HR: 11:35h
AN: G41C-06    [PDF]
TI: Seasonal Vertical Signals in GPS geodesy: Consequences of Using a Linear Reference Frame Model
AU: * Johnson, D J
EM: dj@ess.washington.edu
AF: University of Washington, Department of Earth and Space Sciences, Seattle, WA 98195-1310 United States
AB: Geodetic positions derived from GPS observations are intrinsically relative. Generally, raw GPS solutions at a given epoch are adjusted to assign a meaningful reference to the final positions. The reference frame is, in practice, a subset of GPS sites contained in the solution along with a model that specifies secular motion of those sites. The simplest reference frame is one of a pair of stations that define a baseline. Position and motion at the other site, then, could be expressed relative to the base station which, in most cases, is assumed to be motionless, or "fixed". A more complex approach is to standardize a collection of sites. To first order, the rate of tectonic motion at many locations on the Earth's surface is constant. The real motion at a typical site is also likely to be gradual and progressive on a day-to-day timescale. Hence, a reasonable choice for a model that describes motion of the reference sites is a linear model. Adjustment of the raw GPS observations, then, may be realized by shifting the solution as a whole so that the subset of reference sites matches the positions predicted by the linear model. For extended and global networks, a seven-parameter (3 translations, 3 rotations, and a scale factor) transformation is generally utilized. Position timeseries derived from adjusted GPS observations have several characteristics that may be traced to the reference frame fitting procedure. First, alignment of daily geodetic solutions with a simple model of motion of the reference sites, such as a linear model, tacitly forces day-to-day positional changes at reference sites to be gradual and progressive, similar to the model that the solutions are being fit to. In most cases, this is desirable as much of the day-to-day fluctuation in apparent position of a GPS site is likely noise. Clearly, day-to-day fluctuations that correlate between all sites within a network is not real - this coherent jitter is removed through adjustment to the reference frame model, yielding a smoothed result. The second characteristic is that velocities calculated from the adjusted solutions are linked to the base motion imposed by the reference frame model. The final characteristic - which is not a benefit, but is a problem - is that alignment of GPS solutions with a linear reference frame model means that the resulting individual station timeseries will tend to be linear as well. This is because all common-mode differences between the raw solutions and the reference frame model are subtracted out. Seasonal patterns that deviate from linearity will be removed if are coherent over the entire network. Over 100 globally distributed GPS sites have been analyzed here using GIPSY/OASIS II in a fiducial-free reference frame. The origin of the raw station coordinates is inherently close to the center of mass of the Earth system, meaning that the vertical component of site positions is well determined in the raw solutions. Inspection of the raw coordinates reveals a network-wide seasonal pattern of motion aligned with the Earth's polar axis. This coherent motion is largely removed by applying reference frame constraints. Seasonal deformation cycles related to degree-one loading are poorly managed by typical linear reference frame constraints. Some degree-one vertical motion may be absorbed by the scale parameter in the transformation which, unfortunately, will then propagate the motion to other sites as noise. Due to imbalances in the distribution of GPS sites between hemispheres, this may have the effect of creating an apparent reduction in the amplitude of annual vertical motion in the northern hemisphere and enhanced amplitudes in the southern hemisphere.
DE: 1229 Reference systems
DE: 1243 Space geodetic surveys
DE: 1294 Instruments and techniques
SC: Geodesy [G]
MN: 2003 Fall Meeting