Andrew Howell, Tim McLennan, Camilla Penney, Andrew Nicol, Hannu Seebeck, Russ J Van Dissen, Charles A Williams, Bill Fry

Fault geometry and the connectivity between faults at depth are both important controls on the nucleation, propagation and arrest of earthquake rupture, so modelling these parameters accurately is essential to models of the earthquake cycle. However, simulations involving complex three-dimensional (3D) fault systems rarely explore the sensitivity of results to uncertainties in geometry and connectivity — either in terms of modelled earthquake characteristics or impacts such as ground shaking and surface deformation. In many cases, geometry-related sensitivity testing is limited because it is challenging to construct a suite of alternative fault models that span the range of plausible fault geometries, intersections and connections; such alternative models are especially difficult to construct for systems where faults truncate or cross-cut each other at depth. We present a new, semi-automated method that simplifies creation of 3D models of networks of tens or hundreds of faults, combining open-source python tools with the meshing capabilities of Leapfrog(TM) software. The new workflow reduces the time to create a fault model of 113 faults in central Aotearoa New Zealand by ~80%, from 25 hours to 5 hours of human input. This improvement significantly decreases the effort required to create multiple alternative fault geometries, making detailed sensitivity analyses more feasible. The applicability of the workflow is demonstrated for the creation of three alternative models of fault geometries for central Aotearoa New Zealand.