Journal of the Royal Society of Western Australia, 87(1), March 2004 
Figure 4. Internally estimated precision (la) of the epoch-one GPS-derived coordinates (ITRF2000 epoch 2002.37) of the 48-point 
network across the SWSZ (units in mm) [SZ11 and SZ47 were not placed] 
the solution and suggest that the GPS data were 
generally of high quality and there were no major signal 
interference or reception problems. In addition, 
systematic error effects that vary may over the duration 
of the epoch-one campaign, such as atmospheric delays, 
were well mitigated by sophisticated modelling during 
the data processing. Nevertheless, the figures above are 
by no means an indication of coordinate accuracy. Inter¬ 
instrumental and site-dependent biases, notably un¬ 
modelled multipath, (i.e. systematic errors) will result in 
different coordinates being computed for the same 
station when surveyed with different GPS antennas and 
receivers. 
For instance, the electrical phase centres of the antenna 
at which the GPS signals are received are not necessarily 
coincident with the physical/geometrical centre of the 
antenna due to mechanical imperfections, and also vary 
depending on the antenna make and model. Therefore, 
future surveys should endeavour to use the same 
receivers and antennas at the same ground marks. If not, 
great care will be exercised to account for the above- 
mentioned systematic errors so that any coordinate 
differences are not misinterpreted as surface motion in 
the SWSZ (cf. Lambeck & Coleman 1983; Featherstone 
1998). 
Discussion and Future Work 
Monitoring results from previous studies of geodetic 
surface deformation of the SWSZ have proved 
inconclusive due to errors in the data collection and 
reduction techniques used. Therefore, a 48-point network 
of dedicated permanent ground monuments, with 
forced-centring apparatus, has been established to 
quantify any surface deformation associated with the 
seismic activity across the central part of the SWSZ with 
a view to long-term GPS-geodetic monitoring of their 
deformation. The consortium first occupied the epoch- 
one network during May 2002 using four different types 
of GPS receivers and antennas. Excluding two outlying 
stations (based on statistical analysis and problems 
encountered in the field), the internal precision estimates 
of the computed ITRF2000 (epoch 2002.37) horizontal 
coordinates show a mean precision of ~1 mm in north 
and east (and ~6 mm in the vertical). The vertical GPS 
coordinates will probably not be used because levelling 
is a more precise technique for detecting vertical motion. 
Based on the accuracy with which the horizontal 
coordinates of the monitoring points can be estimated 
and the expected episodic reoccupation frequency of the 
monitoring epochs, indications of the time required to 
detect a particular magnitude of surface deformation can 
be estimated. Using the relationship provided by Coates 
et al. (1985), Dixon (1991) states that the uncertainty o v in 
any station velocity estimate is dependent on the single 
measurement point accuracy (not precision) c m , the 
(assumed constant) epoch separation A t, and the time T 
between the first and last epoch: 
O’,. = 
cj m 1277 Af 
T (l + 77Af)(2 + 77Af) 
For a point measurement accuracy of 1.5 mm, an 
epoch separation (i.e. episodic re-measurement period) of 
two years and a total monitoring period of 10 years, this 
suggests a point velocity uncertainty of 0.2 mmy 1 . 
Extending the monitoring period to 20 years suggests a 
point velocity uncertainty of 0.1 mmy 1 . However, 
regarding the epoch-one results, it is stressed that, whilst 
the maximum horizontal internal precision estimate for 
the SWSZ monitoring points was ~1.5 mm (ignoring 
outliers), this is not an indicator of the accuracy of the 
point measurement and cannot necessarily be blindly 
used to obtain an estimate of the uncertainty in the 
computed velocity. Accordingly, the above velocity 
uncertainties are probably over-optimistic. 
Furthermore, the equation for a assumes negligible 
slowly varying systematic errors, yet coordinate time- 
series computed from the analysis of continuous GPS 
data by many analysts worldwide currently show 
pronounced seasonal variations (e.g. Dong et al. 2002). 
Blewitt et al. (2001) identified the response of the elastic 
solid-Earth to redistribution of surface load via seasonal 
changes in a GPS coordinate time-series. When surface 
6 
