The landowner says that he owns 10 acres, whereas your GIS shows 9.63 — who is correct?
Most people decide which side to take right away—the deed states ten, so the GIS must be incorrect. Alternatively, the GIS might be measuring the actual shape, in which case the deed must be outdated. Both of these responses seem reasonable, and both of them ignore the more appropriate question: what was each number intended to answer?
This is the situation: the land did not experience a loss of 0.37 acres between the registry office and your screen; no one moved any fence. It was the model of the land that changed, and a single parcel typically has not one but five area figures, each of which is produced by a different process, in a different coordinate system, from different measurements.
In this article we’ll look at the origins of those five figures, identify which differences are small and which are large enough to make a difference, and also provide a QA procedure that you can carry out before mismatched GIS acreage is included in a tax roll, a valuation report or a development application. Some of the figures mentioned below are based on our own test calculations, and we will state this each time.
Where Five Different Parcel Areas Come From
Begin by looking at the five numbers; nearly all the confusion disappears when you realise that they are five individual measurements and not five attempts at the same measurement.
The record area
The figure stated in a deed or in a recorded plan is a statement that was made at a particular time using the instruments and methods available at that time; it might be approximate and could have been inherited from an earlier plan and carried forward without anyone having reworked the figures. In Douglas County, Minnesota, the county’s parcel service explains in its notes that the tax acreages originally came from the township plats prepared by the General Land Office as part of the Public Land Survey, work which was carried out between 1858 and 1871, and that these can contradict the later legal descriptions that are on file; hence the “10 acres” in your file may be older than anyone who has ever handled it.
The survey-computed area
The surveyor measures the boundary, calculates the coordinates of the corners, and then determines the area from these coordinates. The area exists on a surface, and this surface is important. It may be the ground, or it might be a projected grid like State Plane. The two methods yield different results, and we will look at that soon.
The three GIS numbers
You can obtain up to three figures. The planar area is computed using the x and y coordinates on a flat basis, regardless of the coordinate system in which the layer happens to be. The geodesic area is calculated on the ellipsoid. There’s also the number in the attribute table; this could be one of the first two methods, or it might have been typed in from the deed, or it could be a figure that was calculated years ago in a different coordinate system by someone who has since left the company. None of the field information will tell you which one it is.
| # | Number | Where it comes from | The question it answers |
|---|---|---|---|
| 1 | Record area | Deed or recorded plan | What was conveyed or registered |
| 2 | Survey-computed area | Corner coordinates, on ground or grid | What the survey measured |
| 3 | GIS planar area | Flat calculation in the layer’s CRS | Area in that projection |
| 4 | GIS geodesic area | Calculation on the ellipsoid | Area on the reference surface |
| 5 | Stored attribute | A field in the table | Whatever happened the last time it was filled |
Now for the interesting part: none of these figures is bad. Each one is a valid answer to a different question. The difficulty arises when we compare them as though they were all responding to the same question.
What should I do: use the geodesic area or the planar area for parcels?
The planar area method treats the map as if it were a flat surface, while the geodesic area method follows the curved surface of the ellipsoid. Esri’s documentation on length and area combines planar calculations with projected coordinate systems (preferably one that preserves area) and geodesic calculations with geographic coordinates, pointing out that a single shape measured using two different coordinate systems will yield two different results.
It sounds concerning, so we carried out some tests.
What the numbers say at parcel scale
We selected a square area of 10 acres (approximately 4.05 hectares) at three different sites and determined its area by using various methods, keeping the same ground shape in each case. These calculations are our own.
| Location | Geodesic (acres) | UTM planar (acres) | Web Mercator (acres) |
|---|---|---|---|
| Denver, USA | 10.008 | 10.000 | about 16.9 |
| Sydney, Australia | 10.001 | 10.000 | approximately 14.5 |
| London, UK | 9.998 | 10.000 | about 25.8 |
If you look at the first two columns, the values for the geodesic and planar methods on a 10-acre plot are about 0.01 acre apart from each other. For the same London square on the British National Grid, it is 9.995, and for the Denver square on the Colorado Central State Plane, it is 10.008. The reason why a gap of 0.37 acres arises is not due to a sensible local projection; that gap is more than forty times the largest difference we observed in our test.
A user of the Esri Community mentioned that there was approximately a four-acre difference between planar and geodesic calculations over 20,000 acres, which is about 0.02 per cent (since it was reported by a user, it should be regarded as an illustration only, not as a benchmark).
When the method does matter
For large parcels—such as ranches, forest tenures, mining concessions, and long corridors—the percentage remains low, but when you add up the acres it amounts to a significant figure, since 0.02 per cent of 20,000 acres is exactly 4 acres. When dealing with small parcels, you should calculate using the local projected CRS that was used in the survey; for large parcels, use the ellipsoid. In either case, make a note of the method you have employed. We will refer to this again in the workflow.
