Geodetic and Engineering Surveying, Control Networks and Deformation Monitoring
This domain tests geodetic literacy: which datum and epoch you are working on, how a control network is designed and adjusted, and how to decide whether an observed movement is real. Deformation monitoring questions are usually marked on the significance test and on reference stability, not on the arithmetic.
Datums and control
Horizontal work in Hong Kong is conventionally on the Hong Kong 1980 Geodetic Datum and HK80 Grid; GNSS-derived coordinates from the Satellite Positioning Reference Station Network (SatRef) are on an ITRF-aligned frame and must be transformed with stated parameters. Heights are on the Hong Kong Principal Datum (HKPD), a tidal datum defined by a tide gauge in Victoria Harbour and not identical to mean sea level; the difference is of the order of a metre, so mixing mPD and MSL values creates a systematic offset. Horizontal control uses trigonometry stations and densified networks; vertical control uses benchmarks and closed level runs.
Adjustment and network design
- Redundancy is what allows blunders to be detected; a minimally constrained network has none.
- Least squares adjustment yields coordinates, residuals and an error ellipse; a large semi-major axis signals weak geometry, not bad observation.
- Precision describes repeatability; accuracy describes proximity to truth. Achieving 2 mm precision around a 20 mm systematic error is still a 20 mm error.
- Free stationing and resection need well-distributed control; collinear control gives poor perpendicular strength.
Deformation monitoring
Define a stable reference network outside the zone of influence, observe it in every epoch, and only then compare object points. Use the same datum, instrument, observation procedure and, where possible, the same observer. Set alert, action and trigger values with the designer in advance. Test significance: a difference should exceed the combined uncertainty at the chosen confidence before it is reported as movement.
Worked example
A settlement marker near a tunnel excavation falls 8 mm between epochs. Levelling precision is 2 mm per setup, so the combined uncertainty of the difference is about 2 x root 2, roughly 3 mm at one sigma. Eight millimetres is well beyond that and is significant - provided the reference benchmark itself has been re-observed and found stable. If the benchmark is on the same moving slope, the 8 mm is relative movement only and must be reported as such.
Traps that cost marks
Comparing coordinates across datums or epochs. Relying on a single reference station. Reporting precision as accuracy. Failing to specify alert and action values. Ignoring environmental corrections on long EDM lines.