How quantitative data combination in earthquake source analyses is paying off

Abstract:

Large and shallow crustal earthquakes reveal characteristics of tectonic faults, often by causing considerable damage. Many of them can be studied by using geodetic data, e.\,g. InSAR data, as well as local to global seismological data. InSAR data provide co-seismic static surface displacements at high spatial resolution that are proportional to the moment and characteristic for the source location, geometry and dimension. Far-field seismic waves carry information about the time-dependent moment release during the rupture. Both datasets are sensitive to the source mechanism. In combination, the weaknesses of each individual data set can potentially be compensated, while information is accumulated.

We have developed methods that enable a strictly quantitative combination of near-field InSAR data and seismic waveforms for kinematic earthquake source modeling. We represent the earthquake with a finite rupture plane of rectangular shape and uniform slip, or a number of planes to allow for segmented faulting if required by the data. Within wide parameter ranges, we estimate the location, dimension and orientation of the rupture plane and the amount and direction of slip. Furthermore, we estimate the nucleation point and the average rupture velocity. The incorporation of Bayesian Bootstrap allows us to rigorously propagate data errors into ensembles of models, representing the model uncertainty.

We apply our methods in a standardized and semi-automated way to about a dozen instrumentally recorded crustal earthquakes with $6<M_{\mathrm{W}}<7$ from different tectonic regimes around the globe.We discuss to which extent our use of open global data helps to robustly constrain the faulting and the hazard-relevant rupture characteristics. We compare our results with independent teleseismic multi-array backprojection.

We investigate the data combination benefits for future robust source catalogs that provide a high degree of source complexity. These simple source models are highly robust and appear to well reproduce the seismic energy excitation up to 1.5\,Hz in synthetic-data backprojection. Potentially source ensembles can be used for informed rupture and ground motion simulations.

The presented methods are published open access and open source as part of the Pyrocko software project (pyrocko.org).