I have a DEM which has some individual pixels lower than those around them. This is a problem as I am using the DEM for pluvial flood modelling and therefore it is providing an incorrect representation of the processes occurring. I was wondering if there was a tool to smooth the DEM or whether it is a raster calculator function? I am using ArcMap 10.3.

  • Have you tried Majority Filter or Filter (both require Spatial Analyst)?
    – GISGe
    Commented Jun 23, 2016 at 11:43
  • @GISGe Getting an error message about the input not being in the required domain. Do you know if focal statistics does a similar job? edit: Thank you filter seems to have worked!
    – Z Herring
    Commented Jun 23, 2016 at 11:44
  • 2
    You might want to look at the Fill tool in the hydrology tools if you are trying to extract drainage basins/flow directions.
    – Dùn Caan
    Commented Jun 23, 2016 at 11:45
  • @Dunk I wanted to try and smooth it before I fill it but thank you :)
    – Z Herring
    Commented Jun 23, 2016 at 11:46
  • 1
    @ZHerring, I would have thought that Fill would give you the required smoothing you need, without losing detail?
    – Dùn Caan
    Commented Jun 23, 2016 at 11:49

1 Answer 1


The smoothing is actually a part of every hydrological analysis in gis (and in arcgis as well). The tool you may want to use is fill. This tool fills sinks and remove peaks, adding functionalities such as the z-limit factor. Shortly, z-limit allow to keep sinks / peaks that exceeds the parameter's value.

  • The majority filter is not a good tool for this; it will generally not fill basins--in fact, in some cases it can expand them and make them larger!
    – whuber
    Commented Jun 24, 2016 at 18:00
  • @whuber thanks for the comment. I edited my answer, although I am not quite sure about these specific cases in which majority filter will expand sinks.
    – dof1985
    Commented Jun 24, 2016 at 19:51
  • 2
    When the DEM's values are given to high precision, such as floating-point values or decimeters, it is likely that all elevations within any small neighborhood will be distinct. The majority filter will be useless then. In some cases, by accident, there will be some duplicate elevations in the neighborhood: the majority filter will give arbitrary results. If there is a sink with a flat horizontal bottom, then the majority value within any neighborhood that appreciably overlaps the bottom will be the bottom elevation itself, thereby propagating that elevation outwards, expanding the sink.
    – whuber
    Commented Jun 24, 2016 at 20:46
  • Thank you @whuber, I was going to make this same comment. So I am not just commenting on a comment, I will add that you will likely bring out bias associated with the interpolation algorithm as well. Truncating the float values is very arbitrary and very often dramatically exaggerates contour bias and other interpolation artifacts. Commented Jun 24, 2016 at 21:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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