A DSM (Digital Surface Model) records the elevation of the topmost thing at every point on a site: tree canopy, roofs, equipment, stockpiles and the ground wherever nothing is standing on it. A DTM (Digital Terrain Model) records the bare ground, with vegetation and structures classified out and the gap beneath them filled by interpolation. Both are produced from the same flight. They answer different questions and they are not interchangeable.
Which one you want
- Measuring something that sits on the ground, such as a stockpile, a waste cell or a building footprint, needs the DSM, because the thing you are measuring is the surface.
- Designing or checking ground, such as grading, roads, drainage, cut and fill or a pad elevation, needs the DTM, because the design is a ground surface and anything standing on it is in the way.
- Volumes usually need both: the DSM for the top of the pile and a ground surface underneath it as the reference, which is the subject of base surface.
- Flood, drainage and runoff work needs the DTM, with the caveat that buildings and berms removed from a terrain model are real obstructions to water and have to be reintroduced as breaklines where they matter.
Where the distinction bites
Three failures come up repeatedly, and all three are the same mistake wearing different clothes.
The first is a contour set generated from a DSM and handed to a designer. The contours follow the top of the grass, the shrubs and the parked equipment, so the grading they produce is wrong by whatever is standing there. The second is a volume computed between a DSM and a DTM that were produced from different flights or different control, where the difference includes the change in method as well as the change in material. The third is a freeboard or crest elevation read off a DSM on a vegetated berm, which reads high because the model is sitting on the grass.
Underneath all three is the fact that a DTM is a filtering decision, not a measurement. Software, or a person, decided which returns were ground and which were not. On bare earth that decision is easy and the terrain model is reliable. Under dense brush, standing crop or heavy canopy there may be no ground observation at all, and the terrain model there is interpolation between the nearest points that were classified as ground. It will still look like a surface.
Photogrammetry and LiDAR do not produce them equally
Photogrammetry measures what the camera can see. Under a closed canopy it sees canopy, so a photogrammetric DTM in a wooded or heavily vegetated area is built from whatever ground was visible in the gaps. LiDAR sends pulses that can reach the ground through gaps in vegetation and record a later return, so it supports a genuine bare-earth model under cover that defeats a camera. That is the practical reason to pay for LiDAR on a vegetated site, and the comparison is set out in LiDAR versus photogrammetry and LiDAR.
On open, bare or lightly vegetated ground, which describes a great deal of Alberta industrial and aggregate work, a photogrammetric terrain model is normally fine and considerably cheaper.
What to ask for
Name the use, not the acronym. Telling a contractor that the surface is going into a grading design, or that it is for measuring a pile, lets them deliver the right product and say which filtering was applied. If you want to be precise in a scope, ask for the DSM, the DTM, and a short note saying how the ground classification was done and where confidence in it is low. The full deliverable vocabulary is covered in drone survey deliverables explained, and which software on your own desktops will open each of them is in who can actually open the drone deliverable.