Using Seismic Interferometry to Map Subsurface Velocities in a Tunnel Complex

We map variations of seismic velocity associated with a legacy underground nuclear explosion (UNE) in a tunnel complex at the Nevada National Security Site using seismic interferometry. Velocity anomalies with respect to the surrounding rock exist in the damage zone and possibly at the near surface due to spall. We deployed 8 broadband sensors around a known legacy (1980's) UNE site. Four sensors were emplaced on the surface with a spacing of 10's to 100's meters and four in tunnels at a depth of approximately 250 meters below the surface. The geology of the location consists of a series of ashfall tuffs with near horizontal layering that is extensively mapped. Geologic and geophysical characterization includes boreholes and subsurface (in tunnel) seismic refraction surveys. These initial models combined with 3D finite difference synthetics suggest that it will be possible to image the expected damage. Several months of passive seismic data were collected. Seismic data in the tunnel was of poorer quality than the surface data due to timing drift, periodic noise due to ventilation, and induced 60 Hz electrical signals at some stations. Green's functions between all station pairs were constructed using ambient noise. Results from the surface stations constrain the shallow (< 100 m) velocity structure well. Green's functions between tunnel stations and from surface to tunnel appear to be composed mainly of body waves. Differences are observed between paths crossing the expected damage zone and paths away from the damage and we are working on a model that is consistent with the observations and accounts for artifacts induced by various source of noise (e.g. ventilation and the induced 60 Hz energy). Several local earthquakes were recorded, and we will investigate the use of coda interferometry as an alternate method. Attenuation at tunnel level (observed frequencies up to ~80 Hz) is significantly lower than observed on the surface (observed frequencies up to ~50 Hz) and this may provide higher spatial resolution. The final model will be compared with existing geologic characterization as well as an extensive set of recently collected gravity data, both on the surface and in the tunnel. Prepared by LLNL under Contract DE-AC52-07NA27344. LLNL-ABS-784280.