As per title, I was wondering how in a DEM (i.e., a bare-earth DEM) one can locate depressions that are deeper than the surrounding ground by at least a given value. The main issue is that, quite obviously, a DEM does not store information about the elevation of points relative to the ground, but relative to sea level. Trying to accomplish that entails to devise a way to estimate the height of any cell relative to its neighbours. Maybe Focal Statistics can be put to work, but I wonder if anyone has specific suggestions to put me on the right track.
Thank you.
One way in which to locate depressions is by conducting a stochastic depression analysis (this can be done easily in Whitebox GAT, a free GIS software). It can be downloaded here: http://www.uoguelph.ca/~hydrogeo/Whitebox/. The tool runs using a Monte Carlo simulation to identify topographic sinks in a given DEM and is very easy to use (it does take a lot of processing power and time to run however).
Have you tried terrain rugosity?
Rugosity has been defined in various ways but is typically standard deviation of elevation in a given neighbourhood. There are a number of simple ArcMap plugins that you can download for this I'm sure.
You could also have a quick look at this:
http://gis4geomorphology.com/roughness-topographic-position/
Dear sir,
Thank you for your question. I have a small idea, maybe match your problem. How about creating a point cloud data (DEM to point SHP) and then find the variable that is very low, or, your can load this raw data into SPSS and do some statistical analysis.
Hope this can help you. Thank you.
I am looking forward to your reply.
You could do the following:
1. Convert your raster to a polygon.
2. Spatial join your polygon with itself and set the match option to "intersect". Also calculate the sum of your elevation attribute while doing so.
3. Now subtract the elevation of the focal cell from the sum of elevations using a field calculate.
4. Divide the remaining elevation sum by 8 to obtain the average elevation in the neighborhood of each cell.
Now you are able to calculate the difference in elevation between a focal cell and the average elevation in their neighborhoods. Defining thresholds allows for the identification of depressions.
Hope this helps.
Dear Collegues,
Thank you all for your prompt and useful replies. Much appreciated.
@Jan: thank you for your detailed description, which I will try on a subset of my DTM. As a matter of fact, the DTM I got is quite large and I do not believe my PC can process it.
I have found at least other 2 solution to the problem. The first, which I could not put into practice due to my PC limitations, is the specific ArcGIS 'cut fill' tool (http://desktop.arcgis.com/en/arcmap/10.3/tools/spatial-analyst-toolbox/how-cut-fill-works.htm). The second viable solution was to use the ArcGIS 'fill' tool (http://pro.arcgis.com/en/pro-app/tool-reference/spatial-analyst/how-fill-works.htm), which fill the "sinks" in the DTM. Afterwards, the output of the fill tool can be subtracted from the original DTM. This results in a raster only representing the sinks and their depth value. Of course, since this procedure is very likely to retain small imperfection present in the original DTM, a little bit of polishing is needed at the end, e.g. thresholding the raster on the basis of the area of the sinks and/or their depth.
Best regards,
Gm
I might be kind of late for answering this but For my master thesis I developed a arcpy script that does exactly what you are trying to accomplish. Even more, the maximum depth and minimum size of the depressions is defined by parameters so no post processing is needed. Moreover it detects nested depressions (smaller depressions within larger ones). The output of this fully automated procedure is a polygon shapefile of the depressions rims. Write me if you need more info on this. The procedure is described in thisArticle Use of digital elevation model and RapidEye satellite image ...
but unfortunately it is only available in Slovenian language, abstract and summary are in English too though.
Dear Gianmarco Alberti,
I think that it is too late to answer your question, but it is exist an index that measure the elevation of points relative to the ground directly and nobody mentioned it. This is the Topographic Position Index: http://www.jennessent.com/downloads/tpi_documentation_online.pdf
Maybe it can still be useful.
@Judit:
thanks so much for stopping by and giving your suggestion. Yes, indeed I know the TPI which doesn't allow to do what I was after. As a matter of fact, it is nothing but the standardized difference between a central cell and the average of the surrounding cells' value within a moving window. It has its merit, but does not address my issue.
But thanks for replying.
Cheers
Gm
Hi
About the solution you mentioned (last year!), don't you think that the "fill" tool in ArcGIS is only for covering the "ERRORs" resulted from lack of rsulation in data, etc. and is not going to find real depression of the ground?
I have the same question for doing my research, and wonder if you used "fill" tool finally and was that the right one? (What was the feedback if your work has been reviewed by some experts/reviewers/etc.?)
- Badges
- Science topic
- Similar topics
- Geoscience
- Geoinformatics
- Geographic Information System