How drone mapping works
From the flight plan to the file your engineer opens in Civil 3D: what happens at each step, what decides accuracy, and what to ask before you order a map.
Thousands of overlapping photos, one measured model
A drone map is not one photo. It is hundreds or thousands of photos taken in a grid, each one overlapping its neighbours, with the camera position recorded for every frame. Photogrammetry software finds the same points of ground in many photos, works out where the camera was for each one, and triangulates the 3D position of millions of points. From that point cloud it builds a surface, and from the surface it produces a true-to-scale, straightened image of the site called an orthomosaic.
LiDAR works differently. The drone carries a laser scanner that fires hundreds of thousands of pulses a second and measures how long each one takes to come back. Combined with a precise position and orientation system, each return becomes a 3D point. Because some pulses slip between leaves and branches, LiDAR can measure the ground under vegetation where photogrammetry only sees the canopy. See LiDAR for when it is worth it.
Both methods end in the same place: measured data in a real coordinate system that a surveyor, engineer, estimator or planner can use in their own software. The quality of that data is decided mostly before the drone leaves the ground.
The uses are wide: stockpile volumes at quarries and yards, monthly construction progress, earthworks and cut/fill against a design, pre-purchase and rezoning context for land development, and as-built records at handover. Each use asks something different of the flight. A volume needs a consistent base surface month after month. A progress record needs the same flight plan and the same control every visit, so the comparison is fair. A design surface needs a bare-earth model in the engineer's datum. Decide which one you are buying before anyone plans the flight.
Seven steps from order to deliverable
1. Define the purpose
A marketing overview, a stockpile volume, a topographic surface for design and a monthly progress record all need different accuracy, resolution and deliverables. Write down what decision the map supports and which software will open it. That single sentence drives every setting that follows.
2. Agree the coordinate system
The map must line up with the survey, the design and the property lines. In Canada that usually means NAD83(CSRS) with a UTM or provincial projection and CGVD2013 heights; in the US, NAD83(2011) with a State Plane zone and NAVD88. Get the horizontal datum, projection, vertical datum and units in writing from the surveyor or engineer. A wrong datum is the most common reason a good map is useless.
3. Plan the flight
The pilot sets altitude (which sets resolution), overlap, flight lines, camera angle and speed in flight-planning software, then checks the plan against airspace, obstacles, terrain and the take-off spot. Tall structures and steep ground need terrain-following or extra oblique passes.
4. Set ground control
A surveyor or the crew places and measures ground control points (GCPs) and separate checkpoints with survey-grade GNSS. GCPs tie the model to the coordinate system; checkpoints are held back to test it.
5. Fly and check on site
The crew flies the mission, then checks image count, sharpness, exposure and GNSS logs before leaving. A reflight on the same day costs minutes. A reflight next week costs a mobilization.
6. Process
Images are aligned, tied to control, densified into a point cloud, and turned into surfaces and the orthomosaic. Ground points are classified from buildings, equipment and vegetation to make a bare-earth terrain model. DroneLink processes in-house.
7. QA and deliver
Checkpoint errors are measured and reported, the outputs are reviewed for holes, blur and artefacts, and the files are exported in the formats your software reads. Deliverables arrive by link in the DroneLink project record, with every flight kept together.
Planning settings and what they control
| Setting | What it means | Typical industry practice | What it changes |
|---|---|---|---|
| Ground sample distance (GSD) | The ground size of one image pixel | About 1 to 3 cm per pixel for construction and topo work; coarser for large rural areas | Detail, and the lower limit on accuracy. Lower altitude gives finer GSD and more photos |
| Front overlap | Overlap between photos along a flight line | Around 75 to 80 percent | Reliable matching; more overlap for trees, water and uniform surfaces |
| Side overlap | Overlap between neighbouring flight lines | Around 65 to 75 percent | Coverage without gaps; more flight time |
| Camera angle | Nadir (straight down) or oblique | Nadir for orthos; add oblique passes for facades and 3D models | Quality of vertical surfaces |
| Ground control | Measured targets on the ground | Several GCPs spread around the perimeter and through the middle, plus independent checkpoints | Absolute accuracy and proof of it |
| Positioning | How camera positions are known | Standalone GNSS, RTK or PPK | How much ground control you need |
| Shutter and speed | Exposure time versus ground speed | Fast enough that motion blur stays under a pixel | Sharpness |
GCPs, RTK and PPK: how the map gets to the right place
There are two kinds of accuracy. Relative accuracy is how well distances and shapes inside the map are measured. Absolute accuracy is how well the map sits on the real coordinates. A map can measure a stockpile perfectly and still be a metre away from where the design says it is.
A drone's ordinary GNSS knows where it is to within a few metres. Without ground control, the whole map can shift or tilt by that much. As general industry knowledge, work flown with RTK or PPK positioning and checked against ground control typically lands within a few centimetres; work without either is typically in the range of decimetres to metres.
