Field Boundary Map

How to reconcile declared area with mapped area in a parcel register

Harvested wheat field with round bales, beyond a stubble strip with green regrowth

Every registry carries two numbers for the same piece of ground: what the claimant wrote on the form, and what the boundary layer says when you measure it. Most years they're close enough that nobody looks twice. Then a claim comes through three-tenths of a hectare over, or a field that was split last season still shows as one parcel, and someone has to decide whether that's a rounding issue or a problem.

Reconciliation is the process of deciding, parcel by parcel, which of those gaps matter and which don't.

Where the gap actually comes from

Declared area is whatever the claimant entered, usually copied forward from last year's form or estimated off an old survey. Mapped area is whatever your current boundary layer returns when you run the polygon through area calculation. The two drift apart for reasons that have nothing to do with fraud:

None of these tell you anything on their own. What tells you something is the size of the gap relative to the parcel, and whether it repeats across seasons on the same claimant's land.

Setting a discrepancy threshold that holds up

A flat hectare cutoff punishes small parcels and lets large ones slide. Most registries that have been through an audit end up expressing the threshold as a percentage of declared area, with a floor for very small parcels so a 0.2 ha garden plot doesn't get flagged for a measurement noise of two square metres.

A workable version looks like: flag anything where mapped area differs from declared area by more than the greater of a fixed small area (commonly somewhere around 0.1 ha) or a percentage of the declared figure (often quoted in the 3-5% range depending on scheme rules). Below that line, the claim passes as declared. Above it, the claimant's paid area gets capped at the mapped figure, and depending on how far over the line it sits, a reduction or penalty calculation kicks in on top.

Write the threshold into the registry's standard operating procedure before the first claim gets checked against it, not after. An auditor will ask to see the rule applied consistently across the whole book, not invented parcel by parcel as exceptions come up.

A reconciliation workflow that scales past a handful of claims

For a few dozen parcels, a spreadsheet and an afternoon will get you through it. Past that, the bottleneck stops being the arithmetic and starts being the boundary layer itself: if your polygons are two or three seasons old, you're measuring against geometry that no longer matches the ground, and every discrepancy you flag is partly a digitising lag problem dressed up as a farmer-declaration problem.

The practical sequence:

  1. Pull the current boundary layer for the claim year, not a carried-forward one.
  2. Join declared area to mapped area by parcel ID and calculate the difference as both absolute area and percentage.
  3. Apply the threshold rule and split claims into pass, cap-and-reduce, and manual review.
  4. Route manual review to someone who can check the actual imagery for genuine boundary change versus stale digitising.
  5. Log the resolution against the parcel ID so next year's reconciliation starts from a corrected baseline instead of repeating the same flag.

Step 1 is where most registries lose time. A digitiser tracing boundary edits by hand across a whole region takes a season to turn around, and that lag is exactly what leaves stale polygons driving false discrepancies. Field Boundary Map machine-vectorises cultivated-parcel boundaries from high-resolution satellite imagery for a given region and reference year, keeping the mapped side of the comparison current with this year's ground instead of a boundary traced two cycles back.

If your reconciliation runs are flagging the same parcels every year for the same reason, it's worth checking whether the boundary layer is the thing that's out of date, and getting a current one against a reference year is a reasonable place to start.

Get started

← Back to the blog