Predicting Co-Seismic Deformation Following Intermediate and Deep Earthquakes: Toward Accurate GNSS Break Estimates and Improved Understanding of Deep Rupture Processes

Despite remarkable advances in understanding earthquake physics and seismic hazards through the use of GNSS data, our framework of understanding the Earth's surface displacements is largely limited to shallow (0-60 km) earthquakes. The amplitude of the surface deformation from deeper events was thought to be small; however, recent GNSS observations of intermediate and deep (I&D) earthquakes—e.g., 600-km deep 2013 MW8.3 Okhotsk event, 560-km deep 2018 MW8.2 Fiji event, and 110-km deep 2019 MW8.0 Peru event—showed considerable (>1 cm) co-seismic displacements, demonstrating that deep rupturing processes contribute significantly to the Earth's surface deformation. This means that failing to incorporate deep events can result in inaccuracies in GNSS break estimates, and hence, bias velocity estimates. The significant surface displacements also suggest GNSS observations can bring new insights into the deep rupture processes that seismic observations have not been able to provide.

To extend current geodetic framework to deeper earthquakes, in this study, we conduct systematic predictions of co-seismic deformation following I&D events. The predictions are made for a series of earthquakes with various depths, magnitudes, and focal mechanisms. In general, we find a relatively broad surface area experiences significant co-seismic deformation from I&D earthquakes. For example, an MW 8.3, 45-degree dip-slip event at 600-km depth can generate deformation larger than 2 mm as far as 1500 km away from the epicenter. Based on these predictions, we come up with (1) empirical equations that provide average deformation as a function of focal depth, magnitude, and epicentral distance and (2) threshold distances within which GNSS stations can record detectable co-seismic breaks.We also explore the influence of simplifying assumptions: flat Earth and homogeneous elastic structure. By comparing co-seismic displacements predicted based on a flat half-space Earth and a layered spherical Earth, we find the effect of these assumptions is insignificant in most cases. Furthermore, we discuss the sensitivity of co-seismic displacements to source properties (e.g., finite fault, fault plane orientations, non-double-couple components) and the potential of using GNSS data to better constrain deep rupture processes.