Drone survey and aerial photogrammetry
Photographs turned into measurements - a week of topo in one sortie, provided somebody put the ground control out first.
Last updated 2026-08-25
What is Drone survey and aerial photogrammetry?
A drone survey is photogrammetry with a flying camera. A small uncrewed aircraft flies an automated grid at a fixed height, taking overlapping photographs, and processing software finds the same features across many images, works out where each photograph was taken from, and builds a dense three-dimensional point cloud from that. Out of it come an orthomosaic - a scaled, map-accurate photograph of the whole site - a surface model, contours and a mesh. Typical flying heights for site work sit around 40-120 m, giving a ground sample distance of roughly 1-3 cm per pixel; overlaps are commonly set around 70-80% along the flight line and 60-75% between lines; a battery gives something in the order of 20-30 minutes in the air, and a single sortie can cover tens of hectares. Lidar payloads do exist and are the answer where vegetation has to be seen through, at several times the cost.
The accuracy comes from the ground, not the aircraft. Ground control points - marked targets surveyed onto project control - are laid out across the site, typically five to ten well spread to the corners, the edges and the middle, with more on a long linear job and more again where there is significant height range. Feed those into processing and the model is pinned to the real world; skip them and you have a beautiful picture floating in roughly the right postcode. With good control, achievable accuracy on open ground commonly lands in the region of 2-5 cm horizontally and somewhat worse vertically - vertical is always the weaker axis. Aircraft carrying precise satellite positioning reduce how many control points are needed, but they never remove the need for independent checkpoints, because a checkpoint is the only thing that turns a claimed accuracy into a measured one. The survey specification states the accuracy required; the checkpoint residuals are the evidence.
Where it falls down is worth knowing before you sell it to anyone. Photogrammetry measures the top surface it can see, so long grass, standing water, scrub and tree canopy all read as ground - a surface model is not a terrain model until it has been classified, and classification through dense vegetation is guesswork. It cannot see into a trench, under a slab or beneath a roof. And it is the one survey method with an operating envelope set by other people: permissions from the aviation authority, flight restriction zones around aerodromes, minimum distances from uninvolved people, constraints over built-up areas, landowner consent for take-off and overflight, third-party insurance, and wind, rain and low winter sun. On a tight urban site, the answer is frequently that you cannot fly at all.
How does Drone survey and aerial photogrammetry work, step by step?
- 1
Step 1: Plan the flight and clear the permissions
Check the airspace over the site and the surrounding area for restriction zones, and confirm the operator holds the permissions and insurance the flight needs. Then plan the ground: take-off and landing point, cordon, observer position, who on site needs telling, and what happens if the link drops or the aircraft has to come down in a hurry. Neighbours, occupied buildings and anyone working outside the site boundary all form part of that assessment. None of this is paperwork for its own sake - flying over people is the single fastest way to end the programme of works and the operator both.
- 2
Step 2: Set and survey the ground control
Lay marked targets - painted crosses, chequered boards or ground mats - spread to the extremities of the area and through the middle, and on a site with real height range put some high and some low. Survey each one onto project control on the same datum as the design model. Then hold several back as checkpoints and do not feed them into processing. The temptation is always to use every point to improve the result; resist it, because a model with no independent checks has no measurable accuracy at all, only a self-assessment.
- 3
Step 3: Fly the mission
Fly the automated grid at the height that gives the required ground sample distance, with the planned overlaps, and keep the camera settings consistent across the sortie. Add a second grid at right angles, or a set of oblique passes, wherever vertical faces matter - a single nadir grid models a stockpile well and a building elevation badly. Fly in even light: high sun or flat overcast, not long shadows and not through changing cloud, because a mosaic assembled from wildly different exposures looks like a patchwork and measures like one too. Watch the wind and the battery, and keep the aircraft in sight.
- 4
Step 4: Process the imagery
Alignment matches features between images and solves for camera positions; ground control points are then identified in the images and the whole block is adjusted onto them; a dense cloud, a surface model and an orthomosaic follow. Expect this to take hours of machine time on a large dataset, so it is not the answer to a question asked at four in the afternoon. The processing report carries the numbers that matter - the reprojection error, the achieved ground sample distance and the control residuals - and it belongs in the deliverable, not on the processor's hard drive.
- 5
Step 5: Check against the checkpoints and classify the ground
Compare the model at every checkpoint held back earlier and report the residuals against the accuracy the specification requires. Then classify: separate ground returns from vegetation, buildings, plant and stockpiled material to get a terrain model out of a surface model. Review that classification by eye against the imagery, particularly along vegetated boundaries and around water, and mark the areas where the terrain is interpolated rather than measured. Anything that has to be certain - a formation level, a drainage invert - still gets a rover on it.
- 6
Step 6: Deliver it and put it to work
Surface against surface gives volumes: this month's model minus last month's, or the model minus the design, with cut and fill shaded across the whole area rather than interpolated between cross-sections tens of metres apart. The orthomosaic drops under the design drawing as a dated progress overlay and settles most of the arguments about what was where and when. Fly the same plan each month and the outputs are directly comparable, which is the whole point - a one-off flight is a picture, a repeated flight is a measurement system.
What are the benefits of Drone survey and aerial photogrammetry?
- Covers large open areas in a fraction of the time a rover walk takes, with nobody walking across a live earthworks face
- Volumes come from surface to surface across the whole area, not interpolated between widely spaced cross-sections
- A dated orthomosaic every month is the cheapest progress record and the strongest dispute evidence on the job
- Reaches roofs, façades, high steelwork and flooded ground without scaffold, rope access or an elevating work platform
- One sortie yields survey data, an inspection record and usable imagery for the client and the marketing team
- Repeatable - fly the same plan monthly and the surfaces line up for direct comparison
What are the limitations of Drone survey and aerial photogrammetry?
- It measures the visible top surface only - grass, water and canopy read as ground until they are classified out, and classification through dense cover is estimation
- Accuracy depends entirely on ground control, and without independent checkpoints there is no evidence it met anything
- Operating restrictions near people, built-up areas and aerodromes rule out or heavily constrain many urban sites
- Weather closes flying days, and low winter sun ruins a mosaic even when the aircraft can fly
- It cannot see into trenches, under slabs, inside buildings or beneath vegetation - services and excavations still need conventional survey
- Processing a large dataset takes hours of machine time and a serious workstation, so turnaround is next day, not next hour
What is Drone survey and aerial photogrammetry best suited for?
What plant does Drone survey and aerial photogrammetry need?
- Small uncrewed aircraft with a survey-grade camera, spare batteries and a ground controller
- Ground control targets or mats, plus a GNSS rover or total station to survey them
- Photogrammetric processing software and a workstation with the memory and graphics to run it
- Anemometer, cordon equipment and signage for the take-off and landing area
- Lidar payload where vegetation has to be seen through rather than around
- A second person as observer, and radio or phone contact with the site team
How is Drone survey and aerial photogrammetry quality-checked?
- Airspace check, operator permissions and third-party insurance confirmed and recorded before the case is opened
- Ground control surveyed from project control, on the same datum and coordinate system as the design model
- Independent checkpoints held back from processing, with their residuals reported in the deliverable
- Processing report issued showing reprojection error, achieved ground sample distance and control residuals
- Ground classification reviewed against the imagery before any volume is calculated from the surface
- Volumes reconciled against a second source - a lorry count, a weighbridge or a rover check - before they are certified for payment