I'm comparing the slope products derived from DEMs of various spatial resolutions (1m, 10m, 30m) for a set area.

The trend I find is that as the resolution improves, the areal extent of various extreme slope breaks (>45%, >50%, >70%) all increase. The areas increase by so much in fact, that it makes me question the reliability of anything I would produce from the 30m DEM - beyond using as just a general reference or guide. I'm inclined to believe that better resolution equates to a better answer, but know that isn't always the case.

Is there a simple enough explanation for this pattern I see?

Or perhaps 1m is overkill for this? Could noise be a big factor?


Measuring the elevation along a transect is a bit like measuring the length of a coastline. If you change the resolution, you change the final result, but none of the result is relevant if you don't mention the scale of the measurement. If you look at a portion of soil with a binocular, every grain of sand will be a cliff. If you fly high on a plane, only mountains are important.

Let's take two examples with possible uses o the DEM:

For hydrological model, you might need a 1 m resolution to follow the runoff of water and all the small obstacle. However, you will have plenty of very small depression that actually fill in very quickly, so after a short time it is like all the slopes "disappeared". some advanced models handle the dynamic filling of sinks, but most of the time you will first remove the small sinks and make everything "flatter" before running or more simple model. For erosion, the slopes that are not part of a sink are the ones that matter.

For animal (or human) movement, you could look at the very steep slopes that are due to the roughness of the terrain (e.g. walking on pebbles beach), but in fact the slope that matters is the one between two steps. So the resolution depends if you are studying ants or kangaroos.

| improve this answer | |
  • 1
    Great answer. Another consideration is to not over-classify slopes in to a lot of classes, because, for example, the "travel-ability" of a route depends on too many other factors, of which slope is just one trend. – danak Sep 25 '19 at 16:22
  • @radouxju "So the resolution depends if you are studying ants or kangaroos". This is in line with what I was thinking - that 1m may be just too fine for the purposes of what I am studying. Assuming that slope is more of a relative concept, is it fair to say that one cannot assume one resolution is better than another in regards to slope - without first defining a use-case? – tordor Sep 25 '19 at 21:29
  • yes, defining use case if necessary to identify the optimal scale for slope. – radouxju Sep 26 '19 at 7:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.