Date of this Version
A central goal of seismology is to understand the physics of earthquakes and other sources of seismic waves in the Earth. We would like to understand how dynamic instabilities are nucleated, how they evolve in space and time, and how they come to rest. To achieve this goal, we need observations that are truly broadband with respect to source process time scales. Because the high-frequency limit of a seismogram directly controls the spatial scale at which we can resolve these processes, the requirement for “broadband” means bandwidth that is sufficient to record the shortest pulse produced by the physical system (a delta function being the ultimate broadband signal). Although there is considerable uncertainty at present about the upper frequency limit needed to capture dynamic processes, it is clearly well above the frequency range of standard seismological instrumentation (typically 30–40 Hz for 100 sample-per-second data). Even when instruments are capable of observing frequencies above 1 KHz, they must be sited close enough to the source to overcome the attenuation of the high frequency waves due to scattering and anelastic loss during propagation from the source. The natural solution to this problem is to emplace the instrumentation within the near-field of the source in boreholes and deep mines. This paper presents a review of some recent results from three deep (>2 km) boreholes in California. The three boreholes considered are the Long Valley Exploratory Well, the San Andreas Fault Observatory at Depth (SAFOD) Pilot Hole, and the SAFOD Main Hole.
The Long Valley Exploratory Well (LVEW) is a 3.0-km-deep research drill hole located near the center of Long Valley caldera in eastern California. The well was drilled in a series of stages beginning in 1989, and completed to a total depth of 2996 m in 1998 (Sackett et al., 1999) as part of the International Continental Drilling Program (ICDP) Long Valley Coring Project. Prior to the final stage of drilling, a 3-component 10-Hz seismometer, installed at a depth of 2050 m from September 1997 through May 1998, recorded tens of thousands of local events during the 1997–1998 seismic crisis in Long Valley Caldera (Prejean and Ellsworth, 2001). The value of recording at depth is clearly evident in the comparison of near-source recordings made at the surface and deep underground (Fig. 1).