Satellite Positioning and GNSS
The exam probes whether you understand why GNSS reaches centimetre accuracy - error correlation, differencing and geometry - rather than quoting specifications from memory.
Observables
- Pseudorange (code) gives metre-level positions; carrier phase gives millimetre precision but carries an unknown integer number of cycles (the ambiguity).
- Dual-frequency observation removes the first-order ionospheric delay, which scales as 1/f². The troposphere is non-dispersive and must be modelled or estimated.
- Ambiguity resolution (a "fixed" solution) needs short baselines, dual frequency and low multipath; a "float" solution is decimetre-level.
Constellations and services
GPS (USA), GLONASS (Russia, FDMA on legacy signals), Galileo (EU) and BeiDou (China). Multi-constellation improves availability and DOP in obstructed sites. RTK uses one base station plus a radio or NTRIP link; network RTK (VRS, FKP, MAC) models atmospheric error across a network of reference stations; PPP uses precise orbits and clocks with no base station but needs a long convergence.
Accuracy and geometry
Dilution of precision describes satellite geometry: PDOP scales the position variance, and HDOP/VDOP split it horizontally and vertically. A published accuracy of "5 mm + 1 ppm" is baseline-dependent - the ppm term absorbs orbit and atmospheric decorrelation.
Worked example: a 12 km baseline with a 5 mm + 1 ppm specification gives 5 mm + (1 mm/km × 12 km) = 17 mm horizontally at 1σ. For a 95% statement multiply by about 2, giving roughly 34 mm, and say so explicitly.
Traps
- Treating a float solution as fixed.
- Quoting RTK accuracy without the baseline length; beyond about 10 km the ppm term dominates.
- Raising the mask angle to suppress multipath without noticing the resulting DOP increase.
- Quoting a single accuracy figure: vertical accuracy is typically 1.5–3× worse than horizontal.