T043-06
Constraining dynamic rupture models using on-fault geological observations from the 2011 Mw 6.6 Fukushima earthquake, Japan.

Monday, 14 December 2020: 08:50
Virtual
Jesse Kearse, Victoria University of Wellington, Dept. of Earth Sciences, Wellington, New Zealand, Clarrie Macklin, GNS Science-Institute of Geological and Nuclear Sciences Ltd, Lower Hutt, New Zealand and Yoshihiro Kaneko, Graduate School of Science, Kyoto University, Kyoto, Japan
Abstract:
Coseismic changes in slip direction recorded by curved slickenlines on fault surfaces are commonly observed following surface-breaking earthquakes. Such observations represent a dynamic record of seismic slip, and may provide a new set of geological constraints on simulations of spontaneous dynamic rupture and hence earthquake dynamics. We test this hypothesis by constructing a detailed dynamic rupture simulation of the 2011 M w 6.6 Fukushima earthquake (Japan), with the objective of reproducing the well-documented field observations of curved slickenlines that formed during coseismic fault displacement at the ground surface (Otsubo et al., 2013). We consider relatively simple dynamic rupture models with a dipping fault embedded into a homogeneous or layered elastic halfspace. Among a wide range of model parameters tested, we find a model with shallow (less than 1.5 km depth) low-velocity layers derived from a regional 3D velocity model combined with a depth-dependent prestress result in a curved surface slip history that matches the curved slickenline observations. This same model also generates a slip distribution and rupture propagation direction consistent with published inversions that are constrained with seismological and geodetic data (Fukushima et al., 2013; Tanaka et al., 2014). However, unlike previous theoretical studies, changes in slip direction occur due to the emergence of multiple slip fronts within the shallow low-velocity medium. Our results indicate that on-fault geological observations can supplement seismological studies of earthquake dynamics beyond traditional datasets.