International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
make an instantaneous judgement on the quality of GPS data 
during measurement, on which basis adequate decision can be 
taken in the field. With the current trend of customisable 
mobile GIS applications, however, it has become possible to 
access the various GPS protocols, such as NMEA, which 
allows users to program their own accuracy assessment and 
figure of merit reports in a customized mobile GIS applications. 
Accuracy experiments to assess the accuracy can then be 
normalized for HDOP statistics and averaging time to provide a 
basis for an error prediction that is better controlled by the user. 
The following provide some suggestions on how to further 
enhance the FOM reports currently employed by non-expensive 
handheld units: 
1. Directional dependencies of the error can currently not be 
assessed. This could be important in applications where 
existing waypoints need to be relocated in the field or where the 
mapping of directional features, such as unit boundaries need to 
be validated. Information on the anisotropy in the distribution 
of measured GPS positions, allows the user to assess directional 
variation of the error. 
2. In the current handheld system the user has only access to 
one percentile in the probability distribution of the error 
estimate (usually the CEP 50 or RMS 68 percentiles). In most 
cases the field surveyor is not that interested in such optimistic 
error estimates where there is still a probability of 50 to 47 
percent that a particular measurement falls outside the specified 
range. More conservative error estimates based for example on 
the 95 or 99 percentiles, are probably more useful to the 
average user, because such estimates minimizes the risk for the 
GPS measurement to be outside the reported error estimate. 
Ideally the user should have access to the modelled error 
distribution during measurement. 
3. The influence of outliers on the error estimate cannot be 
assessed or outliers above a user-defined threshold cannot be 
rejected on the basis of their associated poor DOP value. 
Functions to reject outliers would greatly improve the accuracy 
of time-averaged GPS data. 
7. CONCLUSIONS 
In this paper we discussed methods for the assessment and 
reporting the error of handheld code receivers. The figure of 
merit (FOM) and estimated position error reports (EPE) of the 
code receivers are often misleading, because the effects of bias 
can not be considered. Non-expert users may not bet aware of 
this limitation and may rely on overoptimistic error estimates. 
The following suggestions are meant to overcome these 
limitations: 
I. Non-expert users should be guided by GPS handheld vendors 
in conducting simple experiments in their study area, to 
determine the accuracy of their instrument when reference 
points are available. 
632 
2.The effects of variations of the constellation during the day, 
the masking effects and the effect of averaging time on the 
accuracy assessment should not be underestimated in such 
experiments. 
3.Our experiment suggests that the practical confidence is not 
always in agreement with the theoretical (estimated) 
confidence. For the 99% confidence level, for example, the 
percentage of outliers are higher then estimated, probably as a 
result of the relatively small sample size. 
4.The RMS of the averaged planimetric position, together with 
HDOP at the moment of the observation seems to be a realistic 
FOM for handheld receivers. 
5.More FOM reporting functions, such as scatter plots, time- 
average graphs, DOP as a function of time and probability 
graphs are needed for the demanding users of code receivers. 
REFERENCES 
|- Wilson, D., GPS Horizontal position accuracy in modelling 
of GPS position errors 
2- Guide to GPS positioning, David Wells, ISBN 0920114733, 
University of New Brunswick, Canada 
3- Vanicek, P. and E.J. Krakiwsky, Geodesy, the concepts 2nd 
rev. ed. North Holland, Amsterdam, the Netherlands, 1986 
4- Baarda, W. (1967): Statistical concepts in geodesy. 
Netherlands Geodetic Commission, Publications on Geodesy, 
New Series, vol. 2, no. 4. 
5- de Jong, C.D. (1998): A unified approach to real-time 
integrity monitoring of single- and dual-frequency GPS and 
Glonass observations. Acta Geodetica et Geophysica 
Hungarica, 33, 247-257. 
6- GPS Theory and Practice, B. Hofmann, Wellenhof, 326 pp, 
Springer, New York 
7- The Navstar global positioning system, T. Logsdon, 256 pp, 
Van Nostrand Reinhold, New York 
8- Transfer of vertical geodetic control using only one GPS 
receiver. W. Featherstone, V. Dent, The Australian surveyor, 
Vol. 47, No |. 
KE 
AB 
The 
Ing 
cre 
abc 
Fol 
Thi 
cori 
KU 
Die 
Ing 
Dat 
vor 
In d 
ven 
By 
(PA 
Alto 
for 1 
proc 
the 
map 
heig 
prov 
land 
thesi 
and