Bayesian Dynamic Finite-Fault Inversion of the 2016 Mw 6.2 Amatrice, Italy, Earthquake

Date: 4/25/2019

Time: 08:30 AM

Room: Cascade I

In 2016, Central Italy was struck by three normal faulting earthquakes with Mw>6. The first Mw 6.2 Amatrice event (08/24), caused building collapse and about 300 casualties. The event was recorded by a uniquely dense network of seismic stations. Kinematic source inversions based on such data can be used to characterize the spatio-temporal evolution of earthquake slip, but provide only indirect constraints on rupture physics. Therefore, here we perform dynamic source inversion to directly infer the fault friction parameters and stress conditions that controlled the earthquake rupture. We consider dynamic rupture governed by a linear slip-weakening friction law with spatially variable parameters along the fault. The inversion approach utilizes a novel Bayesian framework, which combines efficient finite-difference dynamic rupture simulations by the FD3D code and a Parallel Tempering Monte Carlo algorithm to sample the posterior probability density function. The main advantage is that subsequent analysis of the posterior samples yields stable features of the results and associated uncertainties. The inversion results in a million of visited models and ~5,000 accepted model samples revealing intriguing dynamic features. In agreement with previous kinematic inversions, the rupture initiated by localized transient nucleation followed by bilateral rupture propagation across two asperities. The rupture accelerates towards the heavily damaged city of Amatrice where peak acceleration of 0.8 g was measured. Dynamic stress drop reaches locally 10-15 MPa, with a mean of 4-4.5 MPa. Friction drop ranges from 0.1 to 0.4. The critical slip-weakening distance spatially varies between 0.2 and 0.8 m. The radiation efficiency is between 0.1-0.2, indicating that approximately 80-90% of the total available energy is spent in the fracture process, while only 10-20% is radiated by seismic waves. This study demonstrates how Bayesian exploration of the parameter space and abundant strong motion data can provide constraints on earthquake source physics of well recorded earthquakes.

Presenting Author: Frantisek Gallovic

Authors