Variability of earthquakes, faults, and the resulting seismic hazard

It will probably come as no surprise to any geologist that earthquakes and tectonic faults are highly variable and complex phenomena. Active (and inactive) faults rarely occur in isolation and have variable geometries, kinematics, and slip rates. This fact, coupled with fault interaction, means that earthquakes are variably distributed in time and space over tectonically active regions. These complexities contribute to some of the challenges associated with probabilistic seismic hazard assessment. In this talk, I will mostly focus on the insights we can gain about normal faulting and earthquake behaviour from the central Apennines of Italy, and how our results can be used to inform potential variability and therefore uncertainties in other tectonically active regions.

Where there are good historical records of earthquakes available, interactions between individual earthquakes and/or faults can be studied by modelling Coulomb stress transfer and calculating how stress accumulates over time. Our results from modelling the ~700 year historical sequence in central Italy supports the hypothesis that past earthquakes affect the location of subsequent events, though the models are not enough to fully hindcast the historical record of damaging events. Furthermore, these models also indicate that variations in the geometry of individual faults and the regional fault network may affect earthquake occurrence.

The geometric complexity of normal faults is evident both via surface observations of faults scarps and from sub-surface techniques such as seismic reflection. Observations from normal faults in the central Italian Apennines indicates that the geometry of normal faults may affect the magnitude and patterns of both short-term (coseismic) and long-term (15 kyr) slip and throw. These geometric variations may also control the propagation and/or termination of earthquakes, as evident from recent earthquakes both in Italy and globally. Extending our observations into the sub-surface using seismic reflection datasets of both active and inactive faults gives us further insights into how fault geometry and slip rates evolve over much longer (million year) time scales than is possible from field studies alone.

These observations of variability are giving us insights into some of the physical mechanisms controlling fault behaviour over a range of timescales and earthquake occurrence. Knowing the variability helps us to understand the uncertainties of inputs into seismic hazard assessment and can also help us to think about other possible approaches to seismic hazard, such as time-dependent hazard.