Mechanisms of slip nucleation during earthquakes

https://doi.org/10.1016/S0012-821X(99)00122-3Get rights and content

Abstract

Slip nucleation during earthquakes is apparently analogous to rupture nucleation within an intact rock sample subjected to triaxial loading. The observations indicate that both these nucleation processes initiate within a relatively small volume and in both the slip propagates unstably along a quasi-planar surface. In both processes a single, pre-existing, shear fracture cannot nucleate the large-scale slip, and in both a `process zone' that includes several interacting fractures in a small volume are required to initiate the unstable slip. Both processes require rupture of intact rocks, generate complex fracture geometry, and are associated with intense energy-release rate during slip. Recent observations and analyses are used to correlate rupture nucleation in laboratory tests with nucleation events of large earthquakes. It is proposed that earthquake nucleation occurs by the interaction among multiple fractures within a small volume that develops into unstable yielding of the healed fault zone.

Introduction

Most large earthquakes of the upper crust are slip events along existing fault zones. The nature of this slip has been the subject of many investigations since the last century. Gilbert [1]in his discussion of the 1872 Owens Valley earthquake stated that: ``… strain increases until it is sufficient to overcome the starting friction along the fracture.'' The recognition of `static' and `dynamic' friction has remained a fundamental concept of earthquake mechanisms. Reid [2], on the other hand, described earthquakes in terms of `rupture', proposing that ``we should expect that the slow accumulation of strain would, in general, reach a maximum value and bring about a rupture in a single, comparatively narrow fault-zone.'' Recent analyses of strong motion and teleseismic wave data revealed that earthquake slip is characterized by a leading rupture front [3]. Heaton [3]proposed that a `rupture pulse' moves from the hypocenter area into the locked parts of the fault zone, and he stated that ``since… no slip pulse will propagate unless a slip pulse already exists, the model clearly begs the question of how the rupture pulse starts in the first place'' ([3], p. 16). The present work focuses exactly on this question: the mechanism of slip nucleation along a locked fault.

The concepts of friction and rupture were studied in two different configurations of laboratory models. In the rupture type tests, intact rock samples under confining pressure, σ3, are subjected to increasing axial loading, σ1 (Fig. 1A). A typical test displays a few stages during load increase [4]: non-linear initial flaw closure, linear elastic loading, yielding with associated onset of acoustic emission, unstable rupture along a quasi-planar surface at ultimate strength, profound stress drop and stable sliding (Fig. 1B). In the typical friction type test, two (or three) solid rock blocks subjected to normal stress, σn, are enforced to slip at a prescribed rate (velocity Vi in Fig. 1C). The friction experiments conveyed the development of rate- and state-variable friction, also known as Dieterich law (Fig. 1D) 5, 6. These models are considered here as idealized end-member configurations for rupture (Fig. 1A) and friction (Fig. 1C); therefore, other testing configurations (e.g. cylinders with saw-cuts or rotary rings) are not discussed.

Some observations of earthquakes are pertinent for the comparison with the laboratory analogs. First, the geometry of fracturing during earthquakes could be highly complex as revealed by surface rupture of recent earthquakes, such as Loma Prieta, 1989 7, 8and Landers, 1992 9, 10. Second, earthquakes nucleate within a small volume at the focal region and propagate by `rupture pulse' along a quasi-planar surface [3]. Third, at a depth of a few kilometers and in the presence of hot water, crushed gouge material is likely to heal and re-cement during periods of interseismic quiescence 11, 12. The healed, re-cemented gouge could behave as an intact rock. Finally, recent analysis of velocity seismograms of several tens of earthquakes revealed that the arrival of the main shock P waves is preceded by a distinct, initial phase 13, 14, 15. This phase was interpreted as representing a few small events that facilitate the nucleation of the main earthquakes. These earthquake features are discussed in detail later.

It will be demonstrated below that earthquake nucleation and slip propagation are similar to rupture nucleation within an intact rock (Fig. 1A,B). The rupture concept that was already proposed in 1910 by Reid was overlooked during the last few decades in favor of the rate- and state-friction law. However, recent observations and analyses of both laboratory rock rupture and earthquake nucleation provide the clear tools to establish this concept. In the following sections we discuss in detail the earthquake features that are in accord with the rupture model and we later present a mechanism for earthquake nucleation based on brittle yielding of intact rocks.

