[IGSMAIL-3698]: GPS Time Series Update

[Michael B Heflin]

******************************************************************************
IGS Electronic Mail      15 Jan 09:52:18 PST 2002      Message Number 3698
******************************************************************************

Author:  M B Heflin

Ambiguity resolved, regionally filtered SCIGN time series are available at URL:

http://sideshow.jpl.nasa.gov/mbh/series.html

Recent improvements include:
------------------------------------------------------------------------------
1.  Ambiguity resolved free-network stacov files provided by Ken Hurst
2.  All results in ITRF2000
3.  Daily three parameter regional filter
4.  Time span 5.8 years:  96JAN01 through 01NOV03
5.  Time series for 265 SCIGN sites
6.  Numerical results available via anonymous ftp
7.  Point positions and filtered results plotted side by side for comparison

How well do the ambiguity resolved, regionally filtered velocity estimates
agree with those from point positioning?
------------------------------------------------------------------------------
WRMS velocity agreement:  0.6 N, 0.9 E, 1.5 V mm/yr

Regional filtering can be written as:

site.lat.filtered = site.lat.unfilitered - common.lat
site.lon.filtered = site.lon.unfilitered - common.lon
site.rad.filtered = site.rad.unfilitered - common.rad

where common.lat, lon, rad are the mean residual latitude, longitude, and
height over regional sites at each time.  The common.lat, lon, rad series are
dependent on the size of the region in question and on the model used to
compute residuals.  In general, the amount of common noise gets smaller as the
region being filtered gets bigger.

Velocities can then be written as:

v.site.lat.filtered = v.site.lat.unfilitered - v.common.lat
v.site.lon.filtered = v.site.lon.unfilitered - v.common.lon
v.site.rad.filtered = v.site.rad.unfilitered - v.common.rad

If the time spans are identical v.common.lat, lon, rad are the same for all
sites so relative velocities are completely unaffected by regional filtering.
In practice v.common.lat, lon, rad are small causing even the absolute filtered and unfiltered velocity estimates to be very similar.

What is the daily repeatability of filtered and unfiltered results?
------------------------------------------------------------------------------
Here is an approximate breakdown on noise characteristics:
Filtered Site Noise = Sf
Common Site Noise = Sc
Point Site Noise = Sp = sqrt( Sf*Sf + Sc*Sc) 
Baseline Noise = Sb = sqrt( 2Sf*Sf + 2Sc*Sc - 2Sc*Sc) = sqrt(2) * Sf

     With Ambiguity Resolution             Without Ambiguity Resolution
          N       E       V                     N       E       V
         mm      mm      mm                    mm      mm      mm
Sf      1.1     1.0     4.4                   1.1     2.1     4.8
Sc      3.7     5.2     9.1                   3.7     5.2     9.1   
Sp      3.9     5.3    10.1                   3.9     5.6    10.3
Sb      1.6     1.4     6.2                   1.6     3.0     6.8