Featherstone et al: GPS-geodetic deformation monitoring 
The 48-point GPS monitoring network will be used 
predominantly for monitoring horizontal deformation 
across the SWSZ, because GPS is inherently less precise 
in the vertical (shown later) due mainly to a combination 
of the geometry of the satellites (with them being only 
above the ground-based antennas) and inaccurately 
modelled atmospheric refraction of the GPS signals. 
Geodetic spirit levelling is the most precise means with 
which to monitor vertical deformation at discrete points. 
Therefore, if vertical deformation is sought over the 
SWSZ, it is probably best to reoccupy and extend the 
levelling network used by Wellman & Tracey (1988). 
GPS Data Processing and Results 
Any geodynamic deformation monitoring campaign 
that uses GPS (episodic or continuous) requires 
sophisticated data processing and error modelling. 
Firstly, geodetic quality (i.e. dual-frequency code and 
carrier-phase) instruments and antennas that reduce the 
effects of multipath must be used. Secondly, scientific 
GPS data processing software must be used. Commercial 
GPS data processing software is inadequate because of 
its approximated algorithms and inability to model the 
small, yet important, biases that would obscure the small, 
if any, surface deformations in the SWSZ. 
The National Mapping Division of Geoscience 
Australia processed the GPS data collected for epoch one 
(Dawson 2002) using the Bernese version 4.2 GPS 
software (Hugentobler et al. 2001), closely following the 
procedures adopted by the regional analysis centres of 
the International GPS Service (IGS; http:// 
igscb.jpl.nasa.gov/). This data processing included the 
rigorous modelling of systematic errors, including 
variable ionospheric and tropospheric signal refraction, 
periodic solid-Earth- and ocean-tide effects, and antenna 
phase-centre variations. The GPS satellite orbit 
information used was that provided in the final IGS 
orbital product, which was held fixed together with 
Earth-orientation parameters. Carrier-phase ambiguity 
resolution was attempted on all baselines for all GPS data 
collected above an elevation angle of 10 degrees. 
Besides the GPS data collected in the SWSZ, GPS data 
were also collated from continuously operating GPS 
receivers at Alice Springs, Karratha, Ceduna and Perth, 
whose 30-second (dual-frequency code and carrier-phase) 
data are made freely available through the IGS. These 
were included since their data are processed daily by the 
IGS analysis centres and hence have extremely accurately 
(sub-cm) known coordinates in a global, dynamic, 
reference frame. When the position of a point is 
expressed in a dynamic reference frame, it is 
accompanied by the particular epoch to which it refers. 
This recognises the fact that the coordinates of the point 
will change due to global plate tectonic motion and hence 
the point also has an associated velocity (curiously, no 
mention is made of intraplate deformations in the models 
used). The estimates of coordinates and velocities for 
Alice Springs, Karratha, Ceduna and Perth are computed 
by the International Earth Rotation Sendee (IERS) using 
all available IGS GPS data and are updated 
approximately yearly in a solution termed the 
International Terrestrial Reference Frame (ITRF). 
The latest realisation is ITRF2000 (http:// 
lareg.ensg.ign.fr/ITRF/ITRF2000), which used all GPS 
and other space-based data up until the end of the year 
2000 to compute the coordinates and velocity estimates 
for each point. Hence, including data from the IGS 
stations (with published ITRF2000 coordinates and 
associated velocities) in the GPS network solution has 
enabled the positions of the 48 SWSZ monitoring points 
to be computed relative to the four nearby IGS stations 
and to express their coordinates in the ITRF2000. The 
velocities of the 48 SWSZ monitoring points could then 
be subsequently computed from the coordinate estimates 
obtained at different epochs. For the epoch-one survey, 
the coordinates of the four IGS stations were fixed at 
their ITRF2000 values at the campaign mid-epoch 
2002.37, therefore the coordinates of the SWSZ 
monitoring points also refer to ITRF2000 epoch 2002.37 
While the GPS data were collected continuously at 
each monitoring point over the 5-7 day observation 
period (excepting the five stations that were occupied 
nearly continuously for 22 days), discrete three- 
dimensional position solutions were obtained using the 
data for each (Universal Time) day, which were then 
combined using the least squares techniques to obtain 
the final ITRF2000 epoch 2002.37 coordinate estimates 
(Dawson 2002). This approach enables some internally 
estimated quality assurance of the solution, via the 
computation of coordinate standard errors (la) from the 
variation of each daily solution from the mean. 
Importantly, such la are only indicators of the internal 
precision of the solution, namely that there is a 68% 
probability that the position solution obtained from any 
given day will be equal to the mean value in each 
dimension. When these three dimensions are combined 
to form a horizontal ellipse or three-dimensional 
ellipsoid, the error magnitudes must be scaled for 
multivariate statistics ( e.g . Anon. 1996). 
Fig 4 and Table 1 summarise the internally estimated 
precision of the ITRF2000 (epoch 2002.37) coordinates of 
the 48 SWSZ stations. The geodetic coordinates (i.e. 
latitude, longitude and ellipsoidal height) are given by 
Dawson (2002). From Fig 4, 46 of the 48 occupied stations 
have an internally estimated precision of 1.5 mm in the 
horizontal [43 have a precision of 1 mm]; the vertical 
precision was generally better than 7 mm. Two stations 
(SZ23 and SZ36) gave very poor results due to Leica 
CRS1000 equipment failure (Fig 4). Therefore, Table 1 
includes a summary of the results excluding these two 
outliers. 
The internal precision estimates summarised in Table 
1 and Fig 4 provide an indication of the consistency of 
Table 1 
Descriptive statistical summary of the internally estimated 
precision (la in mm) of the epoch-one GPS-derived coordinates 
(ITRF2000, epoch 2002.37) of the 48-point network across the 
SWSZ (values in parentheses exclude the two outliers at SZ23 
and SZ36) 
la East 
la North 
la Vertical 
Maximum 
3.4 (1.4) 
8.3 (1.5) 
21.8 (8.8) 
Minimum 
0.6 (0.6) 
0.7 (0.7) 
3.5 (3.5) 
Mean 
1.09 (0.99) 
1.36 (1.16) 
6.80 (6.18) 
STD 
0.50 (0.16) 
1.10(0.17) 
3.20 (1.06) 
5 
