What I am ultimately looking for is a raster file, with a value for each cell stating the minimum distance to any water feature in my study area. For this I am using National Hydrography Dataset shapefile, which originally had thousands of features (ponds,streams,rivers,...). Using the dissolve function, those were split into 13 polygons, while I really wanted them just to be one. I re-ran dissolve on the 13 polygons by a common attribute (which I created, 'dissolve', value=1), but this didn't work either. I decided to just continue with the 13-feature shapefile. enter image description here

I then use the Euclidean distance tool on this file. As output cell size I refer to another raster with a cell size of 985,985 (x,y, meters). This processes quickly, and it's not really what I want as I still need to clip it, but it's a start. Nevertheless, the values already seem off to me (very large). See img 2

I clip the raster to my actual study area, and it looks a lot better! See img 3 Yet, something is off. While clearly distance values are lower for areas with a lot / large water features, smaller features just seem to get 'ignored'. The values of distance are also really high (if they are in fact in meters?), even for cells with water within them (thus where it should be 0). In the example below I 'Identify' the water feature circled in red. See img 4

P.S: This might be enough for a separate question, but it might also have to do with my problem; when checking the distance raster, I notice a discrepancy. When looking at my Raster Properties, cell size (X, Y) = 985,985. Yet when I go to 'Display' and check 'Display raster resolution in ToC', it shows up as 'Res 1: 2,686'. I have no idea how this is even possible!

  • 3
    I think some of the issues is with the cell size which is nearly 1km by 1km. When the Euclidean Distance tool runs I think it internally converts the vectors to a raster and anything smaller than 985m essentially collapses to nothing and is thus excluded from processing. Test this by dropping cell size to say 500m but this will quadruple the raster size, it's a trade off.
    – Hornbydd
    Commented Jul 15, 2021 at 18:00
  • Hmm right, that makes sense! Is there any way to work around this, and not have the vector be converted to a raster? Because ultimately I am interested in how close every pixel is to any water feature, no matter how small. I have just tried it with 250x250 cells and indeed it is already somewhat more accurate.
    – Beardedant
    Commented Jul 15, 2021 at 18:13
  • 1
    Perhaps in this case calculating water feature density per cell makes more sense?
    – Beardedant
    Commented Jul 15, 2021 at 18:41
  • 1
    Try creating a fishnet then run the near tool?
    – Hornbydd
    Commented Jul 15, 2021 at 19:16

2 Answers 2


What projection are you working with? Running distance calcs across three states using a projection that isn't meant for that scale could return some funky results. Be sure you have an appropriate projection and all data created thereafter is in that projection.


Try a finer resolution like hornbydd suggested. Results of the Euclidean Distance tool are depending on the output resolution. When using vector data as input, the first internal step is rasterisation of the vector data. It looks like the method of rasterisation is "Cell Center". If your shape does not match the cell center of output rastercell - it will not be respected in the following distance calculation. The solution is using a finer resolution for the Euclidean Distance tool and resample the result to the desired resolution.

A second solution would be buffering your vector features by more than 71% (half of diagonale) of the desired output resolution. Then use the Euclidean Distance tool and subtract the buffer value of the raster resolution with the raster calculator afterwards.

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.