The role of frictional heterogeneities in the earthquake cycle.

Abstract

The determination of earthquake source models for seismic hazard assessments can be difficult and highly uncertain. Information such as recurrence interval of earthquakes for a given magnitude, earthquake origin and probability of multiple-segment rupture is often poorly constrained, especially for large earthquakes (M>7.0).

Physics-based earthquake simulators offer a means of assessing uncertainties of hazard-related input parameters. Importantly, they also provide a pathway for future generations of seismic hazard models in which ground motions are calculated by modelling the seismic wave field, including the effects of variability in the earthquake source process.
Earthquake simulators such as RSQSim generate earthquakes based on rate-and-state friction law. They can generate long-term synthetic earthquake catalogs on a system of known faults, and help us have more understanding of the recurrence intervals of large earthquakes. However, the frictional stresses being inputted to RSQSim are simplified as a uniform number in previous studies. This could produce characteristic earthquakes in the simulated catalogue. Thus, we introduce heterogeneity to the distribution of the frictional stresses to obtain more realistic and less characteristic catalogues for hazard assessment. We focus on a model of the Hikurangi-Kermadec subduction zone New Zealand is located at the southern end of this subduction zone, which has the potential to generate large (Mw>9.0) subduction earthquakes and is therefore very important to consider in seismic and tsunami hazard assessments.

The state coefficient with fixed rate coefficient (b and a in rate-and-state equation) are also tested for reasonable coseismic slip values. The results are compared with the magnitude-frequency distribution (MFD) of the instrumental earthquake catalogue and the empirical scaling laws from global earthquakes.

Considering the trade-off between the fitness to MFDs (Figure 1) and empirical scaling laws (Figure 2), the models with heterogeneous stresses and state coefficients of 0.002 and 0.003 can be the preferred stress settings for synthetic catalogue modeling of Hikurangi-Kermadec subduction interface. Heterogeneity of other stress-related parameters could also be applied to the RSQSim inputs as the next steps to improve the applicability of the synthetic earthquake catalogs to fundamental and applied seismic hazard problems.