A 3.2 m Shift: GNSS Coordinates in a Legacy Kandawala Grid
A licensed surveyor is preparing a cadastral plan for a 1.6 km road widening in Kurunegala District. The design is based on a 1998 survey in the Kandawala datum, Everest 1830 ellipsoid, with a local transverse Mercator grid. The surveyor observes 14 boundary points with network RTK, obtaining ITRF2014/WGS84 coordinates with stated horizontal precision of 0.02 m. A colleague supplies a simple three-parameter translation (ΔX = -110.2 m, ΔY = +5.4 m, ΔZ = -118.7 m) said to convert WGS84 to Kandawala. Applying it puts the new points about 3.2 m east and 1.1 m north of the old plan, and two check points on the old grid differ by 0.4 m and 0.9 m. The surveyor also notices that the RTK observations were made without a local transformation and that the total station tie to the old control used a combined scale factor of exactly 1.000000. The client wants coordinates that 'match the old plan' so that acquisition notices can be issued. A second surveyor proposes using a single old control point as a translation and rotating the new network to best fit the old plan.
1. Diagnose the likely causes of the 3.2 m offset and explain why the proposed single-point translation and rotation is unacceptable.
Read the worked answer
The 3.2 m offset is too large to be explained by the stated RTK precision. Likely causes are: use of an incomplete datum transformation (a three-parameter translation ignores rotation and scale); confusion between ellipsoidal and grid coordinates; omission of the projection scale factor and elevation reduction; and distortion in the old Kandawala-based plan. The two check points differing by 0.4 m and 0.9 m show that the transformation is not consistent across the area, so a single set of parameters is not fitting the old control.
Using one old control point as a translation and then rotating the network to best fit the old plan is unacceptable because it forces the new survey to agree with an old, distorted framework. It can make the plan look neat while moving boundaries away from their best modern position, and it conceals the residuals that should be reported. Rotation to fit also destroys the independent check between GNSS and old control. The correct approach is to use the authoritative national transformation and common control points, then evaluate residuals rather than suppress them.
2. Set out a defensible workflow for delivering coordinates for the acquisition notices.
Read the worked answer
First obtain the current official transformation between ITRF/WGS84-based coordinates and the Kandawala/SLD99 grid, and obtain at least three to five common control points with reliable old-grid coordinates. Transform the GNSS points with the full transformation, not a single translation. Apply the correct projection and scale factor, and check whether heights are ellipsoidal or orthometric.
Compute residuals at the common points. If residuals are within the accuracy needed for the notice, use the transformed coordinates and report the residual statistics. If residuals are larger, do not rotate the network to hide them; instead report the uncertainty and, where possible, use a local best-fit transformation with a residual map. Deliver two coordinate sets where necessary: modern coordinates for design and old-grid coordinates for legal notices, with metadata stating datum, epoch, projection, transformation and residuals. Independent check measurements should be included.
3. The client insists on a plan that matches the old boundaries to 0.05 m. What should you tell the client?
Read the worked answer
Explain that 0.05 m agreement cannot be guaranteed when the old plan is a paper plan in a legacy datum and the check points already differ by 0.4 m and 0.9 m. The old plan itself has an accuracy limit, and the transformation has residuals. Promising 0.05 m would be false precision.
What can be delivered is a defensible position for each boundary based on the best available evidence, with the transformation and residuals documented. If the acquisition notice requires old-grid coordinates, provide the best transformed values and state their uncertainty. If a particular boundary is critical, re-establish it from original monuments and independent survey rather than forcing the GNSS network to fit the old plan.