Identification and Estimation of Postseismic Deformation: Implications for Plate Motion Models, Models of the Earthquake Cycle, and Terrestrial Reference Frame Definition

Abstract:

Postseismic deformation indicates a viscoelastic response of the lithosphere. It is critical, then, to identify and estimate the extent of postseismic deformation in both space and time, not only for its inherent information on crustal rheology and earthquake physics, but also since it must considered for plate motion models that are derived geodetically from the “steady-state” interseismic velocities, models of the earthquake cycle that provide interseismic strain accumulation and earthquake probability forecasts, as well as terrestrial reference frame definition that is the basis for space geodetic positioning. As part of the Solid Earth Science ESDR System) SESES project under a NASA MEaSUREs grant, JPL and SIO estimate combined daily position time series for over 1800 GNSS stations, both globally and at plate boundaries, independently using the GIPSY and GAMIT software packages, but with a consistent set of a prior epoch-date coordinates and metadata. The longest time series began in 1992, and many of them contain postseismic signals. For example, about 90 of the global GNSS stations out of more than 400 that define the ITRF have experienced one or more major earthquakes and 36 have had multiple earthquakes; as expected, most plate boundary stations have as well. We quantify the spatial (distance from rupture) and temporal (decay time) extent of postseismic deformation. We examine parametric models (log, exponential) and a physical model (rate- and state-dependent friction) to fit the time series. Using a PCA analysis, we determine whether or not a particular earthquake can be uniformly fit by a single underlying postseismic process – otherwise we fit individual stations. Then we investigate whether the estimated time series velocities can be directly used as input to plate motion models, rather than arbitrarily removing the apparent postseismic portion of a time series and/or eliminating stations closest to earthquake epicenters.