Surveys usually include items that are simply things that everybody is already aware of: checking the weather, charging your batteries, taking a tape measure with you. The checks which in fact stop a job having to be done again or a dispute with the client arising are generally the ones that are omitted when there's a time pressure — not since anyone is being careless, but because they seem optional until the moment when they aren't.
Before Either Type of Survey
A handful of checks apply whether you're flying a drone or running a total station, and they're the ones most likely to get rushed:
- Check the airspace and the NOTAMs. Make sure you verify the current restrictions for the site, not just those which you recalled from the last time you flew around it — the airspace situation can change.
- Written permission to use the site. An oral agreement obtained from the person you spoke to on the site does not count as formal authorization, particularly when the facility is active or restricted.
- Existing ground control has been verified, not assumed. You should not place trust in the reliability of a previous benchmark — you have to verify that it is still physically there and also check the distance between your available control points before you decide to use them without making any changes.
- The weather window covering the entire period of the actual survey, and not merely the conditions at arrival.
- Site hazards — such as power lines, ongoing operations, restricted areas, and all other things that prevent you from standing or flying in certain places.
Drone-Specific Checks
The most important checks to carry out before any drone imaging and processing mission are:
- Flight-time calculations against the actual size of the site, done beforehand, not found out during the survey. If a site requires several battery changes and redeployments, planning for this should be carried out right from the beginning.
- GCP target visibility from the air. A target which is accurately positioned on the ground but is hidden by vegetation or by shadow when viewed from above will not be of any use.
- Camera and sensor calibration, confirmed rather than being assumed on the basis of the previous job.
- Local pilot certification and registration requirements applicable to that particular airspace and jurisdiction.
Topographic-Specific Checks
These are the ones that a crew carrying out a topographic survey cannot afford to omit:
- Benchmark distance and reliability, confirmed before fieldwork begins. It is important to detect early on any situation where two reference points are technically usable but too far apart to allow you to confidently establish the required accuracy level, rather than having to identify the problem afterwards.
- Instrument calibration has been checked and recorded.
- Datum and coordinate reference system confirmed before the fieldwork began, rather than being assumed or carried over from the default setting of a previous project.
How Many GCPs Does Your Site Really Need?
This is one of the most frequently asked and most frequently oversimplified questions in the field of drone surveying; there is no single figure that can be applied to all sites, but there does exist a well-established range, which has been supported by both manufacturer recommendations and independent field tests.
The floor is four GCPs; below this number, most photogrammetry software will not be able to reliably georeference the dataset at all. According to Pix4D's own recommendations, a number of 5 to 10 GCPs is adequate for most projects, pointing out that after this figure additional points no longer make a meaningful improvement in accuracy. DroneDeploy also sets a similar minimum of 4, advising 8 for maps covering up to 50 acres. As for Propeller's guidance, a minimum of 5 GCPs is recommended, the number being adjusted according to variations in site elevation and flight altitude.
A field trial carried out by the Nevada DOT provides a useful real-life test of this claim. On a site covering 34 acres, they checked the accuracy against the number of GCPs used: the vertical RMSE remained at about 0.1 feet as the number of GCPs increased from 4 to 10, and only decreased to around 0.05 feet when the number of GCPs was increased to the 14-18 range. That is to say, going from "a handful" to "the range generally recommended" made a difference; going beyond that range brought very little extra gain for a great deal more time spent in the field.
As a general guideline based on the experience of people who have worked on larger sites, as stated in a DroneDeploy community Q&A article: when using a standard (non-RTK/PPK) drone, you should plan to add about one further GCP for every 10 acres that you add to your original set. In the case of a drone that is capable of RTK or PPK, however, this rule is much less strict — often only one extra point is needed for every 100 acres beyond an initial set of 5 to 7, since the drone's own positioning is already carrying out much of the required work.
The positional accuracy classes and the reporting standards that this whole matter comes down to are set out in the ASPRS Positional Accuracy Standards for Digital Geospatial Data — the latest edition (2024) — and this is the industry standard against which the majority of professional accuracy claims, including the ones we have had to justify to a client in the past, are ultimately assessed.
The Check Most Teams Skip Entirely
All the previous items were merely technical checks; this one isn't: you should verify what the client really needs the deliverable for, not just accept the specifications as stated in the contract. A survey that covers all the figures mentioned in the brief but fails to address the actual question the client cares about — such as a particular elevation at a specific location, a feature that is expected to be clearly visible, or a format that their own software can open — is a survey that is technically successful yet still leads to a dispute. It is cheaper to get that discussion right before you go on site than it is to deal with it after the delivery has been made.
Two of the gaps referred to here are not imaginary — they are the very mistakes that occurred on actual projects we have handled in the past. The benchmark-distance check mentioned above is the same one that was picked up at an operational dam site before it turned into a serious issue; this is explained in full in how we proved a GSD spec was actually met following a client's question about it. Likewise, the flight-time calculation mentioned above is the same one used to determine whether or not a site is suitable for a drone, as set out in when a site is too big for a drone.