RTK (real-time kinematic) corrects the drone's position during flight from a base station or a correction network. It depends on a steady radio or cellular link. PPK (post-processed kinematic) records raw GNSS data on the drone and a base, and corrects the positions after the flight. It does not need a live link, which suits remote sites. Either one reduces how many GCPs you need. Neither removes the need for checkpoints, because without them nobody can prove the accuracy.
Vertical accuracy is usually looser than horizontal, and it gets worse on vegetation, water, shiny roofs and dark, uniform surfaces. For any survey-grade use, a licensed surveyor should set the control, set the standard the map is tested against and sign off the result. The drone is a measuring tool in a surveyor's workflow, not a replacement for one. Read more on accuracy.
What you get and what opens it
| Deliverable | What it is | Common format | Opens in |
|---|---|---|---|
| Orthomosaic | A straightened, measurable aerial image | GeoTIFF, JPG with world file, ECW | GIS, CAD, Google Earth (as KMZ) |
| Digital surface model (DSM) | Heights of the top of everything: roofs, trees, equipment | GeoTIFF | GIS, civil design |
| Digital terrain model (DTM) | Bare-earth heights with objects removed | GeoTIFF, LandXML surface | Civil 3D, GIS |
| Contours | Lines of equal elevation at a set interval | DXF, DWG, SHP | AutoCAD, Civil 3D, ArcGIS, QGIS |
| Point cloud | Millions of 3D points, often classified | LAS, LAZ | CAD, GIS, point-cloud viewers |
| 3D mesh | A textured 3D model of the site or building | OBJ, FBX, 3D Tiles | 3D software, BIM, visualization |
| Volume report | Stockpile or cut/fill quantities with method and base surface | PDF, CSV | Anyone |
| Accuracy report | Checkpoint errors and processing summary | Surveyor, engineer, auditor |
Why drone maps fail, and how to avoid it
The map does not line up with the design
Wrong datum, projection, geoid or units. Everything looks right until the engineer overlays it.
Nobody can say how accurate it is
No checkpoints were measured, so the accuracy is a guess.
Volumes that change between surveys
Different base surfaces, different toe lines, different methods from month to month.
Holes and melted edges
Too little overlap, moving shadows, trees, water or wind-blown vegetation.
Files lost across email threads
The survey firm has one version, the site office another, the architect a third.
Who needs what from a mapping flight
Surveyor
- They need from you
- Raw data, control coordinates and the processing report
- You need from them
- The coordinate system, control and sign-off
- On the DroneLink record
- Control and checkpoints agreed before the flight; reports in the project record
Civil engineer
- They need from you
- A bare-earth surface in their design software
- You need from them
- The design surface for cut/fill comparison
- On the DroneLink record
- DTM and contours in DXF or LandXML, same datum as the design
Site superintendent
- They need from you
- Current conditions and quantities this week
- You need from them
- Site access, safe take-off area, crane and haul-truck schedule
- On the DroneLink record
- Scheduled flights logged per date in one record
Architect or visualization studio
- They need from you
- Context: terrain, neighbouring buildings, views
- You need from them
- The model's coordinate origin and format
- On the DroneLink record
- Mesh and ortho exported to their tools; see BDM3d Studios for renderings on real site context
Questions
How accurate is drone mapping?
It depends on the method. As general industry practice, RTK or PPK positioning checked against ground control typically delivers a few centimetres; without either, errors are typically decimetres to metres. Vertical accuracy is usually looser than horizontal, and only checkpoints can prove it.
What is ground sample distance?
GSD is the size of one image pixel on the ground, for example 2 cm. Flying lower gives a finer GSD and more detail, but takes more photos and more flight time.
Do I still need ground control points with an RTK drone?
You need fewer, and sometimes none for relative work, but you still need independent checkpoints to prove absolute accuracy. For survey-grade work, a surveyor will usually want some control as well.
What is the difference between a DSM and a DTM?
A DSM includes the top of everything the camera sees: buildings, trees, vehicles. A DTM is the bare ground with those removed, which is what engineers use for design and cut/fill.
Can drone photogrammetry see through trees?
No. Photogrammetry measures what the camera sees, which is the canopy. LiDAR can often reach the ground through gaps in vegetation, which is why it is used for forested and overgrown sites.
What file formats do drone maps come in?
Orthomosaics as GeoTIFF, point clouds as LAS or LAZ, surfaces as GeoTIFF or LandXML, contours as DXF or DWG, meshes as OBJ, and reports as PDF. They open in AutoCAD, Civil 3D, ArcGIS, QGIS and Google Earth.
How long does it take to process a drone map?
Processing time grows with the number of images and the deliverables. A small site can be processed in hours; large sites with point clouds and classification take longer. Ask for the timeline when you order.
Is a drone map a legal survey?
No. A drone map can be part of a surveyor's work, but a legal survey of boundaries must be done and signed by a licensed land surveyor under provincial or state law.
Tell us about your Toronto site.
An address and what you need is enough to start. We reply with what it will take, who will fly it and when.