Ground control points for drone surveys: how to specify and validate them
What a buyer should require for ground control, independent checkpoints, geodetic reference and positional-accuracy evidence in a drone survey.
A one-centimetre orthomosaic pixel does not prove that its coordinates are accurate to one centimetre. Pixel size describes visible detail; positional accuracy also depends on reference, ground control, flight geometry, processing and independent validation.
Ground control is therefore not a minor line item. It is part of the evidence used to accept —or challenge— the use of an orthomosaic, point cloud, terrain model or derived plan for a particular decision.
For drone surveying, GeoSAT defines that traceability as part of scope. If you are still selecting a method, compare the purpose of a LiDAR aerial survey with the costs and limitations of drone topography.
Establish which decision the product must support
Before counting points, decide how the result will be used: preliminary design, inventory, construction measurement, boundary work, volume calculation, cartographic updating or support for a cadastral process. Each use may require a different level of control, accuracy, completeness, cutoff date and accountable professional.
Do not request “maximum accuracy” without a tolerance tied to the product. Instead, specify the reference system, statistical measure, confidence level, dimensions to assess and the treatment of vegetation, water, shadows or insufficient visibility. Acceptance should measure what the decision needs, not a marketing figure.
Keep control separate from independent checkpoints
A ground control point enters the photogrammetric adjustment to orient or improve the product. A checkpoint is held out of that adjustment to evaluate the result independently. If every point is used in adjustment, a report may show small residuals without demonstrating that the product works beyond known locations.
The USGS guidance on UAS-imagery calibration notes that point accuracy and spatial distribution directly affect geometric accuracy, and cautions that image resolution alone does not establish product accuracy.
The contract should identify every point by role and retain its coordinates, capture method, date, observations, photograph and condition. A shifted, destroyed or misidentified point should not silently reappear in the final report.
Specify the geodetic reference and heights
In Colombia, a coordinate table alone is not enough. The team should declare at least:
- Datum and reference epoch.
- Coordinate system and projection used in fieldwork, processing and delivery.
- Height type —ellipsoidal, levelled or other— plus model and transformation where applicable.
- GNSS equipment, method, corrections, base station or service, and observation period.
- How control connects to the network or documented references.
The IGAC publishes the National Geodetic Network and the MAGNA-SIRGAS framework. For official basic-cartography products, Resolution 471 of 2020 describes the reference system and control points within its specifications. Always confirm which instrument, vertical model and scope apply to the specific engagement.
Design a network that tests the whole area
Placement matters more than repeating points at one convenient location. Distribute control and checkpoints to represent edges, corners, the interior, elevation changes and higher-risk sectors. Use stable targets that are visible in the imagery and safe to occupy; document areas that could not be controlled.
No fixed point count fits every project. Size, shape, terrain, obstructions, camera, flight height, overlap, required accuracy and georeferencing method change the design. A pilot over a representative area can test whether the proposed network and processing flow meet the acceptance criterion before scale-up.
Ask for auditable evidence
A useful specification can require:
- Flight plan, covered area, date, camera and relevant parameters.
- A sketch or layer of control and checkpoint roles, with photographs and field log.
- Observation files, correction records and reference-system rationale.
- Processing configuration, software versions and transformations used.
- Error report by checkpoint, statistical method and units, separated from adjustment residuals.
- Inventory of gaps, non-visible surfaces, low-confidence areas and use restrictions.
- Deliverables with metadata, consistent naming and a means to retain evidence.
ASPRS positional-accuracy standards provide a reference for relating control quality, independent checkpoint assessment and the final product. They do not replace Colombian specifications or project-specific criteria, but they help avoid confusing internal adjustment with external validation.
Close by accepting the result, not the promise
Final testing should compare the product with agreed checkpoints and report what could not be assessed. If an area lacks visibility, reliable control or sufficient geometry, that limitation should travel with the product. Hiding it moves the risk to whoever uses the map next.
GeoSAT can prepare the ground-control specification, review a supplier's evidence or deliver the complete survey. The aim is not to impose more points by default; it is to produce information whose accuracy, reference and limitations are defensible for the intended decision.