Section snippets

Rupture geometry

The 1992, M 7.5 Landers earthquake with maximum horizontal displacement of 6 m produced an 80 km long zone of spectacular surface rupture (Fig. 2). Thanks to a combination of excellent exposures and professional interests, the detailed mapping of this event is accurate and illuminating. Johnson et al. [9], who mapped portions of Landers surface, characterized parts of the rupture zone as belts of shear zones that are 50–200 m wide with subparallel walls and distributed shear (Fig. 2B). The

Experimental observations

It was suggested above that slip nucleation and propagation during earthquakes would be better described in terms of slip nucleation during rock rupture. Slip nucleation within a solid, brittle sample was experimentally explored with three different configurations described below. The first two are based on the assumption that a weak planar surface localizes the shear and initiates macroscopic slip; the third configuration assumes that slip nucleates spontaneously within a solid rock.

Seismic evidence of slip nucleation during earthquakes

An earthquake nucleation stage was detected recently by careful analysis of velocity seismograms of large, shallow earthquakes 14, 15, as well as microearthquakes [13]. In general, a linear velocity increase is anticipated for an earthquake that obeys the scale-independent behavior [15]. The corresponding velocity seismogram should appear as in Fig. 6A for the scheme of slip propagation shown in Fig. 6B. However, Umeda [14]and Ellsworth and Beroza [15]found in the records of few tens of

Conclusions

The nucleation of earthquakes and the associated unstable slip propagation should be analyzed in terms of strength, unstable yielding and rupture of intact rocks. This conclusion is based on observations associated with earthquakes: (1) complexity of the fracture systems in the fault zone; (2) the partial or complete healing of crushed fault gouge at great depth; (3) the high values of energy-release rate during earthquakes. Models of slip nucleation in triaxial experiments and seismic data

Acknowledgements

The concepts presented in this work were developed during discussions with Dave Lockner, Yehuda Ben-Zion, Jim Dieterich, Zvi Garfunkel, Tom Deweres, Amotz Agnon and Gene Scott. None of the above necessarily agrees with these ideas. Liza Heller-Kalai helped in style editing. The critical reviews of Steve Karner and an anonymous reviewer significantly improved the paper. The research was supported in part by the Rock Mechanics Institute, Oklahoma University. [RV]

References (41)

  • A.M. Johnson et al.

    Formation of left-lateral fractures within the Summit Ridge shear zone, 1989 Loma Prieta California, earthquake

    J. Geophys. Res. B

    (1993)
  • A.M. Johnson et al.

    Shear zones formed along long, straight traces of fault zones during the 28 June 1992 Landers California, earthquake

    Bull. Seismol. Soc. Am.

    (1994)
  • S.F. McGill et al.

    Surficial slip distribution on the central Emerson fault during the June 28, 1992 Landers earthquake, California

    J. Geophys. Res. B

    (1999)
  • C. Marone

    The effect of loading rate on static friction and the rate of fault healing during the earthquake cycle

    Nature

    (1998)
  • Y. Iio

    Observations of the slow initial phase generated by microearthquakes — implications for earthquake nucleation and propagation

    J. Geophys. Res. B

    (1995)
  • Y. Umeda

    The bright spot of an earthquake

    Tectonophysics

    (1990)
  • W.L. Ellsworth et al.

    Seismic evidence for an earthquake nucleation phase

    Science

    (1995)
  • A. McGarr et al.

    Observations relevant to seismic driving stress, stress drop, and efficiency

    J. Geophys. Res. B

    (1979)
  • F.M. Chester et al.

    Implications for mechanical properties of brittle faults from observations of the Punchbowl fault zone, California

    Pure Appl. Geophys.

    (1986)
  • F.M. Chester et al.

    Internal structure and weakening mechanisms of the San Andreas Fault

    J. Geophys. Res. B

    (1993)
  • Cited by (46)

    • Earthquake forecast based on its nucleation stages and the ensuing electromagnetic radiations

      2020, Physics Letters, Section A: General, Atomic and Solid State Physics
      Citation Excerpt :

      NSS1 (early nucleation stage): When shear stresses reach critical conditions and shear displacements of the plates initiate, early micro cracking takes place in the weaker in-filling material (Fig. 1a). This initial cracking occurs by compression [19–21], at the space between the corresponding couples of asperities (teeth) – zone A in Fig. 1a1. Note that these cracks are tilted with respect to the fault direction.

    View all citing articles on Scopus
    View full text