding and following the 2008 eruption. In the second, we test the influence of phase-weighted versus linear stacking on the quality of ambient noise tomography (ANT). In the third, we perform a joint inversion of body-wave arrivals and surface wave dispersion to solve for three-dimensional P-wave and S-wave velocity structure and hypocenter locations. Finally, we conduct time series analysis with temporal adjustment of Okmok's deformation between 1997 and 2008 using wrapped phase observations from interferometric synthetic aperture radar (InSAR).;We find two prominent signals in relative seismic velocity in the intereruptive period, strongest on station pairs with paths beneath the caldera. These are a seasonal variation, believed to be due to precipitation and snow loading, overprinted by a gradual increase in velocity until the 2008 eruption. The increase, contrary to typical observations preceding eruptions, may be due to viscoelastic effects decreasing the stresses above the pressurized magma chamber during the late intereruptive period. We find that phase-weighted stacking improves the signal-to-noise ratio of Green's functions and the quality of dispersion curves, group velocity maps, and the resulting S velocity model with respect to linearly stacking. The ANT-derived S model shows two major low velocity zones (LVZs) at depths that agree with previous studies, but their lateral extent is unrealistically large. Joint inversion of body-wave and surface-wave data produces an optimal P model similar to the body-wave-only model, but the S model improves noticeably and suggests slightly greater depth extent of the lower LVZ. From temporal adjustment on InSAR-estimated variations in source strength, we find an adequate fit to a parameterization consisting of twoexponential decay steps, suggesting that viscoelastic processes play a role in deformation during intereruptive periods.