Validating the link between fault geometry and slip distribution

Fault geometry is increasingly regarded as a key parameter that affects all aspects of the earthquake cycle, yet the in-situ geometry of active faults remains poorly resolved, and they are often modeled as largely planar. On the other hand, recent advances in sensing and computational abilities enable measuring the co-seismic slip of large earthquakes in high resolution, both on the surface and at seismogenic depths. Previous analytical and numerical studies showed that the slip distribution of an earthquake is mechanically linked to the fault geometry through its curvature, though this link has not yet been verified. Here we show experimental verification of this link, by measuring at high precision the shear slip profiles along non-planar interfaces in laboratory earthquakes. We trigger dynamic shear ruptures that propagate along the interface between two loaded and matching PMMA plates and, using image correlation of ultrahigh-speed photography, resolve the propagating ruptures and the resulting displacements. The plates are pre-cut along a desired geometry, and we compare the slip calculated from the geometry and the experimental shear slip at each pixel along the interface. We show that, as predicted analytically, fault curvature is correlated to slip gradient when the plates are in contact and anti-correlated when there is opening. These relationships are clearly visible in our results at all measured scales regardless of the rupture complexity. Our results suggest that variations of the co-seismic slip along a fault can be highly indicative of the in-situ fault geometry and, alternatively, that slip distribution may be predicted along a fault with known geometry.