Persistence of Coseismic Rupture Asperities as inferred from Interseismic Geodetic Observations

Inversions of geodetic data from interseismic periods produce models that are locked over spatially smooth and extensive regions, in contrast to the smaller discrete asperities estimated by earthquake source studies. Such smooth, broad regions may be a consequence of a lack of model resolution and the resulting need for regularization inherent to the use of onshore geodetic data. It is also possible that the inferred interseismically coupled regions are larger than the collective asperity sizes for known earthquakes due to an incomplete earthquake catalogue, and may imply the potential for large earthquakes in the future. Hence, the different levels of apparent coupling implied by interseismic and seismic-source inversions have very different implications for regional seismic hazard. Here, we test the hypothesis that mechanical coupling on asperities inferred from the locations of past earthquakes alone is sufficient to explain currently available geodetic observations or alternatively, that these data require additional regions of the megathrust to be coupled. Underlying our approach is the assumption that known asperities persist across multiple earthquake cycles. We use a 3-D mechanical model of stress-dependent interseismic creep along the megathrust, considering frictional rheologies and known spatio-temporal distribution of large earthquakes (Hetland et al. 2009). These mechanical models predict that late in the seismic cycle, there are relatively smooth, long wavelength regions of very low slip-rates on the megathrust interface surrounding these asperities, owing to the "stress-shadow" effect of seismic ruptures. We test whether such "physical" smoothing preserves any signature of the original asperities, in contrast to the artificial smoothing produced by model regularization in inversions of interseismic geodetic data. As a first step to applying the above modeling approach to realistic faults, we have extended the method to handle arbitrary, geo-spatially referenced 3D fault geometries embedded within an elastic half-space. We present benchmark test results of slip evolution on planar faults, as well as preliminary comparisons of model predictions with geodetic observations for the 3D PAC-EUR megathrust interface beneath the Japan Trench off Tohoku.