HR: 10:20h
AN: G41C-01 INVITED [PDF]
TI: Error Analysis of Continuous GPS Height Time Series
AU: * Williams, S D
EM: sdwil@pol.ac.uk
AF: Proudman Oceanographic Laboratory, Joseph Proudman Building
6 Brownlow Street, Liverpool, L3 5DA
United Kingdom
AU: Bock, Y
EM: ybock@ucsd.edu
AF: Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics,, Scripps Institution of
Oceanography, La Jolla, CA 92093 0225 United States
AU: Fang, P
EM: pfang@ucsd.edu
AF: Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics,, Scripps Institution of
Oceanography, La Jolla, CA 92093 0225 United States
AU: Jamason, P
EM: pjamason@ucsd.edu
AF: Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics,, Scripps Institution of
Oceanography, La Jolla, CA 92093 0225 United States
AU: Nikolaidis, R M
EM: niko@ucsd.edu
AF: Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics,, Scripps Institution of
Oceanography, La Jolla, CA 92093 0225 United States
AU: Prawirodirdjo, L
EM: linette@ucsd.edu
AF: Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics,, Scripps Institution of
Oceanography, La Jolla, CA 92093 0225 United States
AB:
Until recently, it was typically assumed that geodetic time series,
in the absence of earthquakes, consisted of a linear trend
representing crustal motion, and measurement errors which
were assumed to be normal (Gaussian) and statistically uncorrelated
from one another (white noise). However, many geodetic data sets
have now provided evidence for signals and noise that introduce large
temporal correlations into the data.
One common statistical model for many types of geophysical signal
may be described as a power-law process, or one with time-domain
behavior that has a power-spectrum of the form $P \propto f^{\alpha}$.
The spectral index, $\alpha$, is typically in the range$-2 < \alpha < 0$.
More than 900 continuous GPS height times series from over 400 individual
sites in 9 different GPS solutions were analyzed for power-law noise
content using Maximum Likelihood Estimation (MLE). The lengths of the
series varied from 500 to 3722 daily position estimates (around 16 months
to over 10 years). In solutions where the sites were globally distributed
the noise can be best described by a combination of white noise plus
flicker noise. Both noise components show latitude dependence in their
amplitudes (higher at equatorial sites) together with a bias to lower
values in the northern Hemisphere. In the regional solutions, where an
attempt has been made to remove a spatially correlated (common-mode)
signal, the noise is significantly lower. The spectral index of the power
law in regional solutions is more varied than in the global solutions and
probably reflects a mixture of origins such as monument instability,
hydrological effects, atmospheric effects, and residual common-mode noise.
A significant reduction in noise can be seen for all solutions since the
first continuous GPS networks began recording in the early 1990s.
Until such time that all unwanted signals in the GPS height time series
can be removed, a suitable stochastic model can serve as a useful description
of the time series. The stochastic noise approach allows us to account for
the effect such noise has on the estimation of parameters (such as rate) and
their uncertainties. Furthermore, stochastic models can provide help in the
search for physical explanations of the observed signal. We examine some
models of expected signals to examine their suitability as candidates for
the noise in a stochastic sense. At the present it appears that certain models
characterise the height time series at specific frequencies, that is, annual
and its harmonics.
DE: 1206 Crustal movements--interplate (8155)
DE: 1208 Crustal movements--intraplate (8110)
DE: 1244 Standards and absolute measurements
DE: 1294 Instruments and techniques
DE: 1299 General or miscellaneous
SC: Geodesy [G]
MN: 2003 Fall Meeting