In ArcGIS 10.1 (advanced license) I have a bare-earth DEM (5m cell size) and I would like to identify depressions which at least, say, 4 meters below the surrounding ground.

What I have tried is to use Focal Statistic to produce a smoothed DEM representing the average elevation within a given moving window, and then subtracting that from the original DEM in order to spot negative areas (=depressions). The problem is that the smoothed DEM so obtained is biased toward high elevations, since (needless to say) them mean is "pulled" toward high elevation values. Therefore, I got not conclusive results.

Edit: I found a solution myself, and I will post the answers in case someone will jump here in the future.

| improve this question | | | | |
  • What happened when you tried that? – PolyGeo Oct 3 '17 at 7:16
  • I did not try yet. I was just wondering if that makes sense. Thank you for replying. Looking forward to comments/suggestions. – NewAtGis Oct 3 '17 at 7:19
  • 1
    I recommend not thinking about GIS SE as being some sort of online GIS tutor or workflow reviewer. For your questions to be answered here they should as much as possible describe not just what you want to do, but precisely what you have tried and where you are stuck trying that. – PolyGeo Oct 3 '17 at 7:45
  • Thank you for the warning. I will rephrase my question to better express my doubts and the issue with the results I get. – NewAtGis Oct 3 '17 at 8:07

I think I found a solution. First, I used the 'Fill' tool (http://desktop.arcgis.com/en/arcmap/10.3/tools/spatial-analyst-toolbox/fill.htm). Then, I took the tool’s output raster and I subtracted it from the original DEM. What I got was a raster representing "pits" (cavities) featuring, needless to say, negative elevation values. Finally, I reclassified the raster to break the values down into some meaninful classes.

| improve this answer | | | | |

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.