Data from 18 GPS stations located in Washington State as well as from 49 sites in eight nearby states and two Canadian Provinces were analysed to quantify co-seismic displacements. Initial positions in the ITRF97 coordinate reference frame were determined using GIPSY/OASIS II software (Heflin, et al., 1992; Zumberge, et al., 1997) in point-positioning mode, which utilizes precise orbits and clock solutions produced by the Jet Propulsion Laboratory. Ambiguity resolution was applied.
A subset of data bracketing the earthquake from the 12 days prior to 12 days after was chosen for further analysis. Longer time intervals give a result similar to the one presented here, and have the disadvantages that they could be biased by error sources that effect long-period data (Mao, et al., 1999) and would require a correction for secular motion. Shorter time intervals have greater position uncertainties, which is evidenced by increased scatter of calculated displacements.
Individual daily station positions were combined into daily network solutions. These solutions were, in turn, combined into a pre and post- earthquake solution by simultaneously solving for transformation parameters (3 translations, 3 rotations, and scale) that adjust each daily network solution to a common best-fit average, and the best-fit average. This step minimizes network-wide position discrepancies, or common-mode errors.
Finally, the post-earthquake network solution was aligned with the pre-earthquake solution with a 7-parameter transformation using all of the GPS sites (with the exception of seven stations located near the earthquake epicenter) and differenced, giving the net displacements.
Formal error estimates for point-positioned GPS solutions are less than actual error, in part because orbital and clock errors are not explicitly accounted for in the solution. Here, formal errors have been scaled by a factor of 1.16 such that the daily solution uncertainties are in agreement with the actual position scatter. Error estimates reported here are 2 sigma, and for displacements between pre- and post- earthquake surveys average 1.36 mm in the north-south component, 1.20 mm in the east-west component, and 4.2 mm in the vertical component. Actual scatter of the displacement vectors in areas outside the region affected by the earthquake are consistent with the above error estimates.
Heflin, M., W. Bertiger, G. Blewitt, A. Freedman, K. Hurst, S. Lichten, U. Lindqwister, Y. Vigue, F. Webb, T. Yunck, and J. Zumberge, Global Geodesy Using GPS Without Fiducial Sites, Geophys. Res. Lett., Vol. 19, 131-134, 1992.