ERROR SOURCES
Description:
The final position of the survey station is influenced by:
(1) The error in the range measurement.
(2) The satellite–receiver geometry.
(3) The accuracy of the satellite ephemerides.
(4) The effect of atmospheric refraction.
(5) The multipath environment at the receivers.
(6) The quantity and quality of satellite data collected.
(7) The connections between the observed GPS network and the existing control.
(8) The processing software used.
It is necessary, therefore, to consider the various errors involved, some of which have already been mentioned.
The majority of the error sources are eliminated or substantially reduced if relative positioning is used, rather than single-point positioning. This fact is common to many aspects of surveying. For instance, in simple levelling it is generally the difference in elevation between points that is required. Therefore, if we consider two points A and B whose heights HA and HB were obtained by measurements from the same point which had an error δH in its assumed height then:
with the result that δH is differenced out and difference in height is much more accurate than the individual heights. Thus, if the absolute position of point A fixed by GPS was 10 m in error, the same would apply to point B but their relative position would be comparatively error free. Then knowing the actual coordinates of A and applying the computed difference in position between A and B would bring B to its correct relative position. This should be borne in mind when examining the error sources in GPS.
Receiver clock error:
This error is a result of the receiver clock not being compatible and in the same time system as the satellite clock. Range measurement (pseudo-range) is thus contaminated. As the speed of light is approximately 300 000 km s−1, then an error of 0.01 s results in a range error of about 3000 km. As already shown this error can be evaluated using four satellites or cancelled using differencing software.
Satellite clock error:
Excessive temperature variations in the satellite may result in variation of the satellite clock from GPS time. Careful monitoring allows the amount of drift to be assessed and included in the broadcast message and therefore substantially eliminated if the user is using the same data. Differential procedures eliminate
this error.
Satellite ephemeris error:
Orbital data has already been discussed in detail with reference to Broadcast and Precise Ephemeris. Errors are still present and influence baseline measurement in the ratio:
δb/b = δS/Rs
where
δb = error in baseline b
δS = error in satellite orbit
Rs = satellite range
The specification for GPS is that orbital errors should not exceed 3.7 m, but this is not always possible. Error in the range of 10–20 m may occur using the Broadcast Ephemeris. Thus, for an orbital error of 10 m on a 10 km baseline with a range of 20 000 km, the error in the baseline would be 5 mm. This error is substantially eliminated over moderate length baselines using differential techniques.