Comparing Calculations of Far-Field Coseismic Deformation

Tuesday, 16 December 2014
Jeffrey Todd Freymueller, University of Alaska Fairbanks, Fairbanks, AK, United States, Jie Dong, UCAS University of Chinese Academy of Sciences, Beijing, China and Wenke Sun, GUCAS Graduate University of Chinese Academy of Sciences, Beijing, China
Coseismic displacements from the largest earthquakes are easily detected in geodetic time series over a large area, up to a few thousand km from the fault. For example, detectable displacements from the 2004 Sumatra-Andaman earthquake were reported as far away as Korea, and detectable displacements from the 2011 Tohoku earthquake were reported as far away as Ulaanbaatar, Mongolia. The great reach of these events means that they can distort geodetic coordinates on a global scale, they impact the ability of users to access the reference frame. Point source (Centroid Moment Tensor) and finite fault models for large events are routinely available from purely seismological data (or potentially from local geodetic data), so it is worthwhile to know hwo well we can predict coseismic displacements at GPS sites from independent or quasi-independent data. Ultimately, a suite of geophysical models based on seismic or near-field geodetic data could provide estimates of far-field displacements and uncertainties that may be used for detection of subtle offsets in geodetic time series, and for correction of global coordinates.

Slip models based only on teleseismic data do not predict displacments well near the earthquake rupture, because teleseismic data do not resolve spatial details of the slip distribution. Farther from the source, coseismic displacements are not so sensitive to the details of the earthquake source model. Different codes for computing displacements for spherical layered Earth models have not been compared carefully before. Here we compare computed displacements using the method of Pollitz (1996) and the method of Sun et al. (2009) and Sun and Dong (2013). These codes are semi-analytical, so results are sensitive to the number of terms summed for computing Greens functions. We also compare the impact of different Earth models, including PREM, modified versions of PREM, ak135f, and a homogeneous sphere. Our preliminary assessment is that the choice of Earth model has a larger impact than the choice of software, as long as a sufficient number of terms are included for numerical accuracy.