GPS Imaging of Solid Earth's Flex and Flow: A New Paradigm

Abstract:

Geodetic GPS data analysis has gone through several paradigm shifts since the 1980s. Initially GPS was used in relative positioning mode to leverage and densify the existing global VLBI network. In the 1990s the new paradigm was to analyze GPS as a self-contained system, in which the global network of GPS stations and satellite orbits could be estimated simultaneously. Computational resources limit this approach to a few hundred stations (n ~ 100), with O(n⁴) computational complexity. Since the last decade, the new paradigm is to estimate GPS orbits first, followed by precise point positioning of single stations with O(n) complexity. This allows for parallel processing of an unlimited number of stations. The Nevada Geodetic Laboratory currently updates GPS time series for over 13,500 stations every week, a number that has been doubling every ~3 years. In some parts of the world, the inter-station distance between GPS stations that we process is now approaching ~10 km. This now brings us to a new paradigm, “GPS Imaging,” for which we use thousands of GPS stations in different continents to generate smooth, yet detailed maps of vertical land motion. Our prototype images show that the striking, first-order signal in North America and Europe is large scale uplift and subsidence from mantle flow driven by Glacial Isostatic Adjustment. Thus we are imaging deep Earth processes with unprecedented scope, resolution and accuracy. At regional scales, the images reveal that anthropogenic lithospheric processes can dominate vertical land motion in extended regions. We have developed prototype techniques that form a foundation to make “GPS Imaging” operational: (1) an automatic, robust estimator of station velocity that is insensitive to prevalent step discontinuities, outliers, seasonality, and heteroscedasticity; (2) a realistic estimate of the velocity errors based on subsampling; (3) a filter of common-mode noise that is globally seamless; (4) a median spatial filter to despeckle the maps; (5) a velocity time series estimator to identify transient behavior; and (6) an integration of InSAR and GPS for fine scale maps of land motion. We predict that “GPS Imaging” is poised to assist the discovery of new natural and anthropogenic signals, and to enable more data-driven approaches to understand Earth processes.