Glossary
GSD (Ground Sample Distance)
The real-world size of one pixel in a drone map, usually quoted in centimetres per pixel. It is set by the camera and the height flown, and it decides what is small enough to be seen at all.
Definition
Ground sample distance is the distance on the ground covered by a single pixel of an aerial image. A map quoted at one centimetre per pixel means each pixel represents a one centimetre square of field, so a two centimetre object spans about two pixels and a thirty centimetre object spans about thirty. It comes out of simple optics: the physical size of a pixel on the camera sensor, multiplied by the height flown above the ground, divided by the lens focal length. Two of those three are fixed once the aircraft is chosen, which leaves flight height as the control the pilot actually has. Fly lower and the pixels get finer, but the camera sees less ground per frame, so the same field needs more passes, more images, more battery, and more processing time. Fly higher and the field is covered quickly at coarser detail. That trade is the entire practical content of the term. The critical thing GSD is not is accuracy. Resolution is how finely the world is sampled; accuracy is whether the sample sits in the right place. A map can be quoted at a very fine centimetres-per-pixel figure and still be shifted, tilted, or sitting at the wrong elevation, which is a separate problem solved by RTK positioning and, where it matters, surveyed control points. Treat the two as independent numbers, because a supplier quoting only the resolution has told you nothing about the accuracy.
In an Alberta Context
For Alberta field work the useful question is never how fine the resolution can be, it is how fine it has to be for the decision at hand, because every step finer costs flight time. Whole-field vigour mapping and NDVI scouting sit at the forgiving end: the job is to find the shape and extent of an area performing differently from its neighbours, and a pattern that covers part of a quarter is still obvious in a coarse map. Flying that job low enough to resolve individual plants buys nothing and can turn a single flight into several. The demanding end is anything counted or identified rather than located. Stand assessment, early weed patches before a canopy closes, and telling one small feature from another all need the feature to span several pixels rather than one, because a target the size of a single pixel disappears into whatever is around it. Elevation work for drainage and tile planning is the case that gets confused most often, because there the deliverable depends on vertical accuracy rather than on resolution: a drainage model needs to know the height of the ground correctly more than it needs a fine image, and that comes from positioning and control, not from flying lower. Alberta adds one more constraint from the other direction. Flight height is capped by the Transport Canada 400 foot above-ground limit, and practical mapping heights are usually well under that, so the ceiling on how coarse a single pass can be is a rule rather than a choice.
Why It Matters
Because resolution is the number most often quoted in a proposal and the one least often tied to a decision. A finer GSD sounds unambiguously better and is not free: it multiplies flight lines, images, battery changes, processing hours, and file sizes, and on a windy afternoon it can be the difference between finishing a field and coming back another day. Paying for detail nobody uses is a real cost with no return. The failure runs the other way too. A map flown too coarse for the job cannot be rescued in processing, because the detail was never captured, and the only fix is flying the field again. Knowing what the map is for before it is flown is what prevents both mistakes, and the question that settles it is simple: what is the smallest thing this map has to show, and does it need to be spotted or identified? Spotting a patch is easy; identifying what is in it is not. The last reason it matters is that fine resolution creates confidence a map may not have earned. A crisp image invites the assumption that everything about it is precise, including where it sits and how high the ground is, and those are governed by positioning and control rather than by pixel size. When a map is going to drive an excavation or line up against another dataset, the accuracy question is the one worth asking, and it is not answered by the resolution figure.
Frequently Asked Questions
+What does ground sample distance mean?
It is the real-world size of one pixel in an aerial image, normally quoted in centimetres per pixel. At one centimetre per pixel each pixel covers a one centimetre square of ground, so a thirty centimetre object spans roughly thirty pixels. It is the practical measure of how much detail a drone map contains.
+What determines the GSD of a drone map?
The physical size of a pixel on the camera sensor, the lens focal length, and the height flown above the ground. The first two are fixed once the aircraft and camera are chosen, so in the field the only variable the pilot controls is height. Lower flying produces finer pixels and covers less ground per image; higher flying covers the field faster at coarser detail.
+Is a smaller GSD always better?
No, because it is not free. Halving the ground size of a pixel means flying lower, which means more flight lines, more images, more battery changes, and more processing for the same field. That is worth paying for when the decision depends on small detail and wasted when it does not. The right resolution is the coarsest one that still answers the question being asked.
+What GSD do I need for NDVI crop scouting?
Less than people expect. Vigour mapping is about locating the shape and extent of an area performing differently from the rest of the field, and a pattern covering part of a quarter is plainly visible in a coarse map. Flying low enough to resolve individual plants adds flight time without improving the decision, since the follow-up step is walking the flagged area anyway.
+Does a finer GSD make a map more accurate?
No, and conflating the two is a common and expensive mistake. Resolution is how finely the ground is sampled; accuracy is whether the result sits in the right place and at the right elevation. A very fine map can still be shifted, tilted, or low, which is a positioning problem solved by RTK and, on work that demands it, surveyed control points. They are independent figures and both are worth asking about.
+What resolution does a drainage or tile survey need?
Resolution is the wrong lever for that job. A drainage model is built on the height of the ground rather than on image detail, so vertical accuracy is what governs whether the design drains. That comes from RTK positioning and, where the work has to agree with a benchmark or a contractor’s grade control, from surveyed ground control points. A finer image with poor vertical control still produces a plan that looks right and runs the wrong way.