Spatial migration of temporal earthquake clusters driven by the transfer of differential stress between neighbouring fault/shear-zone structures

Uncertainty concerning the processes responsible for slip-rate fluctuations associated with temporal clustering of surface faulting earthquakes is a fundamental, unresolved issue in tectonics, because strain-rates accommodated by fault/shear-zone structures are the key to understanding the viscosity structure of the crust and seismic hazard. We constrain the timing and amplitude of slip-rate fluctuations that occurred on three active normal faults in central Italy over a time period of 20–30 kyrs, using in situ 36Cl cosmogenic dating of fault planes. We identify five periods of rapid slip on individual faults lasting a few millennia, separated time periods of up to 10 millennia with low or zero slip-rate. The rapid slip pulses migrated across the strike between the faults in two waves from SW to NE. We replicate this migration with a model where rapid slip induces changes in differential stress that drive changes in strain-rate on viscous shear zones that drive slip-rate variability on overlying brittle faults. Earthquakes increase the differential stress and strain-rate on underlying shear zones, which in turn accumulate strain, re-loading stress onto the overlying brittle fault. This positive feedback produces high strain-rate episodes containing several large magnitude surface faulting earthquakes (earthquake clusters), but also reduce the differential stress on the viscous portions of neighbouring fault/shear-zones slowing the occurrence of large-magnitude surface faulting earthquakes (earthquake anticlusters). Shear-zones on faults experiencing anticlusters continue to accumulate viscous strain at a lowered rate, and eventually this loads the overlying brittle fault to failure, initiating a period of rapid slip through the positive feedback process described above, and inducing lowered strain-rates onto neighbouring fault/shear-zones. We show that these patterns of differential stress change can replicate the measured earthquake clustering implied by the 36Cl data. The stress changes are related to the fault geometry in terms of distance and azimuth from the slipping structure, implying that (a) strain-rate and viscosity fluctuations for studies of continental rheology, and (b) slip-rates for seismic hazard purposes are to an extent predictable given knowledge of the fault system geometry.