Ground GNSS Atmospheric Sensing - GNSS Remote Sensing: Theory, Methods and Applications - page 54

50

40

30

20

10

0

1997 1999 2001 2003 2005 2007

Time (year)

Fig. 3.19 PWV time series at ALGO, Canada

be derived from GPS-observed ZTD, surface synoptic observation-estimated P s
and NCEP/NCAR reanalysis-estimated T m with errors of about 1.0-1.5 mm due
to errors in ZTD, Ps and Tm. Also our GPS-derived PWV data are consistent with
PWV estimates from IGS-provided combined ZTD (1997-2007) with a RMS of
less 0.2 mm. For example, Fig. 3.19 shows the PWV time series at ALGO station
(Canada).

3.4.2 Comparison with Independent Observations

The IGS GPS-derived PWV data are compared with another co-located inde-
pendent technique-Very Long Baseline Interferometry (VLBI). Here VLBI-PWV
was derived from co-located meteorological observations data. Snajdrova et al.
( 2005 ) analyzed 15 continuous days of VLBI data during the Continuous VLBI
2002 (CONT02) campaign and found that the ZTD from VLBI and GPS were in
good agreement at the 3-10 mm level as well as with the Doppler Orbitography
Radio positioning Integrated by Satellite (DORIS). Meanwhile, VLBI and GPS
observed ZTDs are also quite good agreements with those from the European Centre
for Medium-Range Weather Forecasts (ECMWF) and WVR. Niell et al. (2001)
compared the results of a 2-week VLBI campaign in August 1995 (CONT95) at
Westford (USA) with GPS, Water Vapor Radiometry (WVR) and radiosondes and
found the VLBI technique was the most accurate for the determination of ZTD.
These show that the VLBI can obtain a high accuracy reliable ZTD estimation. The
Analysis Centers (Acs) of the International VLBI Service (IVS) process all available
VLBI observation data and corresponding products (e.g. ZTD) are transferred to the
IGG AC (Institute of Geodesy and Geophysics, Vienna University of Technology,
Austria), for combination. The combined ZTD time series are available from