Long Term Monitoring of EM Signals Near Parkfield CA

Fluctuations of resistivity and anomalous electromagnetic (EM) signals have often been reported as precursors to earthquakes. Most of these reports are based on anecdotal observations of unusual phenomena associated with distant earthquakes, with anomalous signals that are often orders of magnitude larger than expected based on laboratory measurements. In an attempt to assess the validity of these reports, and to understand how such signals might be generated, anomalous EM signals and resistivity have been monitored since 1995 using magnetotelluric (MT) instruments at the site of the focused earthquake prediction experiment at Parkfield, CA. This EM monitoring array was fully operational and producing high quality data when the long awaited 28 Sept Mw=6.0 Parkfield earthquake occured. The Parkfield MT site, consisting of three magnetic induction coils and replicated orthogonal electric dipoles, was situated within a few kilometers of the northwestern end of the surface rupture, providing unprecedented observations of EM signals at a well calibrated site in very close proximity to a moderate (M~6) earthquake. A second MT site near Hollister, CA was also functioning well before, during, and after the earthquake, providing a callibrated reference for cancellation of normal EM variations due to ionospheric and magnetospheric sources. Analysis of this data using a variety of techniques has not revealed any anomalous signals which were unambiguos precursors to the 28 September Earthquake. In particular, residual E and B fields computed in both the time and frequency domains over the year 2004 using the remote Hollister site for prediction have been analyzed statistically, revealing no long term trends or changes in anomalous EM signal or noise levels in the months to days preceeding the earthquake. Similarly, no anomalous bursts of EM activity are seen immediately preceeding the earthquake. Based on our analysis of these residuals we conclude that any anomalous magnetic signals would have to be at least 2-3 orders of magnitude weaker than those reported to preceed the Loma Prieta earthquake. A strong co-seismic signal was observed. Much of this signal can probably be explained by motion of the sensors in the Earth's main field, although local electrokinetic effects in the vicinity of the sensor may also have contributed to these signals. Statisitcal analysis has also been performed on variations of daily estimates of apparent resistivity and phase at Parkfield. Systematic variations in these parameters are observed, including slow seasonal modulations and more rapid changes on time scales of a few days. However, comparison to weather records suggest that these signals are most probably related to ground moisture and precipitation events, modifying near surface distortion of the electric fields. Multivariate statistical analysis, including principal components and canonical coherence analysis have also been applited to the data, allowing alternate views of temporal variations of signal and noise characteristics. Although there are some anomalous signals deserving more careful study, there is no evidence from this analysis for significant anomalous EM signals preceeding the Parkfield earthquake.