Sect_6.2.10

6.2.10 Measurement Biases and Errors

COMMENTS ON RESIDUAL BIASES & GPS ERRORS

The residual biases remaining after the appropriate differencing of phase data, collected simultaneously by two GPS receivers (section 6.3.2), are primarily due to the:

differential ionospheric refraction between the two sites (if not eliminated in the L3 "ionosphere-free" combination),
differential tropospheric refraction between the two sites, and
muted, second-order effect of orbit error on a baseline solution.

The uncertainty in the baseline solution caused by a combination of these effects is unlikely to be more than a few parts per million for relatively short baselines (less than about 30km in length). This uncertainty may have different characteristics -- for example, be manifest as a scale error, or a height error -- but is considered to be acceptable for standard GPS surveying as the level of uncertainty is still much lower than that generally required for conventional geodetic surveying. Attention will instead be focussed on several sources of error that can significantly impact on baseline accuracy:

Carrier phase cycle slips.

Multipath and signal interference.

Antenna phase centre offset.

Random measurement error is far too small to be considered of significant concern even for the highest accuracy GPS applications.

ERRORS & RESIDUAL BIASES

CARRIER PHASE CYCLE SLIPS:

SOLVE for in a pre-processing step -- aim is to obtain a "clean" dataset.
DIFFERENCE between-epochs to obtain a more "robust" triple-differenced observable -- solution will be less influenced by "slipped" data.

MULTIPATH & OTHER DISTURBANCES:

AVOID reflective or jamming environments.
OBSERVE at a site for more than 15 minutes -- to average out multipath effect.
USE large ground planes, RF absorbing material or choke-ring antennas.

ANTENNA PHASE CENTRE OFFSET:

IGNORE -- generally below 1cm.
BETTER antennas -- use stable micro-strip or dipole antennas.
Keep antenna and receiver unit together during survey, and always orient antenna in the same direction -- effect on baseline is constant (hence indistinguishable from scale error).

RESIDUAL, UNMODELLED BIASES:

IGNORE -- residual effect on solutions is a function of baseline length, but should be of the order of only a few parts per million.

RANDOM MEASUREMENT ERROR:

Measurement noise is typically at the millimetre level.

Back To Chapter 6 Contents / Next Topic / Previous Topic