Repeating Earthquakes With Remarkably Repeatable Ruptures On The San Andreas Fault At Parkfield

We calculate rupture directivity and velocity for earthquakes in three well-recorded repeating sequences (2001-2016) on the San Andreas Fault at Parkfield using P waves from borehole recordings and the empirical Green's function method. The individual events in each sequence all show the same directivity; the largest magnitude sequence (M similar to 2.7, 8 events) ruptures unilaterally NW (at similar to 0.8Vs), the second sequence (M similar to 2.3, 9 events) ruptures unilaterally SE, and the smallest magnitude sequence (M similar to 2, 11 events) is less well resolved. The highly repetitive rupture suggests that geometry or material properties might control nucleation of small locked patches. The source spectra of the M similar to 2.7 sequence exhibit no detectable temporal variation. The smaller M sequences both exhibit a decrease in high-frequency energy following the M6 earthquake that recovers with time. This could indicate a decrease in stress drop, an increase in attenuation, or a combination of the two, followed by gradual healing.Plain Language Summary Sequences of small earthquakes with very similar seismograms have been observed on the San Andreas Fault at Parkfield and many other faults that are also observed to be creeping. The similarity of the seismograms suggests that these earthquakes represent repeating slip on overlapping pieces of fault. Under this assumption they have been used to investigate changes in slip conditions and the slip rate on major faults with time. To date, most studies have treated these repeating earthquakes as just points because most are too small and too poorly recorded to observe details of the rupture area or direction. Here we identify three exceptionally well-recorded repeating sequences and are able to resolve the direction of rupture. All the earthquakes in one sequence propagate toward the NW, and all those in another propagate to the SW. The rupture velocities are similar to those of much larger earthquakes. This improved resolution of the earthquake sources will help constrain modeling of the rupture and hence our understanding of the factors that control earthquake nucleation and dynamics.