Reconsidering Fault Slip Scaling

Tuesday, 15 December 2015: 17:00
301 (Moscone South)
Joan S Gomberg, US Geological Survey, Seattle, WA, United States, Aaron Wech, Alaska Volcano Observatory, Anchorage, AK, United States, Kenneth C Creager, Applied Physics Laboratory University of Washington, Seattle, WA, United States, Kazushige Obara, Earthquake Research Institute, University of Tokyo, Tokyo, Japan and Duncan Carr Agnew, University of California San Diego, La Jolla, CA, United States
The scaling of fault slip events given by the relationship between the scalar moment M0, and duration T, potentially provides key constraints on the underlying physics controlling slip. Many studies have suggested that measurements of M0 and T are related as M0=KfT3 for ‘fast’ slip events (earthquakes) and M0=KsT for ‘slow’ slip events, in which Kf and Ks are proportionality constants, although some studies have inferred intermediate relations. Here ‘slow’ and ‘fast’ refer to slip front propagation velocities, either so slow that seismic radiation is too small or long period to be measurable or fast enough that dynamic processes may be important for the slip process and measurable seismic waves radiate. Numerous models have been proposed to explain the differing M0-T scaling relations. We show that a single, simple dislocation model of slip events within a bounded slip zone may explain nearly all M0-T observations. Rather than different scaling for fast and slow populations, we suggest that within each population the scaling changes from M0  proportional to T3 to T when the slipping area reaches the slip zone boundaries and transitions from unbounded, 2-dimensional to bounded, 1-dimensional growth. This transition has not been apparent previously for slow events because data have sampled only the bounded regime and may be obscured for earthquakes when observations from multiple tectonic regions are combined. We have attempted to sample the expected transition between bounded and unbounded regimes for the slow slip population, measuring tremor cluster parameters from catalogs for Japan and Cascadia and using them as proxies for small slow slip event characteristics. For fast events we employed published earthquake slip models. Observations corroborate our hypothesis, but highlight observational difficulties. We find that M0-T observations for both slow and fast slip events, spanning 12 orders of magnitude in M0, are consistent with a single model based on dislocation theory. Remarkably, a few parameters can explain the first-order features of these observations: the average slip zone width or boundaries, slip front propagation velocities, and stress drop. While a continuous distribution of slip modes likely exists, this distribution may be bimodal.