T51J-04
Connecting depths of seismicity, fault locking, and coseismic slip using long-term fault models
Abstract:
The depth of faulting and its temporal evolution are important for understanding fault physics and seismic hazard. The seismicity cut-off depth (Dseis) and (interseismic) fault locking depth estimated geodetically (Dgeod) give independent constraints on the locked-creeping (LC) transition on faults. How the two are related to each other and to the depth extent Drupt of large earthquake ruptures is still poorly understood.We explore the relation between Dseis, Dgeod and Drupt in fault models governed by rate-and-state friction and enhanced dynamic weakening. We find that: (i) transition from locked zones (e.g., creeping rates < 0.1 Vpl) to zones creeping with near-plate rates (> 0.9 Vpl) occurs over a certain depth range, which can be significant; (ii) Dseis is predominantly affected by the location and amplitude of the stress concentration front (SCF) near the top of the LC transition, while Dgeod depends on the spatial distribution of fault slip rates; (iii) Dseis either stays in place or becomes shallower due to the up-dip migration of the SCF, while Dgeod tends to deepen as the stress shadowing region expands throughout the interseismic period; and (iv) Drupt and Dseis can differ due to deep seismic slip below the seismogenic zone, while the discrepancy between Dgeod and Drupt depends on the extent and amplitude of postseismic slip. Therefore, Dseis and Dgeod reflect different aspects of fault behavior and could diverge, especially in cases of deeper penetration of earthquakes and/or significant afterslip. Assuming Dgeod equal to Dseis could lead to an underestimation of fault slip rates and a discrepancy with geological estimates.
Our current work is directed towards (a) establishing whether Dgeod could constrain the maximum depth extent of Drupt, as suggested by our simulations so far, and (b) exploring how the presence of the realistic LC transition affects interpretation of the inferred Dgeod and Dseis based on simplified models.