SCEC Award Number 16053 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title Using low frequency earthquake families on the San Andreas as deep creep meters
Investigator(s)
Name Organization
Amanda Thomas University of Oregon N. Beeler United States Geological Survey David Shelly United States Geological Survey Roland Bürgmann University of California, Berkeley
Other Participants Rand Michie
SCEC Priorities 5d, 5e, 2b SCEC Groups Seismology, FARM, EFP
Report Due Date 03/15/2017 Date Report Submitted 06/30/2017
Project Abstract
 The central section of the San Andreas fault hosts tectonic tremor and low-frequency earthquakes (LFEs) similar to subduction zone environments. LFEs are often interpreted as persistent regions that repeatedly fail during the aseismic shear of the surrounding fault allowing them to be used as creepmeters. We test this idea by using the recurrence intervals of individual LFEs within LFE families to estimate the timing, duration, recurrence interval, slip, and slip rate associated with inferred slow slip events. We formalize the definition of a creepmeter and determine whether this definition is consistent with our observations. We find that continuous families and the short timescale of episodic families appear to be inconsistent with our definition of a creepmeter. However, when these families are evaluated on timescales longer than the interevent time a meaningful relationship between recurrence interval and duration emerges and these events can be used to meter slip. A straight-forward interpretation of episodic families is that they define sections of the fault where slip is distinctly episodic in well defined SSEs that slip 15 times the long term rate. In contrast, the frequent short-term bursts of the continuous and short-timescale episodic families likely do not represent individual creep events but rather are persistent asperities that are driven to failure by quasi-continuous creep on the surrounding fault. Finally we find that the moment-duration scaling of our inferred creep events are inconsistent with the proposed linear moment-duration scaling. However, caution must be exercised when attempting to determine scaling with incomplete knowledge of scale.
Intellectual Merit This work formalized the definition of a creep meter and applied it to a declustered catalog of over one million low-frequency earthquakes on the deep San Andreas fault. This definition, along with the duration and recurrence intervals of episodes (bursts of LFEs), can be used to estimate the velocity of deep slow slip events in environments where slow slip is below the geodetic detection limit. This work also shows that not all LFE families reflect surrounding larger scale fault creep, many families simply represent persistent asperities driven to failure by long-term fault creep, akin to shallow repeating earthquakes. We also investigate the potential role of interepisode slip and the moment-duration scaling of our inferred slip events.
Broader Impacts This award facilitated collaboration between an early-career female PI, and early career postdoctoral researcher, and three established scientists. Also, this study contributed to our understanding of the deep roots of faults and in particular, whether LFEs can be used in real-time to measure ongoing slip transients. A number of recent studies have suggested that deep slow slip has preceeded significant shallow earthquakes and our results suggest that some LFE families could be used for real time monitoring.
Exemplary Figure Figure 2: Median duration vs recurrence interval of slip episodes in the 88 LFE families color coded by their MFD75 value (high values correspond to continuous families while low values correspond to episodic families). The squares and triangles represent episodes defined by the long and short timescales while circles represent continuous families. Events separate into three populations. The continuous and short timescale episodic families show no systematic trend while episodic families with episodes defined by the long tr timescale have td that increases as a function of tr. The dashed line is the best fit to this group of families and has a slope of 0.066 corresponding to V/VL~15.