RTK control for a bridge setout beside a steel portal
A 90 m concrete bridge is to be built over a tidal river. The design requires pier positions and deck levels to plus or minus 10 mm relative to a site datum. You establish four control marks around the abutments. Static GNSS on two marks gives a 2.4 km baseline to the LINZ passive control network, with horizontal residuals of 8 to 12 mm and ellipsoidal height residuals of 25 mm. RTK on all four marks gives horizontal residuals of 15 to 20 mm and ellipsoidal height residuals of 35 to 45 mm; the best fixed solution is 18 mm horizontal and 42 mm vertical. The site is beside a steel portal and a line of mature trees. The contractor wants to set out the piers tomorrow using RTK and the NZVD2016 geoid model. The design engineer asks whether the control is fit for purpose and what independent check will be used before setout.
1. Assess whether the RTK-derived control meets the plus or minus 10 mm setout requirement. State what additional observations or adjustments you would require before approving setout.
Read the worked answer
The RTK control does not meet a 10 mm setout requirement. A best fixed solution of 18 mm horizontal and 42 mm vertical is the observation noise, not a guaranteed error bound, and the site has steel and trees that can cause multipath and height bias. For a 10 mm design tolerance, the control should be significantly better than the setout tolerance, not comparable to it. I would treat the static GNSS marks as the primary control, adjust the static baseline in the national datum, and then establish secondary control by total station traverse or network from those marks. The total station work should use forced centring, reciprocal observations where practicable, calibrated equipment and a least-squares adjustment with redundancy. I would require an independent check across the site, for example a closed traverse or a check shot between two marks not used in the setout, and I would compare the result with the static control. RTK may be used for reconnaissance or for non-critical earthworks, but not for approving pier positions and deck levels at this tolerance. The setout should be released only after the adjusted control closes within a stated control tolerance and the independent check agrees.
2. Explain how you would treat the ellipsoidal heights and the NZVD2016 geoid model, including any check against a local benchmark.
Read the worked answer
GNSS gives ellipsoidal heights, not heights above the local vertical datum. The NZVD2016 geoid model converts ellipsoidal height to normal-orthometric height, but the conversion carries model uncertainty and can be affected by local datum distortions. I would not take a single RTK height and use it directly for deck levels. I would compute the static GNSS ellipsoidal heights, apply the geoid model consistently, and then check the result against the nearest reliable NZVD2016 benchmark by differential levelling or a total station height traverse. If the check shows a systematic offset, I would investigate before applying any local correction. For the bridge itself, I would transfer the site datum by levelling from the checked control and use total station height observations for the pier and deck setout. I would record antenna heights, the geoid model version, the datum and epoch, and the results of the independent height check. The key point is that the geoid model is a conversion tool, not a substitute for a local height check.
3. The contractor proposes to set out from a single RTK base placed on one control mark. What risks and controls would you require?
Read the worked answer
A single RTK base on one control mark makes the entire setout depend on that one coordinate. If the mark is disturbed, miscoordinated or affected by a local distortion, the error propagates to every pier without detection. There is also no independent check of the base position, and the rover residuals can look good while the base is wrong. I would require the base to be set on a checked primary mark, with at least one independent check mark occupied before setout. The check should be a different technique where possible, such as a total station shot or a static GNSS check, not simply another RTK fix from the same base. I would require the contractor to record the base coordinate, antenna height, fixed solution status and check-shot residuals. If the check fails, setout stops. For the bridge, I would still require total station setout of the piers from the adjusted control, with RTK used only as a supporting check. The control network, not the convenience of a single base, must govern the works.