Long-term monitoring of electromagnetic signals along the San Andreas Fault

There have been frequent reports of anomalous electromagnetic (EM) signals (or changes in ground resistivity) preceding seismic activity. Although there are a number of physical mechanisms which might plausibly explain such EM-seismic phenomena, observations often seem inconsistent (e.g., precursors are reported to be primarily electrical in some cases, but in others are primarily magnetic) or difficult to reproduce (the phenomenon appears to occur in some special locations or circumstances, but not in others). As a result, the very idea that tectonic activity might have a measurable EM signature is justifiably regarded with skepticism by many scientists. One major problem with many previous reports of EM precursory phenomena has been the lack of redundant measurements which would allow signals to be corroborated and separated from natural magnetotelluric (MT) and cultural sources of EM noise. In an effort to address some of these issues systematically, EM monitoring arrays have been established near the Parkfield segment of the San Andreas Fault. Here we will present results from an array of 2-3 magnetotelluric instruments which has been occupied for much of the past 7 years, with sites near Hollister and Parkfield, CA. Each site is instrumented with an EMI MT-1 system, consisting of three orthogonal induction coil magnetometers, 2-4 orthogonal dipoles to measure induced electric field variations, and Quanterra data loggers digitizing at 40 Hz. With multiple sites it is in principal possible to remove the relatively large scale external MT signals, increasing sensitivity to more localized signals that might be associated with tectonic activity. In the simplest approach, the horizontal magnetic fields at a reference site are used to estimate the MT signal at the local site. We compare several variants on this basic idea, including time and frequency domain approaches, and approaches based on multivariate statistical methods. We find that large scale cultural noise from the San Francisco Bay Area (mostly due to DC trains) complicates residual estimation significantly. During the period that these arrays have been occupied, we have not yet detected any pattern of anomalous signals that is obviously associated with small earthquakes in the Parkfield or Hollister areas.