GPS data processing

The result shown here are derived from the network of continuous GPS sites covering the New Madrid area (GAMA) installed by CERI and from additional continuous GPS sites of the CORS and IGS networks.

The picture on the left (courtesy G. Mattioli, University of Arkansas) shows two GPS antennas at site HCES in the New Madrid seismc zone. Note the different type of monumentation.

The GAMIT-GLOBK analysis uses double-differenced GPS phase and solves for station coordinates, satellite state vectors, 7 tropospheric delay parameters per site and day, horizontal tropospheric gradients, and phase ambiguities using IGS final orbits and earth orientation parameters. We apply elevation dependant antenna phase center models following the tables recommended by the IGS, solid Earth, polar tide, and ocean loading corrections following the IERS standards (IERS, 1996). We process 15 subnetworks of about 25 stations each, totalling about 400 sites covering the stable interior of the North American plate. 10 IGS sites are common to all subnetworks to tie them together and with a global reference frame, if needed. We then calculate position time series for each site, from which we estimate process noise parameters. The daily solutions for each subnetwork are then combined into a single loosely-constrained position-velocity solution covering 1996-2005. We finally implement a North America-fixed reference frame by minimizing the velocities of 23 stations in the North American interior while estimating an orientation and translation transformation. These stations are chosen to (1) have a random-walk noise component lower than 0.6 mm/sqrt(year), (2) have at least 5 years of continuous operation, and (3) be located south of 40N in order to avoid glacial isostatic adjustment effects and east of 110W in order to avoid tectonic deformation in the Western U.S.

The GIPSY-OASIS-II solution uses final precise orbits, clocks, and earth orientation parameters (all from JPL's global analysis) with a free-network, absolute point-positioning strategy. Ocean-loading coefficients for each site were obtained using the Scherneck model of the Onsala Space Observatory. Site coordinates determined from the analysis of the GPS phase pseudo-range observables are initially defined in a fiducial-free reference frame and subsequently transformed into the ITRF2000 using 7-parameter Helmert transformations that align the daily coordinates of a selected subset of ITRF2000 fiducial sites with the fiducial-free coordinates of the same subset of sites. The secular motion of the North American plate was removed based on current realization of the DeMets et al. (2000) NOAM-ITRF00 euler pole. Common mode errors at each site were removed (up to Sep. 2004) using the NLIB cGPS site as a regional reference with a method modified after Wdowinski et al., 1997. Velocities are calculated by a linear fit to position time series using a linear least-squares inversion with formal estimates of white and flicker noise and an assumed random walk monument noise of 1 mm/sqrt(yr) as discussed by Mao et al. (1997).