Pairing geodesy and earthquake catalogs to get at seismic hazard

Chris Rollins

Published August 3, 2020, SCEC Contribution #10248, 2020 SCEC Annual Meeting Talk on TBD

Geodesy can provide crucial constraints on a region’s seismic hazard, particularly when paired with earthquake catalogs and other data. For example, geodetic data might show that tectonic strain is accumulating much more quickly than known earthquakes have released it, implying that the seismic hazard may include 1) larger, less frequent events and/or 2) periods of increased seismicity. Conversely, geodetic data might show that strain is accruing more slowly than the rate at which recent earthquakes have collectively released it, implying that they have constituted an “overshoot” and that long-term-average seismicity may be comparatively quiescent. Such constraints, however, have traditionally been hampered by the limited spatial sampling of geodetic data, as this results in ambiguities about where and how quickly strain is accruing and therefore where and how often earthquakes might occur. High-resolution observations of surface motion help address this limitation and are now within reach thanks to recent advances in SAR coverage and processing. These include the COMET-LiCSAR pipeline, which performs large-scale automated processing and timeseries analysis of Sentinel-I InSAR data.

A recent study [Weiss et al., 2020] combined LiCSAR products with GNSS data to generate high-resolution maps of surface strain rates across Anatolia. Here we assess what these strain rate maps imply for seismic hazard. Building on previous work [Rollins and Avouac, 2019], we develop a fully probability-based method to pair geodesy and seismic catalogs to estimate the recurrence times of large, moderate and small earthquakes in a region. We assume that earthquakes 1) obey a power-law magnitude-frequency distribution up to a maximum magnitude and 2) collectively release seismic moment at the total rate that we estimate it is accumulating from the strain maps. Iterating over candidate magnitude-frequency distributions and their governing parameters, and formally incorporating uncertainties in moment buildup rate, the magnitudes of recorded earthquakes and the magnitude of completeness, we build a probabilistic long-term-average earthquake likelihood model for Anatolia. We also use arguments from dislocation modeling to identify key signatures of a locked fault in a strain rate field, providing a potential way to convert strain rate maps to “effective locked fault maps” and distinguish individual faults’ contributions to deformation and hazard.

Key Words
strain, InSAR, GPS, hazard, probabilistic

Citation
Rollins, C. (2020, 08). Pairing geodesy and earthquake catalogs to get at seismic hazard. Oral Presentation at 2020 SCEC Annual Meeting.

Related Projects & Working Groups
Tectonic Geodesy