Episodic tremor and slip on the Cascadia Subduction Zone: The chatter of silent slip
PDF version [PDF, 795.1 kb, viewer]
This article was originally published in ScienceXpress magazine, May 2003.
Repeated slow slip events observed on the deeper interface of the northern Cascadia subduction zone, at first thought to be silent, have now been found to have unique, non-earthquake, seismic signatures. Tremor-like seismic signals have been found to correlate temporally and spatially with slip events identified from crustal motion data spanning the past six years. During the period between slips, tremor activity is minor or nonexistent. We call this associated tremor and slip phenomenon Episodic Tremor and Slip (ETS) and propose that ETS activity can be used as a real-time indicator of stress loading of the Cascadia megathrust earthquake zone.
The Cascadia subduction zone is a region that has repeatedly ruptured in great thrust earthquakes of moment magnitude (Mw) greater than 8 (1, 2). Recently, slip events have been detected on the deeper (25 to 45 km) part of the northern Cascadia subduction zone interface by observing transient surface deformation on a network of continuously recording Global Positioning System (GPS) sites (3). The slip events occur down-dip from the currently locked, seismogenic portion of the subduction zone (4), and, for the geographic region around Victoria, British Columbia, (Fig. 1), repeat at 13 to 16 month intervals (5). These slips were not accompanied by earthquakes and were thought to be seismically silent. However, using records from the regional digital seismic network, unique non-earthquake signals have been identified which accompany the occurrence of slip. These pulsating, tremor-like seismic signals are similar to those reported in the fore-arc region of Japan (6, 7), but the signals observed in Cascadia correlate temporally and spatially with the six deep slip events over the past seven years. At other times, this tremor activity is minor or nonexistent. These tremors have a lower frequency content than nearby earthquakes, and they are uncorrelated with the deep or shallow earthquake patterns in the region. They have been observed only in the subduction zone region and specifically in the same region as the deep slip events. We refer to this associated tremor and slip phenomenon as Episodic Tremor and Slip (ETS).
Figure 1A
Map of seismic network sites (numbered circles) and approximate source region (shaded ellipse) for tremors used for correlation with observed slips. Note that both tremors and slip have been observed to migrate parallel to the strike of the subduction zone to the south, through, and to the north of this shaded region.
The seismic tremors described here are different from small earthquakes. The frequency content is mainly between 1 to 5 Hz, whereas small earthquakes have most of their energy above 10 Hz. A tremor onset is usually emergent and the signal consists of pulses of energy, often about a minute in duration. A continuous signal may last from a few minutes to several days in duration. Tremors are strongest on horizontal seismographs and move across the seismic network at shear wave velocities. Tremor on an individual seismograph is unremarkable and does not appear different from transient noise due to wind or cultural sources. It is only when a number of seismograph signals are viewed together that the similarity in the envelope of the seismic signal at each site identifies the signal as ETS (Fig. 1).
Figure 1B
Sample seismic records of tremor activity at selected sites. It is the similarity of the envelope of the seismic signal on many seismographs that identifies ETS activity.
The tremor activity migrates along strike of the subduction zone in conjunction with the deep slip events at rates ranging from about 5 to 15 km per day. Sometimes there is a gradual migration, but other times there is a sudden jump from one region of the subduction fault to another. Tremors range in amplitude and the strongest can be detected as far as 300 km from the source region. During an ETS event, tremor activity lasts about 10 to 20 days in any one region and contains tremor sequences that have amplitudes at least a factor of 10 larger than the minimum detectable tremor amplitude. Because of the emergent nature of the tremors, they are difficult to locate as precisely as nearby earthquakes using standard earthquake location procedures. Arrival times of coherent bursts detected across the network suggest source depths of 20 km to 40 km with uncertainties of several kilometres. Deeper solutions are to the north east and all solutions are near the subduction interface or just above it. The fact that surface displacement patterns have been satisfactorily modeled using simple dislocations of 2 to 4 cm on the plate interface bounded by the 25 km and 45 km depth contours (3) strongly suggests a spatial correlation with the source region of tremors.
