Abstract
The 2016 7.8 Kaikōura earthquake is one of the most complex earthquakes in recent history, rupturing across many disparate faults with varying faulting styles. The complexity of this event has motivated the need for multidisciplinary geophysical studies to ascertain the underlying source physics in order to better inform earthquake hazards models in the future. However, events such as that in Kaikōura beg the question of how well (or how poorly) such earthquakes can be modeled automatically in real time and still satisfy the general public and emergency managers. To investigate this question, we perform a retrospective real‐time Global Navigation Satellite System (GNSS) analysis of the Kaikōura earthquake with the G‐FAST early‐warning module. First, we perform simple point‐source models of the earthquake using peak ground displacement scaling and a coseismic offset‐based centroid moment tensor (CMT) inversion. We predict ground motions based on these point sources, as well as simple finite faults determined from source‐scaling studies and validate against true recordings of peak ground acceleration. Second, we perform a slip inversion based upon the CMT fault orientations and forward‐model near‐field tsunami maximum expected wave heights to compare against available tide gauge records. We find remarkably good agreement between recorded and predicted ground motions when using a simple fault plane, with the majority of disagreement in ground motions being attributable to local site effects and directivity, not to earthquake source complexity. Similarly, the near‐field tsunami maximum amplitude predictions match tide gauge records well. We conclude that, even though our models for the Kaikōura earthquake are devoid of rich source complexities, the CMT‐driven finite fault is a good‐enough average source and provides useful constraints for rapid forecasting of ground motion and near‐field tsunami amplitudes.
