87. Bird, P., and R. S. Stein [2024] Majority of ruptures in large continental strike-slip earthquakes are unilateral: Permissive evidence for hybrid brittle-to-dynamic ruptures, Seismol. Res. Lett., 95, 3406–3415, doi: 10.1785/0220240175.

Abstract.  Finite-element models of neotectonics require transform faults to rupture seismically even where preseismic shear stresses are low, presumably by dynamic-weakening mechanisms. A long-standing objection is that, if a rupture initiated at an asperity with high static friction stresses, which then transitioned to low dynamic-weakening stresses, local stress drop would be near total and on the order of 80 MPa, which is 4×–40× greater than observed. But the 5 Mw ≥ 7.8 transform earthquakes since 2000 initially ruptured on the branch faults of small net slip (Stein and Bird, 2024). If the slip initiates on a branch fault with different slip physics and no dynamic weakening, this solves the stress-drop problem. We propose that most large shallow earthquakes are hybrid ruptures, which begin on branch faults of small slip with high shear stresses, and then continue propagating on a connected dynamically weakened fault of large slip, even where shear stresses are low. One prediction of this model is that most large shallow ruptures should be unilateral. We test this prediction against the 100 largest (Mw ≥ 6.49) shallow continental strike-slip earthquakes 1977–2022, using information from the Global Centroid Moment Tensor and International Seismological Centre catalogs. The differences in time and location between the epicenter and the epicentroid define a horizontal “migration” velocity vector for the evolving centroid of each rupture. Early aftershock locations are summarized by a five-parameter elliptical model. Using the geometric relations between these (and mapped traces of active faults) and guided by a symmetrical decision table, we classified 55 ruptures as apparently unilateral, 30 as bilateral, and 15 as ambiguous. Our finding that a majority (55%–70%) of these ruptures are unilateral permits the interpretation that a majority of ruptures are hybrids, both in terms of geometry (branch fault to transform) and in terms of the physics of their fault slip.

 A diagram of different types of fault

Description automatically generated

Figure 1.  Suggested division of shallow continental fault surfaces by the physics of slip, into Byerlee type, Rice type, and creeping faults. All spatial dimensions are hypothetical. When a single earthquake rupture crosses a boundary between different domains of slip physics, we refer to it as a “hybrid rupture.” Abbreviations: EQ, earthquake; G–R, Gutenberg–Richter. The “weak minerals” mentioned in the Creeping fault-physics section are probably talc and/or montmorillonite, as discussed in the text. Parameters a and b refer to the rate-and-state friction theory of Dieterich (1978, 1979, 1981). See Noda and Lapusta (2013) regarding infrequent huge ruptures with dynamic weakening on creeping faults.

 A screenshot of a cell phone

Description automatically generated

Figure 2.  Graphical summary of effective friction coefficients (defined for quasi-static slip and hydrostatic pore pressure) inferred in eight neotectonic-dynamics modeling studies with the Faults/Plates/Shells family of finite-element (F-E) programs by the first author. Abbreviations Earth2 and Earth5 represent global lithosphere models at different spatial resolutions. All optimal coefficients for the modeled faults are significantly less than “Byerlee’s law” friction of 0.85, requiring that some dynamic- weakening process is active in modeled faults (but not in the continuum between them). In that case, each of these effective friction coefficients can be interpreted as one-half of the ratio of preseismic shear traction to preseismic effective normal stress on the modeled faults.

 A map of a meteorite

Description automatically generated with medium confidence

Figure 3.  Sample map of a continental strike-slip earthquake that we classified as a unilateral rupture. This example is the Mw 7.88 Susitna Glacier–Denali–Totschunda earthquake of 3 November 2002 in Alaska. Traces of active faults from Styron and Pagani (2020) shown with thin red lines. Epicenter from Global Centroid Moment Tensor (Global CMT) (which obtained it from National Earthquake Information Center) shown with black triangle. Epicentroid from Global CMT shown with a focal mechanism on lower focal hemisphere. The migration vector is shown with a heavy black arrow from epicenter to epicentroid and is labeled with migration velocity (if greater than 1.62 km/s). Early aftershocks from ISC are shown with small gray circles. The ellipse summarizing early aftershock locations (Kagan, 2002) is shown with a heavy gray curve.

 A map of a satellite

Description automatically generated with medium confidence

Figure 4.  Sample map of a continental strike-slip earthquake that we classified as a bilateral rupture. This example is the Mw 7.61 Izmit earthquake of 17 August 1999 in Türkiye. Heavy green line is a plate-boundary fault trace from model PB2002 [Bird, 2003]. Other graphical conventions as in Figure 3. In this case, the migration velocity was large but the migration direction was orthogonal to mapped traces of active faults, and therefore implausible. In addition, the distribution of early aftershocks was symmetrical about the epicenter.

 A map of a storm

Description automatically generated

Figure 5.  Sample map of a continental strike-slip earthquake that we classified as an ambiguous rupture. This example is the Mw 7.80 Balochistan earthquake of 24 September 2013 in Pakistan. Graphical conventions as in Figure 3. In this case, the migration velocity was large and roughly parallel to the traces of active faults, but the distribution of early aftershocks was roughly symmetrical about the epicenter. This rupture was classified as Ambiguous because of these conflicting indicators.

 A table with text and images

Description automatically generated with medium confidence

Figure 6.  Our decision matrix, with (row) axis of migration velocity, and (column) axis of aftershock symmetry/asymmetry. Another consideration (in subcells) is whether the migration velocity was parallel to the fault traces and/or the aftershock lineament/ellipse, or perpendicular. Figures in parentheses show the fraction of the 100 events in our study that were assigned to each cell.


SUPPLEMENTS:

(1) an Excel spreadsheet containing Table 1: details of all 100 earthquakes that we studied.

(2) a PDF file of 100 pages, containing 100 maps of all the earthquake ruptures we studied.