Short answer: a well-planned drone survey with RTK or PPK positioning and ground control points typically achieves accuracy of a few centimetres. The real figure for your project depends on the ground sample distance (GSD), the ground control, the terrain and the processing, and it should always be proved with independent checkpoints in an accuracy report.
1. Ground sample distance (GSD): the level of detail
GSD is the size of one image pixel on the ground. A GSD of 2.5 cm means each pixel covers 2.5 cm × 2.5 cm. It depends on the camera sensor, the lens and the flying height: fly lower or use a higher-resolution camera, and the GSD gets smaller.
GSD sets the limit on what you can see and measure. You cannot reliably map a feature smaller than a few pixels. A common rule of thumb is that horizontal accuracy with good ground control is around one to two times the GSD, and vertical accuracy around one and a half to three times the GSD.
For reference, a typical 20-megapixel survey drone camera flown at about 100 m above ground gives a GSD of roughly 2.5–3 cm.
2. Ground control points (GCPs): tying the map to the real world
A GCP is a clearly marked target on the ground whose coordinates are measured with DGPS. During processing, GCPs anchor the drone map to the project coordinate system and correct errors in the drone's own position data.
- Spread GCPs around the edges of the site and through the middle, not in one cluster.
- On long corridors such as highways and railways, place them along the full length, alternating sides of the alignment.
- Measure them in the same coordinate system and height datum the design team will use.
3. Checkpoints: proving the accuracy
Checkpoints are measured exactly like GCPs but are not used in processing. After the map is built, the coordinates read from the map are compared with the surveyed coordinates. The difference, usually reported as RMSE (root mean square error) in X, Y and Z, is the honest accuracy of the survey. A survey without independent checkpoints has an unproven accuracy.
4. RTK and PPK: better positions for every photo
RTK (real-time kinematic) corrects the drone's GNSS position during the flight using a base station or correction network. PPK (post-processed kinematic) applies the same corrections after the flight. Both give each photo a position accurate to a few centimetres, which reduces the number of GCPs needed and makes results more consistent. They do not remove the need for checkpoints.
5. What limits accuracy in the field
| Condition | Effect | What helps |
|---|---|---|
| Dense vegetation | Photogrammetry maps the top of the canopy, not the ground | LiDAR, or ground survey under trees |
| Water and shiny surfaces | Few matching features, so poor reconstruction | Bathymetry for water; extra control nearby |
| Low image overlap | Gaps and distortion in the model | Plan around 75–80% forward and 65–70% side overlap |
| Uniform terrain (sand, fresh earthwork) | Hard to match images | Lower flights, more overlap, more control |
| Wind, haze and harsh shadows | Blurred or inconsistent images | Fly in stable conditions and good light |
Photogrammetry or LiDAR?
For open ground, photogrammetry is usually the most economical choice and also gives a true-colour orthomosaic. Where the ground is under vegetation, or where you need wires and fine 3D detail, LiDAR is the better tool. Many corridor projects combine both. See our photogrammetry page for typical deliverables.
What to ask your survey provider
- What GSD will the survey be flown at?
- How many GCPs and how many independent checkpoints, and where?
- Which coordinate system and height datum will be used?
- Will you supply an accuracy report with RMSE in X, Y and Z?
- How will vegetated areas, water and structures be handled?