What fraction of common-mode errors which have been mitigated in the NA12 datum are artifacts of GPS data processing? Abstract


  • GPS has become the preeminent tool in studying current crustal motion. The sub-millimetric precision of GPS time series requires some sophistication in coordinate frame design and definition. When GPS time series are used in regional studies it is often useful to develop regional reference frames. The NA12 reference framed was developed for studies in North America to provide a context which makes intraregional motions easier to interpret. Furthermore, the use of regional frames can significantly reduce the daily and higher-frequency scatter of these series about their longer-term trends. Both the horizontal and vertical time series from solutions produced in NA12 at the Nevada Geodetic Laboratory have about half the scatter of the corresponding series expressed in the global reference frame IGS08. This reduction in scatter is due to the reduction of common-mode error, i.e. unmodeled errors that are highly correlated among a group of stations. These errors can have many different physical causes such as regional-scale tropospheric delays, but we cannot discount the influence of the particular GPS processing methodology used. Two software packages are in common use to produce high-precision GPS time series. The Nevada Geodetic Laboratory use GIPSY, which implements a precise point positioning. On the other hand, the University of Arizona Tectonic Geodesy Laboratory uses GAMIT/GLOBK, which uses double differences in a relative positioning model. Relative positioning is fundamentally different than precise point positioning in that some physical common-mode errors are automatically eliminated but, again, any method can introduce its own artifacts. At NGL we transform directly from GIPSY solutions based on JPL fiducial-free orbits directly into NA12 using the full set of 299 reference stations that are part of the datum definition. By contrast, the Tectonic Geodesy Laboratory in Arizona computes the initial GAMIT/GLOBK solution and then transforms these series into NA12 using a subset (usually about half) of these stations. Although in past studies computed station velocities obtained using GIPSY are identical at the level of statistical significance, the daily solutions may differ. The intent of this study is to gain some insight into the nature of these differences. Because there are a large number of paths to obtaining solutions nominally in a common datum using GIPSY and GAMIT/GLOBK, we must compare daily solutions at several stations using several processing combinations. By analyzing how the daily solutions vary when the solution methodology is changed we infer the proportions of common-mode errors which is an artifact.

publication date

  • 2014

presented at event