When the Coordinate System Is the Problem
Web Mercator inflates everything
Take the third column of the table once more; at Denver’s latitude the same 10 acres is about 16.9 in Web Mercator, and at London’s it is about 25.8. Esri’s documentation on Web Mercator is direct concerning the cause: the system is not conformal, and the distortions in area and distance increase rapidly the further away from the equator you go.
However, consider the direction involved here. Since Web Mercator causes areas to become larger rather than smaller, it cannot generate a value of 9.63, although it is the usual reason when a web-map layer displays an acreage that is much higher than the actual recorded amount. Data that remains in latitude and longitude is similar in nature, as carrying out a planar calculation on degrees does not give you the acreage that you should report. We discussed the broader CRS issue in The Datum Mismatch That Costs You a Client’s Trust.
Grid versus ground
In many parts of the United States, survey coordinates are stored in a State Plane grid, and the lengths on the grid do not correspond to actual ground lengths. The combined scale factor (which is the product of the grid scale factor and the elevation factor) is what connects the two. Since area follows the square of the linear factor, the actual ground area is the grid area divided by the square of the combined factor. A presentation given in New Jersey on State Plane and GIS points out that in the typical situation the grid area is less than the ground area, and San Diego County’s ordinance relating to subdivisions requires the combined factor to be indicated on maps that include both grid and ground distances.
What is its size? According to our calculation, taking into account only the elevation factor, a piece of land of 10 acres changes by about 0.005 acres when the elevation is 1,600 metres and by about 0.009 acres at 3,000 metres. On a ranch of 2,000 acres situated at an elevation of 1,600 metres, this amounts to almost one full acre. Therefore, the difference between grid and ground measurements can have legal significance on large properties, but it cannot account for a 3.7 per cent discrepancy on a 10-acre plot.
Does reprojecting affect the real size of a parcel?
No. When you reproject, the coordinates and the distortion associated with them change, but the land remains in exactly the same position. What changes is the value obtained when the area is calculated in the new coordinate system. And if the area is contained in a typed or imported field, then it is simply a number and will not update itself upon reprojecting the layer.
Reading the gap
The magnitude and the direction of the discrepancy show you which area to examine first.
| What you see | Likely cause to check first |
|---|---|
| GIS area far larger than the record, by tens of percent | Web Mercator or another unsuitable CRS; data in degrees |
| Difference of about 0.1 per cent or less | Planar versus geodesic, grid versus ground, rounding |
| Difference of a few percent | Boundary position, digitising or scan scale, rounded record figure |
| Area of zero, or parcels that do not add up to the block | Invalid geometry, overlaps, gaps |
| Stored value differs from a fresh calculation | Stale field, or calculated in another CRS |
Geometry That Lies: Overlaps, Gaps, Slivers and Invalid Polygons
Invalid polygons that still draw
There’s a good example in the PostGIS validity workshop. A figure-eight shape, consisting of two one-unit squares that meet at a point, displays correctly and appears to have an area of two units. However, if you ask the database for its area, it will return zero, since the area calculation takes the assumption that rings do not intersect themselves. When you use a function such as ST_MakeValid to repair it, the result is a multipolygon that covers the same visible area. The point to take away is simple: a polygon which displays on screen is not necessarily one which has the correct measured area. Since repair can alter the structure, it is important to compare the area before and after the operation. The summary of ST_IsValid also states that transformations or heavy coordinate rounding can cause a valid polygon to become invalid. Moreover, reprojection can invalidate the geometry you were intending to measure.
Overlaps, gaps and slivers
If two adjacent plots have an overlap of 0.5 metres along a common 200 metre boundary, the area of 100 square metres, or about 0.025 acre, is counted twice. A gap of 10 centimetres along that same edge amounts to about 0.005 acre, which belongs to no one. This is a negligible amount for a single plot. When you add up thousands of plots and then compare the total with that of the block or the tax district, the discrepancy becomes apparent. Topology rules (namely, that there must be no overlap and that there must be no gaps) should be applied before relying on any aggregate figures.
When the Digitised Boundary Isn’t the Surveyed One
Turn the area gap into metres
Here’s a method of making a percentage seem tangible: according to our calculation, a 3.7 per cent gap on a 10-acre square is equivalent to moving each of the four sides inward by approximately 1.9 m. Such a degree of positional error is considerable for a modern survey but relatively small when the boundary is taken from an old map.
The shape affects the situation; on a strip which is 20 metres wide and approximately 2 km long (that is still 10 acres), a mismatch of 3.7 per cent only requires about 0.37 metres of shift along each of the long sides. Thin plots, for example road reserves, utility corridors and long narrow lots, are the ones that are most affected. If the parcels that are mismatched are the thin ones, then check the positional accuracy first.
