A ground motion model for GNSS peak ground displacement

Abstract

We present an updated ground‐motion model (GMM) for M w "> M w Mw 6–9 earthquakes using Global Navigation Satellite Systems (GNSS) observations of the peak ground displacement (PGD). Earthquake GMMs inform a range of Earth science and engineering applications, including source characterization, seismic hazard evaluations, loss estimates, and seismic design standards. A typical GMM is characterized by simplified metrics describing the earthquake source (magnitude), observation distance, and site terms. Most often, GMMs are derived from broadband seismometer and accelerometer observations, yet during strong shaking, these traditional seismic instruments are affected by baseline offsets, leading to inaccurate recordings of low‐frequency ground motions such as displacement. The incorporation of geodetic data sources, particularly for characterizing the unsaturated ground displacement of large‐magnitude events, has proven valuable as a complement to traditional seismic approaches and led to the development of an initial point‐source GMM based on PGD estimated from high‐rate GNSS data. Here, we improve the existing GMM to more effectively account for fault finiteness, slip heterogeneity, and observation distance. We evaluate the limitations of the currently available GNSS earthquake data set to calibrate the GMM. In particular, the observed earthquake data set is lacking in observations within 100 km of large‐magnitude events ( ⁠ M w > 8 "> M w > 8 Mw>8 ⁠ ), inhibiting evaluation of fault dimensions for earthquakes too large to be represented as point sources in the near field. To that end, we separately consider previously validated synthetic GNSS waveforms within 10–1000 km of M w "> M w Mw 7.8–9.3 Cascadia subduction zone scenario ruptures. The synthetic data highlight the importance of fault distance rather than point‐source metrics and improve our preparedness for large‐magnitude earthquakes with spatiotemporal qualities unlike those in our existing data set.