Ground control is the least glamorous part of a drone survey and the part that determines whether the results can be trusted. It is also the first thing cut when a provider is competing on price. Here is what it is, why it matters, and how to tell whether a quote includes enough of it.
What ground control actually is
A ground control point, or GCP, is a marked target placed on the site whose position has been measured by conventional survey methods, usually GNSS. During processing, the target is identified in the imagery and its known coordinates are used to anchor the model to the real world.
Without control, photogrammetry produces an internally consistent model that floats. Shapes and relative distances can be good while the whole thing sits several feet from where it belongs, tilted, or slightly scaled. Control removes that ambiguity.
Control points and checkpoints are not the same thing
This distinction is the single most useful thing to understand as a buyer.
Control points are fed into processing. The software adjusts the model to fit them, so by definition the model will agree with them closely. Reporting error at your control points tells you almost nothing.
Checkpoints are surveyed the same way but deliberately withheld from processing. Once the model is built, it is compared against them. That comparison is an independent test, and it is the only accuracy figure that means anything.
When a provider tells you their accuracy, ask whether the number came from control points or checkpoints. The answer is revealing.
Does RTK remove the need for it?
Partly, and this is where marketing outruns practice.
An aircraft with onboard RTK or PPK positioning records accurate camera positions at the moment of capture, which substantially reduces how many ground targets are needed. It does not eliminate the value of independent verification. The aircraft can only tell you where it thought it was; it cannot audit its own result.
A common and defensible approach is RTK capture with a reduced number of ground targets, some used as control and at least a few reserved as checkpoints. Any workflow with no ground measurement at all has no way of proving its own accuracy.
Placement matters as much as quantity
Targets should be distributed across the full extent of the site rather than clustered near the access road, and they need vertical spread as well. A site with control only along the flat entrance and none at the top of the cut will show good numbers where the targets are and drift where they are not.
Targets also need to be visible from above, large enough to identify at the flight altitude, and stable between placement and capture. A target that gets driven over by a scraper between survey and flight is worse than no target, because it looks valid and is not.
What this means for repeat flights
If you intend to compare flights over time, and for progress tracking you do, consistency matters more than any single flight’s accuracy.
Permanent or semi-permanent control that survives between visits, measured once and reused, makes each flight directly comparable to the last. Sites that get fresh, arbitrary control every visit produce a series where apparent change includes the wobble between setups, and small real changes get lost in it.
Questions worth asking
How many ground points will be placed, and how many are reserved as checkpoints? What coordinate system and vertical datum will the deliverables use, and does it match your civil model? Will control be re-established or reused on subsequent flights? Will the QC report state the achieved accuracy against independent checks?
A provider who answers these easily is telling you something. So is one who does not.
