On the geometry of an earthquake fault system

Abstract

Invariants of a two-point correlation function of the seismic moment tensor were used to investigate an earthquake fault system. The geometry of the fault system is significantly different from the standard model of an earthquake fault, i.e. coherent rupture on a planar surface. Contrary to the ‘flat-fault’ model, we see clear evidence for non-planarity of the fault system geometry, and observe that the focal mechanisms of neighboring events may have very different orientation, i.e. they undergo large three-dimensional rotations. Therefore, earthquake deformation models need to be fully three-dimensional and should include large rotations. The spatial behavior of the invariants is approximately the same for earthquakes in different depth intervals: shallow, intermediate, and deep. The temporal behavior of the invariants differs only in that shallow earthquakes are clustered in time, whereas for deeper events the clustering is much less pronounced; as soon as we ‘decluster’ shallow seismicity, the invariants' temporal properties become similar for earthquakes of all depths. This demonstrates that the basic geometrical properties of earthquake rupture do not depend on depth, and therefore they are generally independent of rheological properties of rocks, lithostatic pressure, or the presence of a free boundary for strong shallow earthquakes.