Glossary
DSM vs DTM (Surface Model vs Terrain Model)
The two elevation products a mapping flight can produce. A DSM is the height of the top of everything, including crop, bush, and buildings. A DTM is the height of the bare ground underneath. Drainage and tile design need the DTM.
Definition
A mapping flight produces two kinds of output: an image you can measure across, and elevation data describing how high the ground is. The elevation side is where the naming gets confused, and the confusion costs real money. A digital surface model, or DSM, records the height of the topmost thing at every location. Where there is standing crop, that is the top of the canopy. Where there is a shelterbelt, that is the top of the trees. Over a bin yard it is the top of the bins, and over a powerline it is whatever the processing made of a wire it could only partly see. A digital terrain model, or DTM, records the height of the bare ground, with everything standing on top of it removed. Digital elevation model, or DEM, is the loose umbrella term and is often used to mean either one, which is exactly why asking which product you are getting matters more than recognising the acronym. The important part is how each one is produced, because they are not two equally available options. Photogrammetry from drone photos natively produces a DSM and nothing else, because a camera can only measure what it can see, and what it can see is the top surface. A DTM has to be derived from that: the processing classifies which measured points are likely to be ground, discards the rest, and interpolates a ground surface underneath the gaps where vegetation or structures were removed. That derivation is reliable where there was not much standing in the way, which means bare soil, short stubble, or thin cover. It becomes an estimate under a dense standing crop and a poor one under heavy bush, because the camera never saw the ground there at all and the software is filling in a surface it has no measurement for. LiDAR behaves differently and better in that respect, since a laser pulse can find gaps in a canopy and return from the ground beneath, which is why bare-earth models under vegetation are a LiDAR strength. The practical consequence for a farm is that the cheapest way to get a trustworthy DTM from a camera is to fly when there is nothing on the field, rather than to pay for processing that tries to reconstruct ground it could not observe. One more product is worth knowing because it is the useful by-product of having both: subtract the DTM from the DSM and what remains is the height of everything standing above the ground, which is a canopy height model. That is how crop height, windrow volume, and tree or shelterbelt dimensions get measured from the air.
In an Alberta Context
The reason this matters on central Alberta ground is that the elevation jobs here are mostly water jobs. Tile drainage, surface drainage, ditching, culvert placement, and working out where water actually goes in a wet spring all depend on the height of the bare ground, which means they depend on a DTM. Run the same analysis on a DSM and a standing crop canopy, a shelterbelt, and a bush margin all behave like dams in the model, so the computed flow routes water around obstacles that are not there and the design comes out wrong in a way that looks entirely plausible on screen. Parkland fields make this worse than open prairie would, because they are ringed and divided by exactly the tall features a surface model cannot distinguish from a hill: bush margins, shelterbelts, creek crossings, slough rings, and yard sites. That points at an obvious scheduling rule, and it is the single most useful thing in this entry. Fly elevation work when the field is bare or in short stubble, which in practice means early spring before seeding or after harvest before freeze-up, since a DSM over bare ground is very nearly a DTM already and almost no reconstruction is required. Booking the same flight over a full canopy in July and asking for a bare-earth model is asking the software to invent the part you are paying for. Accuracy terms are worth separating as well. Vertical accuracy is normally the weaker of the two in photogrammetry, and drainage is the application that cares about it most, since the grades being designed are small. RTK positioning is the baseline requirement rather than an upgrade, and where the model has to tie to a benchmark or to a contractor’s grade control, surveyed ground control points are what make that connection. It is also worth distinguishing relative from absolute accuracy: for deciding where water moves inside one field, consistent relative heights across that field carry most of the value, while absolute accuracy becomes the requirement as soon as a contractor, a neighbour’s project, or a municipal culvert is involved. Satellite elevation data, which does exist and is free, is far too coarse for any of this and is not an alternative.
