Very Broad Band Observations of Strain Transients from Borehole Strainmeters

The time constants of geodetically and seismically observable strain signals range from a fraction of a milli-second to a few decades. While modern broad band seismometers are capable of observing signals with period of about an hour and shorter, and GPS is best at detecting strain changes at the period of a few months and longer, several examples show that borehole strainmeters are ideal for signals with a period of a few minutes to a month, and are capable of detecting multi-year transients. Here we present two types of strain transients revealed from our systematic analysis of borehole volumetric strain data recorded along the San Andreas Fault system over the past decade. The first type is a slow earthquake which we define as a slip event with a time-constant ranging from hours to a month. Because of the currently sparse distribution of such instruments, there have been only a few cases of redundant (2 or more instruments, e.g. Linde et al., Nature 1996) detections of slow earthquakes. Here we report four slow events for which we have clear signals but only on 1 instrument per event. The durations of the events range from 4 to 14 days, and the amplitudes range from 200 to 3500 nanostrain. For each event, we considered and eliminated instrumental artifacts and non-tectonic effects such as precipitation and barometric pressure, as possible causes for the signals. In one case there is a simultaneous creep event (within 10 minutes) recorded on 3 creepmeters 30 to 40km northwest of the strainmeter. Each of the strain events is similar in shape and also similar to the Linde et al event, suggesting that this kind of transient may be a characteristic feature of slip along the San Andreas fault system. The second example is a significant multi-year transient in slip-rate along the San Andreas fault at Parkfield. This transient is seen most clearly from two-color geodimeter data, and consists of two phases: decreased slip-rate (by 16 ± 5%) during 1991.0 - 1993.0 followed by increased rate (by 34 ± 6%) from 1993.0 for at least 3 years. This transient is also clear in the other deformation data used in this study: borehole strain dilatometers and creepmeters. The second phase was observed by Gwyther et al. (1996) from tensor strain data. Compared with pre-1991 levels, seismicity is greater during this transient, especially for the period of increased slip-rate, during which the four largest earthquakes in the observing period occurred. These two examples of transients illustrate the broad temporal range of strain events that can be observed from borehole instrumentation. Because of inadequate coverage by instrumentation (borehole strainmeters) capable of detecting them, such events are probably more common than is usually assumed; expansion of our capabilities may shed light on the deformational context of earthquake occurrence