2

I have 2 feature class files of contour lines for an area of interest - from 2 different sources. We would like to select just one to use and I'm looking for any methods to effectively compare the files.

I'm using ArcGIS as my primary application to run an analysis.

Wondering if converting the lines to surfaces and comparing the surfaces is appropriate, or comparing values only along the lines itself?

  • 1
    You need to define criteria on which to compare them. A goal function if you will. – lynxlynxlynx Feb 4 '15 at 16:24
  • @lynxlynxlynx I suppose a source to compare the contours to would be a standard surface file/DEM to see how much the lines deviate from the DEM ? – Jennifer Feb 4 '15 at 16:47
  • Sure, that's an option. I was getting at that you need to specify what you want to achieve concretely and how you will measure success. What other data is available is also important, yeah. – lynxlynxlynx Feb 4 '15 at 17:19
  • @lynxlynxlynx using standard deviation to achieve the least deviated line. Whatever source DEM is used, has to be used to compare both source contour files. this will create consistency in the analysis. Spatial statistics could be used to confirm analysis. – Jennifer Feb 4 '15 at 17:49
  • 1
    Values along the line, because type of interpolation greatly affects derived rasters by itself. Also make sure you use the same extent. I'd go first placing points on the lines at fixed interval, e.g. 15 m, if DEM cell size is 10. – FelixIP Feb 4 '15 at 23:42
1

Contours are a model of the surface, a DEM is also a model of the surface, debating the relative inaccuracies of one model vs another seems a bit pointless. Contours don't exactly follow the same elevation on the source DEM and are usually generalized or smoothed for a more cartographic appearance, if you try to compare them in the Z with a DEM both are going to show significant differences - even against the DEM that they are derived from.

Categories to consider:

  1. Contour interval: a 5m contour interval is too much for 1:250k but might be more at home at 1:50k. When mapping land parcels of a few hectares less interval would be appropriate.

  2. Importance: Is the user looking at this map to establish slope/terrain or are the contours simply to give a bit of 'depth' to the product but the focus is primarily on other aspects.

  3. Intended Use: Rounder contours may be less 'accurate' but are nicer to look at which is good for the casual map reader, conversely surveyers and engineers would generally prefer more 'edgy' contours.

  4. Accuracy: Yes, that's important too, but not so much in the Z. Do creeks flow through where they're supposed to? Is the cut/fill on the roads in the right spot? The contours should agree with other map features. If you also have a DEM it's probably best that contours agree (in an X,Y sense).
  5. Source: Contours derived from LiDAR are more likely to be a better product to those derived from mapping or SRTM. If you're in a position where both are generated from LiDAR then choose the latest survey date to minimize temporal change.

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.