Why It Matters
Because a supplier can hand over elevation data, call it a DEM, be entirely truthful, and leave you with a product that cannot do the job you bought it for. A DSM delivered for a tile project is the commonest version of that, and it will not announce itself. Contours generated from a surface model look like contours. Slope shading looks like slope shading. A flow accumulation layer produces confident blue lines. All of it is wrong wherever something was standing on the ground, and nothing downstream reveals the error, because the model is internally consistent and only disagrees with reality. The fix is a question asked before the flight rather than an inspection afterwards: am I getting a surface model or a bare-ground terrain model, and if it is a terrain model, how was the ground classified and what was the field condition when it was flown. A good answer names the field condition, because that is what determines whether the DTM was measured or estimated. There is a second reason to understand the distinction, which is that sometimes the DSM is the product you want. Measuring crop height, estimating windrow or pile volume, checking shelterbelt growth, or finding the tall obstacles that constrain a flight path are all surface-model questions, and asking for bare earth in those cases strips out precisely the information of interest. The same flight can yield both, so the right request is usually both plus an explicit statement of which one the design work will be built on. The last point is money. Drainage and tile are among the most expensive things done to a field, they are difficult to reverse once installed, and they are designed against numbers that come off an elevation model nobody re-measures. An error in the base model propagates silently into grade, into tile depth, and into where water is expected to leave, and it surfaces as a wet spot that will not go away rather than as an obvious fault in a file. Getting the right product at the start costs a question. Getting the wrong one costs a season at best.
Frequently Asked Questions
+What is the difference between a DSM and a DTM?
A DSM, or digital surface model, records the height of the topmost thing at every location, so over standing crop it is the top of the canopy and over a shelterbelt it is the top of the trees. A DTM, or digital terrain model, records the height of the bare ground with everything standing on it removed. DEM, digital elevation model, is a loose umbrella term used for either, which is why it is worth asking which product you are actually getting rather than relying on the acronym.
+Which one do I need for tile drainage?
The DTM. Drainage, tile, ditching, and culvert work all depend on the height of the bare ground, because that is the surface water actually runs on. Run the analysis on a DSM instead and the crop canopy, shelterbelts, and bush margins behave like dams in the model, so the computed flow is routed around obstacles that are not part of the ground. The output still looks convincing, which is what makes the mistake expensive.
+Does a drone mapping flight give me a DTM automatically?
No. Photogrammetry from camera imagery natively produces a DSM, because a camera can only measure the surface it can see. A DTM is derived afterwards by classifying which measured points are likely to be ground, discarding the rest, and interpolating a surface underneath the gaps. That derivation is dependable over bare soil or short stubble and becomes an estimate under a dense canopy, so a DTM is a processing step you should confirm is included rather than something that arrives by default.
+When should elevation work be flown?
When the field is bare or in short stubble, which in central Alberta means early spring before seeding or after harvest before freeze-up. A DSM over bare ground is very nearly a DTM already, so almost no reconstruction is needed and the result is measured rather than inferred. Flying the same job over a full canopy in July and asking for bare earth means paying the software to invent the part that matters most.
+Why is LiDAR better than photogrammetry for bare earth?
Because a laser pulse can find gaps in a canopy and return from the ground beneath it, while a camera only ever records the top surface. That gives LiDAR genuine measurements of the ground under vegetation, where photogrammetry has to interpolate across an area it never observed. The practical implication for most farms is not that they need LiDAR, but that the cheap alternative is to fly a camera when there is nothing standing on the field.
+Can I measure crop height from these?
Yes, and it is the useful by-product of having both. Subtract the DTM from the DSM and what remains is the height of everything standing above the ground, which is a canopy height model. The same arithmetic measures windrow or pile dimensions and tracks shelterbelt growth. This is also the case where the DSM is the product you actually want, so asking for bare earth would strip out the information of interest.
+How accurate does the elevation have to be?
More accurate than people expect, because drainage designs small grades and vertical accuracy is normally the weaker side of photogrammetric accuracy. RTK positioning is the baseline rather than an upgrade, and surveyed ground control points are what tie a model to a benchmark or to a contractor’s grade control. It is also worth separating relative from absolute accuracy: consistent heights across one field carry most of the value for deciding where water moves internally, while absolute accuracy becomes necessary as soon as a contractor, a neighbour, or a municipal culvert is involved.
+Can I use free satellite elevation data instead?
Not for design work. Public satellite elevation data exists and is free, but it is produced at scales far coarser than any field drainage or tile plan needs, so it cannot resolve the small grades that decide where water goes within a quarter. It can be useful for understanding broad landscape drainage direction. For anything a contractor will build against, the elevation data has to be flown.
+What should I ask a supplier before booking elevation work?
Three questions. Am I receiving a surface model, a bare-ground terrain model, or both. If a terrain model, how was ground classified and what was the field condition on the day it was flown, since that determines whether the ground was measured or estimated. And what positional accuracy, vertical as well as horizontal, is being delivered, including whether ground control points are used. A good answer names the field condition without being pressed.