Ambient Noise Observables from USArray: Site Amplification, Scattering and Attenuation

Abstract:

Previous work has demonstrated the usefulness of tracking wavefronts from ambient noise cross correlations across small, dense arrays (e.g., the Long Beach array). We now apply a similar method to USArray data to explore the information available on a continental scale. The approach measures the travel times and amplitudes of a wave propagating across the array and through numerical computations of spatial gradients corrects for geometric spreading, focusing and defocusing; any remaining signal is attributed to local site amplification effects, intrinsic attenuation, or the presence of sources/scatterers within the array. Because the USArray was active in different regions at different times and therefore subject to a variable background noise field, we are careful to divide the US into smaller regions and ensure that noise correlation pairs can be stacked over an identical time period before measuring a given spatial gradient. USArray observations of distant earthquakes have been shown to recover site amplification terms and attenuation through helmholtz tomography, but the use of ambient noise allows for such measurements at shorter periods and with greater flexibility. Site amplification maps produced with ambient noise show promising variations correlated with local geology, tectonics and velocity models, for periods at least as low as 12 seconds. Additionally, source differences between the incoming and outgoing wavefronts of ambient noise (negative and positive lags) may provide the ability to image noise sources or local scatterers within the continental US. These types of observations, especially site amplification and attenuation, may be used to supplement or improve traditional velocity-based tomographic methods, as they provide different constraints on temperature and density. They may also be useful for seismic hazard studies as they provide direct observations of site effects independent of complex inversions and forward model simulations.