Bayesian Inference of Seismogenic Stress for the 2016 Mw 7.8 Kaikōura, New Zealand, Earthquake

Abstract

ABSTRACT We use a Bayesian Markov chain Monte Carlo estimation strategy to estimate the state of pre-earthquake stress from eight published coseismic slip models (CSMs) of the 2016 Mw 7.8 Kaikōura, New Zealand, earthquake. Our estimation relies on the Wallace–Bott assumption, which equates the direction of slip and the orientation of maximum shear stress along a fault plane, and is analogous to methods that infer stress from focal mechanisms. We infer the orientations and relative magnitudes of the principal components of tensorial stress preceding the 2016 Mw 7.8 Kaikōura, New Zealand, earthquake as stress posterior probability density functions (PDFs). We find that the orientations of the principal stresses are well resolved, with weaker to no resolution of the relative magnitudes of the principal stresses. We form a nonmutually exclusive composite posterior PDF as an aggregate of the individual posterior PDFs estimated from each CSM, which allows for the true coseismic slip to be included in any or all of the CSMs. We find that the Kaikōura earthquake can be described by a homogeneous pre-earthquake tensorial stress, despite the complex pattern of slip on multiple fault segments. This state of crustal stress is best represented by an Andersonian thrust regime with a west-northwest–east-southeast trending, horizontal most compressive stress. In addition to describing the state of pre-earthquake stress, the nonmutually exclusive composite posterior PDF allows for the evaluation of the mechanical consistency of the features within CSMs, both individually and as an ensemble. The estimated stress posteriors allow for testing of the CSMs for consistency with known slip mechanisms on the Hope fault, as well as arbitration between differences in inferred slip presented by the CSMs.