Automatic InSAR Derived Earthquake Fault Planes as a Ground-Truth Seismic Dataset

Summary Accurate earthquake hypocentre determination is essential for a wide range of applications in seismology. Traditionally, researchers estimate hypocentres by inverting seismic body-wave arrival times. The accuracy of such hypocentres depends on the geographical distribution of stations, the accuracy of seismic travel-time models, and the precision of the arrival time picks. Past research has sought to validate location accuracy with a set of well-constrained ground-truth locations. We instead use surface deformation determined via Interferometric Synthetic Aperture Radar (InSAR), and develop an independent ground-truth set of 42 shallow, Mw 5.5–6.5 (USGS) events using an automated inversion scheme for InSAR-based source parameter estimation. Using this new independent ground-truth dataset, we investigate the relative effects of 3D velocity models, station distribution, and pick error on seismic event location. We demonstrate that InSAR source inversions derived via an automated Bayesian processing pipeline can serve as independent ground-truth for validating seismic event locations, with fully represented model error. This dataset can be used to test the accuracy of hypocentral locations derived using travel-time corrections from a 3D velocity model and show that, for our limited dataset, there is no improvement on average towards the InSAR-derived ground-truth location. We also show that InSAR-derived locations can be used to test pick accuracy, showing a great improvement towards the InSAR planes with manual picking and proper error handling. Through these demonstrative studies we show the usefulness of the new InSAR catalogue for use in studies requiring a ground-truth location.