What legacy digitising does to area
We have witnessed this ourselves. As part of the Survey of Pakistan’s national cadastral programme (2022 to 2024), our teams carried out their work in eight divisions, converting decades of hand-drawn Mussavi and Jamabandi land records into a georeferenced digital cadastre. The records had been scanned and were faded as well as inconsistently scaled. The work was carried out right down to the individual Khasra parcels, Murabba grids and Mouza boundaries, with strict quality control and assurance, using up to 200 digitizers working on three shifts and approximately 4,000 ground control points measured by DGPS.
The reason why scale is the one aspect worth remembering is that area increases with the square of the scale. When a scanned sheet is enlarged or reduced by less than 2 per cent along each axis, the area changes by approximately 3.7 per cent (according to our calculation), equivalent to the whole 0.37-acre discrepancy, even though no boundary has been incorrectly drawn on the paper. This is one of the reasons why a programme such as this relies on ground control rather than relying on the printed scale of the sheet.
It isn’t just a story from Pakistan; according to a 2025 paper by FIG relating to Queensland’s digital cadastral database, a cadastre was constructed from existing maps, and the positional accuracy in this case varied from 0.1 m to 250 m.
Old measurements, old datums
Historic plans have also made use of older instruments, units and datums, which means that even a perfectly accurate transcription will differ from current coordinates. We examined this history in The Survey That Measured a Subcontinent by Hand, and the same drone-versus-record trap appears in Your Drone Map Shows the Fence. It Doesn’t Prove the Property Boundary.
What parcel area is legally correct?
It all depends on your location and the purpose of the number, and no legal advice is being given here. But the pattern is consistent enough to be described.
In West Virginia, the guidance regarding acreage states that, for tax purposes, the deeded acreage must be treated as superior to the calculated acreage when the two figures differ. The disclaimers issued by the US Fish and Wildlife Service Land Mapper distinguish between the actual legal acreage—that is, the amount recorded by a survey, patent, deed, or similar document—and the GIS-computed acreage, which they refer to as an approximation and stress that it should not be regarded as the legal acreage.
In these systems, the record has control over the record; although this does not mean that the deed is physically correct, it does mean that it is the authority for legal purposes, while the GIS figure provides an estimate of the actual shape on the ground.
Compare to the uncertainty, not to zero
This point is one that many teams fail to consider. Certain systems incorporate the tolerance directly into the rules. A study from Lithuania, published in Baltic Surveying, describes a permissible area error, set in the real property cadastre regulations, by which a recomputed parcel area may differ from the registered one. A Polish study makes a similar point: the difference between cadastral area and coordinate-based surveying area should stay within the error of determining the surveyed area.
The stakes can be high. A FIG 2022 paper by Yildiz and colleagues discusses Turkish court decisions on state liability for differences between the actual boundary and the registered area, where uncertainty was often left out of the reasoning. That is the same logic as in Two Surveys of the Same Stockpile Give Different Results. Which One Is Correct?: You do not crown a winner; you ask whether the difference is bigger than what the measurements can support.
A Parcel-Area QA Workflow You Can Run This Week
What you have to do is simple to state, and it works in this order.
- Label every number. For each of the five, write down the source, date, CRS and datum, units, and method. If you cannot, mark it unknown. That alone settles many arguments.
- Check the geometry first. Run a validity check, then topology rules for overlaps and gaps, before you calculate anything.
- Calculate in the right place. Use the local projected CRS suited to the parcel, or the survey’s own projection, plus an ellipsoidal calculation for large parcels. Never use Web Mercator for area.
- Match the surface. If you are comparing against a ground survey held in grid coordinates, apply the combined factor (ground area equals grid area divided by the factor squared) and state the direction you applied it.
- Build a reconciliation table. Record survey, GIS planar, GIS geodesic, and stored value, side by side, with the difference in acres and in per cent.
- Convert the gap into metres. Compare it with the positional accuracy of the boundary source. A 1.9 m equivalent shift on a boundary digitised from a 1:10,000 sheet means something very different from the same shift on an RTK-surveyed one.
- Decide, then escalate. If the gap sits inside the uncertainty, document it and keep the record. If it does not, get independent survey evidence from a licensed surveyor, and for drone data, independent checkpoints.
- Store the lineage. Add fields for area value, method, CRS, source, and date. A number that travels without them will start the same argument again.
If you need this kind of check across a whole parcel layer, that is the GIS data work our GIS team does day to day.
So, Who Was Right About the 9.63?
Probably nobody, and probably everybody, for their own question. The deed answers what was registered. The survey answers what was measured. The GIS answers what shape the polygon has in the coordinate system you chose. The land did not change; our description of it did.
So the useful move is not to pick a winner. It is to trace each figure back to its lineage, convert the gap into metres, and compare that against what the measurements can honestly support. If you are sitting on a parcel layer where the acreage will not agree with the record, start with the symptom table above. If it still does not add up, we are happy to take a look.