The magnitude distribution of dynamically triggered earthquakes

PUBLICATIONS Geochemistry, Geophysics, Geosystems RESEARCH ARTICLE 10.1002/2014GC005404 Key Point: Magnitude distributions of triggered and untriggered small quakes are indistinguishable Supporting Information: Readme Supplemental_v5_2 Correspondence to: S. Hernandez, shernan8@ucsc.edu Citation: Hernandez, S., E. E. Brodsky, and N. J. van der Elst (2014), The magnitude distribution of dynamically triggered earthquakes, Geochem. Geophys. Geosyst., 15, doi:10.1002/ 2014GC005404. Received 5 MAY 2014 Accepted 22 AUG 2014 Accepted article online 27 AUG 2014 The magnitude distribution of dynamically triggered earthquakes Stephen Hernandez 1 , Emily E. Brodsky 1 , and Nicholas J. van der Elst 2 Department of Earth and Planetary Sciences, University of California, Santa Cruz, Santa Cruz, California, USA, Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York, USA Abstract Large dynamic strains carried by seismic waves are known to trigger seismicity far from their source region. It is unknown, however, whether surface waves trigger only small earthquakes, or whether they can also trigger large earthquakes. To partially address this question, we evaluate whether current data can distinguish between the magnitude distribution of triggered and untriggered small earthquakes. We use a mixing model approach in which total seismicity is decomposed into two classes: ‘‘triggered’’ events initiated or advanced by far-field dynamic strains and ‘‘untriggered’’ spontaneous events consisting of every- thing else. The b-value of a mixed data set, b MIX , is decomposed into a weighted sum of b-values of its con- stituent components, b T and b U . We utilize the previously observed relationship between triggering rate and dynamic strain amplitude to identify the fraction of triggered events in populations of earthquakes and then invert for b T . For Californian seismicity, data are consistent with a single-parameter Gutenberg-Richter hypothesis governing the magnitudes of both triggered and untriggered earthquakes. 1. Introduction Transient strains delivered by large amplitude seismic waves are frequently associated with seismicity rate increases in the far field at both active margins and stable plate interiors [Hill et al., 1993; Velasco et al., 2008]. This triggering phenomenon is frequently attributed to dynamic stresses since static stresses decay quickly at such large distances (2–3 fault lengths) [King et al., 1994]. One of several outstanding problems associated with remote dynamic triggering is whether the magnitudes of triggered earthquakes are signifi- cantly different from the magnitudes of ambient seismicity. For instance, Parsons and Velasco [2011] investi- gated whether large (M 7) events are capable of dynamically triggering other large (5 M 7) earthquakes in the far field and found that they were unable to observe near-instantaneous triggering of large events in the far field. From these observations, they concluded that dynamic stresses must be incapa- ble of affecting faults above a certain length scale. This conclusion was somewhat upended by the 2012 Sumatra-East Indian Ocean earthquake, which triggered over a dozen remote earthquakes of magnitude 5.5–7.0 over the next several days [Pollitz et al., 2012]. The question remains as to whether this new observa- tion of large triggered earthquakes reflects an extraordinarily rare event or whether triggered earthquakes simply follow the same size distribution as the population of earthquakes as a whole. In the latter case, the observation of large triggered earthquakes is rare, but expected. This is an open access article under the terms of the Creative Commons Attri- bution-NonCommercial-NoDerivs License, which permits use and distri- bution in any medium, provided the original work is properly cited, the use is non-commercial and no modifica- tions or adaptations are made. HERNANDEZ ET AL. Some of the difficulties in interpreting these intriguing events stem from the relative paucity of large earthquakes in general in the catalog. Since earthquake magnitude distributions generally follow a power law, small earthquakes are much more abundant and amenable to statistical studies than large ones. The magnitude distributions of small earthquakes are also in themselves important for seismicity modeling. Cascade models of earthquake sequences take a parsimonious approach of assuming that triggered earthquakes follow an identical magnitude distribution as untriggered ones [e.g., Ogata, 1998]. As a result, the models predict that a cascade of events drawn from a single magnitude distribution can culminate in large, societally significant earthquakes [Ogata, 1998; Felzer et al., 2002, 2004; Helmstetter et al., 2005]. The assumption that the earthquake magnitude distribution is unaffected by the triggering process bears inspection. Assessing the assumption is observationally intricate as it relies on distinguishing the triggered and untriggered populations. Near-field aftershocks from large earthquakes are routinely observed to follow C 2014. The Authors. V