5/18/2994G21D-04 Spring AGU Realization of a Stable North America Reference Frame Thomas Herring Department of Earth Atmospheric and Planetary, Sciences, MIT
5/18/2994G21D-04 Spring AGU Overview Construction of a stable North America Reference Frame: –Secular motion field (alignment to GIA model) –Temporal evolution (day-to-day realization of frame) Concentrate on comparing analysis with Bernese and GAMIT software with fitting to GIA Models
5/18/2994G21D-04 Spring AGU Acknowledgements Jim Davis, SAO Glacial Isostatic Adjustment (GIA) Models: 3 lithosphere thicknesses (120,96,71 km); 8 Upper Mantle viscosities (0.05-1x10 21 Pa-s; 9 Lower Mantle viscosities (1-50x10 21 Pa-s) Eric Calais, Purdue University, 29,000 processed networks of east-coast GPS data. SOPAC facility with ftp access to processed networks (up to 600 global and North America sites per day). ftp://garner,ucsd.edu/pub/hfiles CODE weekly sinex files submitted to IGS.
5/18/2994G21D-04 Spring AGU Secular Field Construction Necessary steps: –Account for offsets from antenna type and radome changes. –Establish sites that have “linear” motion (this step is relative easy with correlated noise models). –Determine a “stable” region of North America With stable region selected, GIA models can be evaluated This is the most complicated (and unresolved) step. Fit to GIA models depends on choice of stations (examples shown) –With stable North America defined, frame can be transferred to other linear motion sites (extrapolation issue: the stable region is small. Adding sites on other plates can make system more robust, but adds complication of motion of other regions).
5/18/2994G21D-04 Spring AGU GIA Model Evaluation Basic approach: –Select sites that have linear motion. We do this by fitting a correlated noise model and then selecting sites with small correlated noise (and long time series) –Select from these sites, those who motions are expected to be due GIA only (not clear at moment) –Determine the rotation/translation of the loosely combined GPS solution that best aligns it with the GPS model –Examine 2-D (horizontal) and 3-D fits to GIA –Evaluate with different GPS analyses: CODE Bernese solutions (sinex files) SOPAC analysis SOPAC analysis+Purdue analysis
5/18/2994G21D-04 Spring AGU Fit of COD to GIA: 9/1995-3/2004: 17 sites This minimum moves with increasing LT Lithosphere Thickness (LT) 71 km Details here depend geographic sites distribution
5/18/2994G21D-04 Spring AGU Fit of SIO to GIA: 1/1996-3/2004: 28 Sites Basic structure similar to CODE; similar spatial distribution of sites
5/18/2994G21D-04 Spring AGU Fit of PUR to GIA: 1/1994-4/2004: 12 sites selected to be GIA sensitive Sites selected in Canada and down the middle of the US Min UM 1, LM 2x10 21 Pa-s
5/18/2994G21D-04 Spring AGU Comparison of PUR solution (red, 50% confidence ellipses) with GIA model 71 km LT, UM 1, LM 2x10 21 Pa-s Fit: 26-sites N 0.6 mm/yr E 0.3 mm/yr U 1.9 mm/yr
5/18/2994G21D-04 Spring AGU Comparison of PUR solution (red, 50% confidence ellipses) with GIA model 120 km LT, UM 1, LM 2x10 21 Pa-s Fit: 26-sites N 0.5 mm/yr E 0.4 mm/yr U 1.8 mm/yr
5/18/2994G21D-04 Spring AGU Comparison between COD, SIO and PUR for Reference Sites Red: CODE RMS 3-D 0.7 mm/yr; 2-D 0.3 mm/yr Blue: SIO RMS 2/3-D 0.2 mm/yr Black: PUR
5/18/2994G21D-04 Spring AGU Motion estimates in “North America” for sigma less than 0.5 mm/yr and rates less than 2 mm/yr NOTE: Scale change
5/18/2994G21D-04 Spring AGU Time series of Site GDAC for North and Height
5/18/2994G21D-04 Spring AGU Conclusions In general none of the GIA models fit the GPS observations at the expected noise level: –Fits are 0.5 mm/yr horizontal, 1.5 mm/yr vertical –Agreement between the GPS results is better than with the GIA models Coherent non-GIA residuals seem to present in the results Caveat: With a decade of GPS data, satellites constellation has evolved and there are un-accounted satellite phase center variations that will effect the results