Key takeaways
- A drone measures surfaces it can see from the air. Canopy, water, walls, soil and enclosure are the four things that break that, and no amount of better equipment changes it.
- The expensive version of this is not a refused job. It is a complete-looking deliverable where the blind spot was filled by interpolation and nothing on the page says so.
- Every blind spot has a standard ground pairing: LiDAR or a crew under canopy, sonar or a survey boat over water, an internal robot or borescope inside enclosures, locates and ground-penetrating radar below grade.
- Ask what is excluded before the flight, not after the deliverable arrives. The exclusion list is the most useful page in a drone survey scope.
Most of what gets written about commercial drone work is about what it can do. That is the easy half. The half that decides whether a project goes smoothly is knowing what aerial capture cannot reach at all, because those gaps are rarely visible in the finished product.
A drone survey measures surfaces that can be seen from the air. Everything below is a consequence of that one sentence. None of it is a limitation of our aircraft or a problem that a better sensor solves next year. These are properties of the method.
Ground under a closed canopy
Photogrammetry builds a surface by matching the same physical point across overlapping photographs. Over a forested area, the points it can match are leaves and branches. What comes back is a canopy surface, not a terrain model, and it will look entirely convincing because a canopy surface is a smooth continuous thing with contours on it.
LiDAR does better, because individual pulses find gaps in the foliage and return from the ground beneath. How much better depends on how closed the canopy is, the time of year and how the flight was planned. Under dense mature cover the ground returns get thin, and a terrain model built from thin returns is mostly interpolation between the few places the sensor actually reached. Our LiDAR versus photogrammetry comparison covers where that line sits.
Pair it with: LiDAR rather than photogrammetry where the canopy is open enough, and a ground crew with a rover or a total station where it is not. If terrain under trees is what the project actually needs, say so at scoping so the method is chosen for it.
Anything below a water surface
A camera sees the water surface, plus whatever glare and reflection the sky is putting on it. It does not see the bottom of a pond, a settling cell, a borrow pit, a ditch that is holding water or a river channel. In shallow, clear, still water you can sometimes make out the bed, but that is an image, not a measurement, because refraction has bent the light path and the apparent depth is wrong.
This one catches out volume work more than anything else. A borrow pit or a pit floor holding spring water produces a perfectly ordinary surface model where the water surface is treated as ground, and the volume computed from it is wrong by however much material sits under the water. The number is not flagged, because to the software the water surface was just a surface.
Pair it with: a bathymetric survey, a sonar boat, or dewatering before the flight. At minimum, the deliverable should state where the water line was at capture, so anyone reading the volume later knows what was excluded.
Inside a structure or an enclosure
The outside of a tank, a silo, a stack, a vessel or a building is aerial work. The inside is not. An exterior flight tells you about the shell, the roof, the insulation, the coatings and the accessible fittings. It tells you nothing about internal corrosion, internal liner condition, deposits, scale or a defect that has not yet reached the outer surface.
Internal inspection is a separate discipline that uses a collision-tolerant aircraft flying inside the asset, which is a different aircraft, a different pilot qualification and a different job. We do that work too, described on the tank inspection case study, but it is not something an exterior survey quietly includes.
Pair it with: an internal drone inspection, a borescope, or the inspection procedure your integrity program already specifies. Treat exterior and interior as two scopes, because they are.
Anything below grade
Buried utilities, pipe, cable, foundations, tank bottoms, contaminated soil and voids are all invisible to any aerial sensor. Surface expression sometimes hints at what is underneath, such as a settlement scar, a differential thaw pattern or a vegetation line following a trench, and thermal capture occasionally makes those hints clearer. A hint is not a locate, and nobody should dig on one.
Pair it with: a utility locate, ground-penetrating radar, or the as-built records. Where a drone genuinely helps is afterward: the locate marks get flown and captured into a dated, georeferenced record instead of living on a set of paint marks that the next rain removes.
Inside a wall, a roof or an assembly
Thermal imagery is the blind spot people are most likely to misread, because it looks like it is seeing inside something. It is not. A thermal camera measures the temperature of the outermost surface it is pointed at. When a roof scan shows a warm area, it is showing a surface temperature difference that is consistent with trapped moisture in the assembly below, which is a strong indication and not a direct observation. Where the water got in is usually somewhere else entirely, as we covered in thermal versus visual imagery on a roof inspection.
Pair it with: a physical verification on the roof. An anomaly should be confirmed with a moisture meter or a core cut before a repair scope is written against it.
The things that are not blind spots but still are not ours
Two more worth naming, because they cause as many misunderstandings as the physical limits.
The first is legal boundary. A survey-grade aerial deliverable can be extremely accurate and still not be a boundary determination, which has a defined process and a defined signatory in Alberta. Who signs off drone survey data sets out which work needs a professional stamp and which does not.
The second is interpretation in your discipline. We produce measured imagery and data and we will tell you plainly what it supports. The geotechnical, structural, silvicultural and integrity judgements stay with the professionals who carry the responsibility for them.
What to ask before the flight
The most useful section in a drone survey scope is the exclusion list, and it is the section most often missing. Four questions cover nearly all of this:
- What on this site will the flight not see, named specifically rather than as a general disclaimer?
- Where the surface is uncertain, how is that marked in the deliverable rather than smoothed over?
- Where was the water line, the canopy edge and the snow cover at the time of capture?
- Which of these gaps matter for the decision this data is supporting, and what covers them?
An operator who answers those without hedging is describing a method they understand. One who says the drone captures everything is describing a sales position, and the gap between the two shows up later, in a number somebody has already built a decision on. If you are putting a job out, the clauses worth writing down are in what to put in a drone survey RFP.