What causes tectonic tremor? Investigating tremor's origins and implications with seismology

Lead Research Organisation: University of Oxford
Department Name: Earth Sciences

Abstract

Over the past two decades, improving seismic and geodetic data have revealed that many faults accumulate their slip via a suite of phenomena that are not predicted by conventional friction laws: via slow earthquakes, or fault slip events whose average slip rates are between 0.1 microns/s and 1 mm/s, a factor of 1 thousand to 10 million slower than the 1 m/s slip rates typical of earthquakes. Slow earthquakes are now found at most subduction zones, where they accommodate about half of the plate interface slip in the region down-dip of the seismogenic zone. But currently, we do not know which fault zone processes generate the aseismic slip we observe in slow earthquakes.

It is important to improve our understanding of slow earthquakes because they occur next to the seismogenic zone. They are capable of triggering large and damaging earthquakes. In this project, we focus on the smallest but most abundant slow earthquakes: tremor. Tremor consists of hundreds to millions of small, closely spaced, slow earthquakes. The earthquakes can be rapidly observed and could be used to track larger-scale aseismic slip variations and to assess whether that slip could trigger hazardous seismic slip. But like other slow earthquakes, tremor remains poorly understood. The goal of this project is to determine which physical process creates tremor and limits its slip rates to around 1 mm/s.

Several explanations of tremor's low slip rates have been proposed. It is possible that tremor is governed by the same frictional sliding process that governs normal earthquakes. Tremor may be slow only because the fault's frictional strength or normal stress is low, and thus is unable to drive rapid slip. Alternatively, a more novel physical process could limit tremor's slip speeds. Changes in pore fluid pressure might pull the fault shut, inhibiting rapid slip. Or tremor could be a collection of failed earthquake nucleations, which arise because of stress perturbations on a nominally stable fault.

In the proposed work, we will use targeted seismological analysis to assess five proposed models of tremor generation. We will test specific model predictions using high-quality seismic data from some of the best-observed tremor in the world: that near Parkfield, CA.

To test our model predictions, we will first examine how tremor is related to shorter and longer slow earthquakes. If tremor is governed by the same novel fault zone physics that governs larger slow earthquakes, there should be a continuum of slow earthquakes with a wide range of sizes and slip rates. The presence or absence of the continuum will be important for constraining the processes governing large and small slow earthquakes, as only a few of the proposed models of large slow earthquakes are consistent with the continuum's wide-ranging slip rates. We will search for 0.05 to 1-second-long events in this continuum using recently developed seismic analysis techniques. And we will examine the clustering of tremor, in order to (1) identify larger, hours-long slow earthquakes potentially within the continuum and (2) to constrain the relationship between tremor and larger-scale slip.

Finally, to further test the models, we will move into the details of individual tremor events and probe the evolution of slip in individual tremor earthquakes. We will closely examine the seismic signals produced by tremor in order to determine how tremor's earthquakes' durations, sizes, and complexities vary from event to event. These data will let us determine how much of tremor's properties are controlled by particular rheologies and how much is due to local fault zone structure.

By pursuing a suite of features that can test our models, we will be able to determine which physical processes generate the numerous small earthquakes that constitute tremor, so that we may better understand slow earthquake slip and more confidently use tremor to track large-scale slip at depth.

Planned Impact

The potential non-academic beneficiaries of our work are
1. geological surveys who create seismic hazard maps, who could use our physical models and our tremor locations as they incorporate tremor into seismic hazard estimates.
2. oil and gas exploration companies, who could use our seismic techniques to locate microseismicity and use our physical models to understand how microseismicity constrains deformation in exploration areas.

An understanding of the physical process that generates tremor will improve seismic hazard estimates because (1) the small earthquakes in tremor can be used to track large-scale aseismic slip, which is capable of triggering large, damaging earthquakes and because (2) tremor observations could constrain how individual earthquakes nucleate. Our physical understanding of tremor will help reveal how good tremor is at tracking large-scale slip and which properties of tremor we should use to infer that slip.

We note that it will be only after years of work and vetting by the community that our understanding of tremor can be implemented in published seismic hazard estimates. However, there are several steps we can take now to facilitate the uptake of our understanding and encourage tremor's use in seismic hazard.
First, we will communicate our results to the scientists who create and publicise seismic hazard estimates. In California, where our analysis is focused, hazard estimates are created and publicised primarily by members of the United States Geological Survey and the Southern California Earthquake Center. We regularly interact with members of the USGS and SCEC via other collaborations and at meetings such as AGU. Project partner David Shelly is a research geophysicist at the USGS. To further facilitate this communication, PI JCH will attend the SCEC annual meeting in 2020.

Over the next ten years, seismic hazard models are likely to incorporate an increasing amount information and physics. An ideal goal for us will be to have our tremor understanding incorporated into one of the earthquake forecasts involved in CSEP, the Collaboratory for the Study of Earthquake Predictability. We will engage with a range of researchers who could incorporate tremor into their forecasts and encourage them to use and test the tremor-slip relationships we obtain.

We will also maintain a publicly available catalogue of the tremor we identify in the Parkfield area, so that it can be used in seismic hazard forecasts and tests. We will commit to making our tremor detection code robust, so that we can continue to systematically search for tremor for at least five years after the end of the project, downloading data and publishing new detections at least every six months.

Another community that could benefit from this work is the hydrocarbon industry. They could use the seismic techniques we develop to detect, locate, and analyse microseismicity. Our techniques may be especially useful in probing the numerous small earthquakes that are created during unconventional shale gas plays.
The hydrocarbon industry might also benefit from our understanding because transient aseismic slip events are observed at oil wells, especially when the stress state has been perturbed by large-scale oil extraction or fluid injection. Our results will provide better insight into how small earthquakes reflect larger-scale deformation.

To make our work to available to the industry community, we will post the seismic processing codes we develop online, along with an explanation that is accessible to the industry. And we will present our analysis directly to the industry community. As part of our ongoing slow slip investigations, we already have plans to present the seismic technique development at the EAGE meeting in 2020. In that presentation, we will incorporate the results from our search for tremor and other slip events in Parkfield, and we will discuss the implied relationships between seismic and aseismic slip.

Publications

10 25 50