RTK GNSS setout on a constrained CBD site
RTK GNSS is being used to set out a 200 m retaining wall and adjacent pad levels on a tight CBD site surrounded by 5 to 8 storey buildings. The base receiver sits on a mark 4.2 km away with a single-base radio link; the base coordinate was derived from an earlier RTK occupation. Positions show 18 to 21 satellites with PDOP between 1.8 and 2.3, all fixed. The same mark occupied 20 minutes apart differs by 45 mm in plan. A total station resection to three building corners and two cadastral marks puts the RTK-derived coordinate 30 mm away. RTK ellipsoidal heights reduced with the local geoid model differ from a trigonometric height by 60 mm. The wall has a plus or minus 20 mm setout tolerance.
1. Which RTK results would you trust for setting out this wall, and what would you do before any setout?
Read the worked answer
None of them, as they stand, because 45 mm of repeatability cannot meet a plus or minus 20 mm requirement - and that is before accuracy is even considered.
Separate the two questions:
- Precision is the 45 mm spread between repeat occupations of the same mark. It is already more than twice the tolerance. Good PDOP and a fixed flag say the ambiguity was resolved, not that the position is good; under an urban canyon, multipath from the buildings is the dominant error and it does not show up in PDOP.
- Accuracy is the 30 mm disagreement with the total station and the 60 mm height disagreement. Both point to a systematic problem in the base or the coordinate transformation, not to random noise.
Actions before setout:
- Establish or recover verified control - at least three marks, preferably on a network/CORS solution, with the base set on a known mark rather than on a coordinate derived from an earlier RTK occupation (which doubles the error).
- Run an independent check to at least three marks spread around the site and look at the residuals, not just whether they are 'close'.
- Verify the datum and transformation: confirm whether the base is in NZGD2000 and whether any local circuit or site calibration is in play, and check the epoch/projection parameters.
- Do the actual setout with the total station from the verified control, using RTK only for reconnaissance and for locating services. The total station gives an internal consistency and a checkable residual that RTK cannot in this environment.
2. The 60 mm height difference between RTK-plus-geoid and the trigonometric height. Explain the likely sources and how you would resolve the height control for the pad levels.
Read the worked answer
A 60 mm discrepancy is large for a height that will control pad levels, so treat it as a warning, not a rounding error.
Likely sources, roughly in order of size here:
- Base height error propagating as a constant bias. The base coordinate was RTK-derived, so its ellipsoidal height carries the same order of error and it is transferred to every rover position.
- Geoid model limitation. The model is smooth and nationally fitted; local gravity anomalies and deep fill can produce decimetre-level departures. Its stated uncertainty is typically tens of millimetres, so a 60 mm disagreement is plausible on the geoid side alone.
- Antenna height and plumb. A mis-measured pole height goes straight into every height; a tilted pole adds both a height and a horizontal error.
- The check itself. The trigonometric height carries curvature and refraction corrections and its own vertical angle and distance errors; it is not automatically the truth either.
Resolution for pad levels:
- Hold a verified vertical control mark on NZVD2016 with a documented height and provenance.
- Transfer height to the pads by level run or total station from that mark, closing back onto a second mark.
- If RTK heights must be used, do a site calibration from at least four well-distributed known marks with independent checks, and report the residuals. Remember a calibrated RTK system only reproduces the control it was calibrated on - its accuracy away from those points is unverified.
For a plus or minus 20 mm element, I would not use RTK heights as the construction height datum at all.
3. What would you record so that the setout can be defended later if the wall is challenged?
Read the worked answer
The record needs to let someone else reproduce the result and see where the limits were.
- Control: the marks used, their coordinates and heights, their provenance (CORS, previous survey, cadastral mark), the adjustment or calibration report with residuals.
- Base station: location, coordinate source, antenna model and height, and the date/time and epoch of occupation.
- Rover: instrument and antenna, pole height and measurement method, and for each key point the fix status, satellite count, PDOP, and horizontal/vertical residuals.
- Independent checks: the total station resection and check-shot results, the differences, and the decision made about which method governed.
- Datum and transformation: the projection, datum, geoid model and version, and whether a site calibration or local transformation was applied.
- Method statement: a short note that RTK was used for reconnaissance and the total station for setout, with the tolerances each method can actually deliver on this site.
The point of the record is not to prove the numbers are perfect; it is to show that the limitations were recognised and that the method chosen was appropriate to the tolerance.