To establish the temporal correlation between the slip events and the tremor activity (Fig. 2), the timing of six slip events observed on southern Vancouver Island since 1997 was established by cross-correlating changes in the east-west component of the Victoria GPS site (ALBH) with a symmetric 180-day saw tooth function which replicated an average slip time-series (8). This approach allowed the resolution of the mid-point of the slips to within 2 days. The duration of the slips, estimated from slope breaks in the ALBH time series, varied from 6 to 20 days. Seismic data were then examined at corresponding times to check for tremor activity. In each case, sustained tremor activity on southern Vancouver Island was observed to coincide with the occurrence of slip (Fig. 2). For five of the slip events, tremors continued to migrate north along the axis of Vancouver Island moving beyond the region of diagnostic GPS coverage.
Figure 2
Comparison of slip and tremor activity observed for the Victoria area. Blue circles show day-by-day change in the east component of the GPS site ALBH (Victoria) with respect to Penticton which is assumed fixed on the North America plate. Continuous green line shows the long-term (interseismic) eastward motion of the site. Red saw-tooth line segments show the mean elevated eastward trends between the slip events which are marked by the reversals of motion every 13 to 16 months. Bottom graph shows the total number of hours of tremor activity observed for southern Vancouver Is. within a sliding 10-day period (complete annual records examined from 1999 onward). 10 days corresponds to the nominal duration of a slip event. The graph ends 10 March 2003, with the slip and tremor activity that was predicted for the spring of 2003.
To test a one-to-one correspondence, continuous digital seismic data from the beginning of 1999 to the end of the 2003 tremor event were examined to look for significant tremor activity outside the time windows of the slip events. None was found for southern Vancouver Island, although a few periods with scattered low-amplitude tremor activity were observed in most months. Sustained large amplitude tremor events have also been observed in mid and northern Vancouver Island, both as continuations of tremors migrating from the south and as independent tremor events. Implications are that the ETS process occurs over the full length of the northern Cascadia subduction zone but GPS coverage at the northern end is sparse and surface displacements indicative of slip at depth have not been identified.
The cause of the tremor is not clear. Obara (6) has suggested fluids as a source for similar tremors in Japan. Because the tremor observed in Cascadia is mainly composed of shear waves and because it correlates with slip that is relieving stress due to convergence (3), a shearing source seems most likely. However, because of the abundance of available fluids from the subducting plate in the subduction forearc (9), fluids may play an important role in the ETS process by regulating the shear strength of rock.
If the one-to-one correlation between transient slip and seismic signatures proves to be robust, then the tremor-like seismic signals can provide a real-time indicator of the occurrence of slip. Because slip events on the deep slab interface increase the stress across the locked plate interface located up-dip, it is conceivable that a slip event could trigger a large subduction thrust earthquake (10, 11). Consequently, the onset of ETS activity could lead to recognized times of higher probability for the occurrence of megathrust earthquakes in the Cascadia subduction zone.
- B. F. Atwater et al., Earthquake Spectra 11, 1 (1996).
- K. Satake, K. Shimazake, Y Tsuji, K. Ueda, Nature 378, 246 (1996).
- H. Dragert, K. Wang, T. S. James, Science 292, 1525 (2001).
- H. Dragert, R. D. Hyndman, G. C. Rogers, K. Wang, Geophys. Res. 99, 653 (1994).
- M. M. Miller et al., Science 295, 2423 (2002).
- K. Obara, Science 296, 1679 (2002).
- A. Katsumata, N. Kamaya, Geophys. Res. Lett. 30, 1020 (2003).
- Precise dates for slip events were derived as follows. First, linear regression was used to estimate the long-term linear trend of the observed time series, corrected for offsets introduced by GPS antenna changes. Next, approximate dates (to within a week) for slip transients were identified from a visual inspection of the time series, and regression was again used to estimate an annual signal, the magnitude of the offsets at the time of the slips, and the average linear trend for the time periods between slip events. A 180-point symmetric sawtooth time series was generated using the average inter-slip trend and the average value of the offsets. This abbreviated sawtooth function was crosscorrelated with the original observed (but de-trended) time series and the maxima for the cross-correlation function were used to establish the mid-point of slip events.
- S. M. Peacock, Science 248, 329 (1990).
- W. Thatcher, Nature 299, 12 (1982).
- A. T. Linde, P. G. Silver, Geophys. Res. Lett. 16, 1305 (1989).
- We thank M. Schmidt and Y. Lu for their support in GPS network operations and R. Baldwin, T. Claydon and R. Hall for their support in seismic network operations. The help of J. Cassidy and W. Bentkowski in the location tremors is much appreciated. This paper is Geological Survey of Canada contribution no. 2003022. Copyright, Her Majesty the Queen in right of Canada (2003). 21 March 2003; accepted 29 April 2003