Earthquake rupture dynamics frozen in exhumed ancient faults

Abstract

fault zone14. Fault segments carrying only pseudotachylytes (that is, Most of our knowledge about co-seismic rupture propagation is derived from inversion and interpretation of strong-ground- not associated with cataclastic precursor) have displacements of less motion seismograms1–3, laboratory experiments on rock4,5 and than 1.5 m, which are typical of seismic fault ruptures of about 10 km rock-analogue material6, or inferred from theoretical and numeri- in length17. Field and microstructural data indicate that pseudota- cal elastodynamic models7–9. However, additional information on chylyte is produced at the final stage of fault slip14. Evidence of dynamic rupture processes can be provided by direct observation multiple generations of pseudotachylytes is rare along the Gole of faults exhumed at the Earth’s surface10. Pseudotachylytes Larghe fault14; pseudotachylytes within each individual fault segment (solidified friction-induced melts11,12) are the most certain fault- are therefore the result of a single seismic rupture. rock indicator of seismicity on ancient faults13. Here we show how Many pseudotachylyte veins were injected into the host tonalite the asymmetry in distribution and the orientation of pseudo- from pseudotachylyte-bearing fault segments; an example of a fault tachylyte-filled secondary fractures around an exhumed fault can segment with injection veins is shown in Fig. 2. We measured the be used to reconstruct the earthquake rupture directivity, rupture orientation of 624 injection veins that branch off 28 different fault velocity and fracture energy, by comparison with the theoretical segments, in exposures sub-parallel to the fault slip direction. Linear dynamic stress field computed around propagating fractures. In fault segments that were at least 2–3 m away from the closest segment particular, the studied natural network of pseudotachylytes is were selected for the measures, to avoid potential perturbations due to fault irregularity18 and interference with neighbouring faults. The consistent with a dominant propagation direction during repeated cumulative data from all measures (Fig. 3) reveals two dominant seismic events and subsonic rupture propagation close to the orientations of injection veins, at about 30–2108 (referred to as set 1) Rayleigh wave velocity. and 90–2708 (set 2) with respect to the fault trace. Both vein sets are asymmetrically distributed with respect to the fault trace with distinct dominance (67.7%) of the veins intruded into the southern bounding block. The veins of set 1 (mostly less than 2 mm thick and less than 50 cm long) intruded pre-existing minor cataclastic faults