The 58-Metre Shift in the Parcel Layer

A constant 58 m east / 12 m north shift confined to one LGA is a systematic error, not random noise. The tests below separate the candidates.

Test 1 - Is the shift constant across all sheets in the LGA? Select 8–10 known survey beacons with published coordinates in the LGA and compare digitised positions against them. If the residual vector is identical everywhere, the cause is a single transformation/offset applied to the whole group. If it varies smoothly from sheet to sheet, the cause is georeferencing.

Test 2 - Datum check. Minna datum to WGS 84 is a real transformation; applying none, or applying the wrong zone or false-easting convention, produces a large constant shift. Test by transforming the known beacons with the ministry's documented parameters and comparing. A 58 m shift is too large to be a geoid or projection scale effect at this scale, but exactly the size produced by a wrong or absent datum shift or a wrong UTM zone/false easting.

Test 3 - Georeferencing check. Four-corner georeferencing of a scanned sheet with a graticule-only margin is a low-order fit with control only at the extremes. Errors in reading the margins, paper shrinkage, and scanner distortion all appear as an affine or higher-order warp. Test by digitising identifiable features (road junctions, beacons) and examining residuals: an affine fit will give residuals that grow towards the sheet centre and vary between sheets.

Test 4 - Sheet-to-sheet consistency. Overlay adjacent sheets and check that common features coincide. If they do not, each sheet has its own georeferencing error on top of any datum error.

Most likely finding: a group-level datum/offset error affecting the LGA, plus per-sheet affine error. Both must be removed; fixing only one will not align the layer.

Accuracy budget

• Source map: a plotted point on a 1:2,000 sheet is conventionally good to about 0.2 mm at map scale = 0.2 × 2,000 = 0.4 m before any other error.
• Scanning and georeferencing: add the georeferencing residual. With a proper 6–9 point control and a first- or second-order polynomial fit, residuals of 0.3–1.0 m are realistic for scanned analogue sheets; with four corner points they can be several metres.
• Digitising: cursor/operator error, typically 0.2–0.5 mm on screen, which at 1:2,000 is another 0.4–1.0 m.

Adding these in quadrature gives a realistic expectation in the region of ±1 m, and certainly no better than ±0.5 m, for the corrected layer. It cannot be claimed as a 0.1 m cadastral dataset, no matter how many decimal places the coordinates carry.

Publishing

1. Compute and report a statistically defensible figure from checkpoints against independent control - do not quote the nominal map accuracy alone.
2. Record full lineage in the metadata: source sheets, survey datum, scan resolution, georeferencing method, control points used, transformation parameters, residuals, and date.
3. Label the layer explicitly as indicative, not cadastral. It may support planning and visualisation; it must not be used to determine a boundary, issue a title, or resolve a dispute.
4. Keep the original sheets and the unedited digitised versions so that the correction is auditable and reversible.

The maths A3 is half the area of A1: linear reduction = √(1/2) ≈ 0.707. The printed map is therefore at 2,000 / 0.707 ≈ 1:2,828, not 1:2,000. Everything plotted at true scale on the sheet is now wrong by 41%.

Consequences

• Any user scaling a distance from the sheet with a ruler will under-measure by about 29%.
• Line weights, text sizes, and symbol dimensions shrink by the same factor, so a 0.2 mm line becomes 0.14 mm and may break up in printing.
• If a coordinate grid or a stated ratio is relied on for a setback check, the error is material.

Correct practice

1. A graphic scale bar is the only scale indicator that survives reproduction, because it shrinks with the map. Every printed sheet should carry one.
2. A stated representative fraction is only valid on the sheet size for which the layout was designed; if a reduced-format edition is needed, re-lay it out at the new size and state the correct ratio, or annotate it clearly as a reduction.
3. Note on the sheet: source scale, reduction if any, and the date of the reduction.
4. Re-scale line weights and text for the reduced format rather than simply shrinking the A1 layout, and check legibility of the smallest annotation at final print size.
5. Add the same warning to the digital service: users of a web map must be told that screen scale is meaningless and only the scale bar applies.