Skip to main content
Outlook Universal Aerial Services, LLC. Built for the far side of the map. Outlook UAS, LLC. Built for the far side of the map.

What accuracy
really means.

Better accuracy means less positional error. We measure it by comparing mapped positions with independently surveyed reference positions.

Understand the error, see how it is measured, then decide what your project needs.

See what error tells you

Why we
measure error.

Accuracy describes closeness to the correct position. The difference between a mapped position and an independently surveyed reference gives us something we can measure: positional error.

A smaller difference means closer agreement. That is why an accuracy report expresses the result in units of error, such as centimeters.

Less measured error.
Better positional accuracy.

How do we put a number on it?

  1. Compare independent positions.At each checkpoint, compare the mapped coordinate with its surveyed reference. The difference is called a residual.
  2. Summarize the differences.Root mean square error (RMSE) squares the differences, averages those squares, and takes the square root. Larger differences have more influence on the result.
  3. State what the result covers.Name the accuracy component, units, and checkpoint count. The ASPRS product-accuracy result also accounts for checkpoint survey uncertainty.
See a simple RMSE calculation

A hypothetical vertical example: one mapped height is 3 cm below its surveyed reference; another is 3 cm above. The differences are −3 cm and +3 cm.

The signed average is zero, but both heights are off by 3 cm. Squaring prevents those opposite signs from canceling.

  1. Square the differences: 9 cm² and 9 cm².
  2. Average the squares: (9 + 9) ÷ 2 = 9 cm².
  3. Take the square root: √(9 cm²) = 3 cm RMS error.

These two invented differences explain the arithmetic only. This is residual RMS before checkpoint survey uncertainty is included, not a complete ASPRS accuracy test.

Read the measurement definition of accuracy

What evidence does your project need?

Start with how the map will be used and how much positional error that use allows. Then choose the production target, independent testing, or licensed survey responsibility your project requires.

Lower errorHigher error
8 GCPs shownMinimum for target RMSE: project-specific.
Independent test not includedTests with <30 checkpoints require qualified reporting.

Illustrative layouts and error fields. GCP counts are not prescribed minimums; colors are not measured site RMSE. Greener areas show a possible reprocessing outcome, not guaranteed improvement.

Built to the agreed specification.

Established production methods and documented quality checks support the specified accuracy class. Independent accuracy testing is a separate scope.

Produced to ASPRS

What comes with this scope

  • An agreed accuracy class supported by established methods and documented quality checks.
  • A control plan matched to the project.
  • A production record, without an independent tested-accuracy claim.

An accuracy class is the target specified for the product. A production target alone does not prove the finished result meets it.

Map detail

Illustrative point layout and error field. Scroll across the map to read the numbered checkpoints.

More about produced and tested mapping

A few terms.
Clearer decisions.

You do not need to learn the entire standard to ask the right questions.

Looking for the checkpoint minimum?
Does a sharp image mean an accurate map?Resolution shows detail. Testing measures position.

Ground sample distance (GSD) is the ground distance represented by one image pixel. Small pixels can show fine detail, but they do not establish how close mapped features are to their actual positions.

Ask for the accuracy evidence as well as the image resolution.

How are control points and checkpoints different?Control helps build the map. Checkpoints test it.

Ground control points (GCPs) have known reference coordinates used in processing. Independent checkpoints have higher-accuracy reference coordinates reserved for testing the finished result. Those coordinates may be surveyed for the project or come from a suitable established reference.

A point used to adjust the model cannot independently test that same result. Checkpoints measure error; they do not correct the map.

ASPRS sets no universal GCP minimum for UAS photogrammetry. The control plan depends on the site, capture system, and accuracy target. The counts shown in the illustration are not validated minimums for a real project.

Does RMSE tell me the error everywhere?It summarizes the checkpoint test, not each location.

RMSE is an aggregate measure. A reported value is not a maximum-error guarantee for every point in the map.

For comparable tests of the same accuracy component, 3 cm RMSE indicates less measured positional error than 8 cm RMSE. Read either value with the test method, area, checkpoint count, and survey accuracy.

GCP spacing alone cannot supply measured accuracy between points. The heatmaps above illustrate a possible pattern; they are not measured site RMSE.

Is it 20 checkpoints or 30?The current edition uses 30 as the baseline for full testing.

Edition 2, Version 2 (2024) uses 30 independent checkpoints as the baseline for full accuracy testing; Edition 1 (2014) used 20. These are test points, not a minimum number of GCPs.

Fewer checkpoints can still provide useful results. Section 7.16.1 provides qualified reporting for that situation, with the actual count disclosed. It does not make a smaller test equivalent to meeting the full testing requirement, and it does not define the Produced scope.

Checkpoint references must be at least twice as accurate as the required product accuracy for the component tested. For example, a 5 cm horizontal RMSE target requires reference horizontal RMSE of 2.5 cm or better. Checkpoints must also be well distributed and separate from processing control.

Larger projects and some product types need additional testing. Point count alone does not establish conformity.

Read the ASPRS edition update

Read the evidence behind the number.

A useful accuracy statement tells you what was tested, how it was tested, and where the result applies.

What to look for in your report

  • The product and areaIdentify the data version, tested extent, and coordinate reference system.
  • The accuracy resultLook for the class, units, and horizontal, vertical, or 3D component.
  • The independent testCheck the checkpoint count, distribution, residuals, and survey accuracy.
  • The limits of the claimReduced counts, exclusions, and the applicable standard should be stated.
Accuracy statements in the FAQ

Go deeper when you need to.

The ASPRS standard Published by the American Society for Photogrammetry and Remote Sensing. Edition 2, Version 2 (2024), Section 7.16 and Addendum V. Practical project questions Find answers about deliverables, scope, timing, and working with Outlook UAS in the master FAQ.

Tell us what the map
needs to support.

Share the intended use and your reviewer's requirements. We will identify the accuracy scope and evidence your project needs.

Talk through your project Already have data? Verify it