# Two different cell sizes represented in a raster

I am currently working with a DEM with a cell size of 10 meters. I only needed a certain area of this DEM and wanted it at a 1 m cell size so I extracted by mask. I set the appropriate environmental settings to obtain a cell size of 1 m and my mask layer was also set to 1 m cell size (same coordinate system as well).

My output does indeed have a cell size of 1 m, but you can see from the following image two different sized cells. Along the edges measure 1 m, but the cells in the middle measure about 8 meters (this is the cell size of the original DEM before I clipped it).

I'm wondering if my clipped DEM does in fact have a 1 m cell size or is something wrong here?

The point of setting my cell size at 1m was to obtain a smooth look. Possibly those larger cells really are 8 1m cells, all of the same values? I have tried resampling with bilinear interpolation and I get the exact same result.

I am using ArcGIS 10.5.1 Version 10.5.1.7333.

The most similar question I have been able to find, but which did not seem to have a final solution/one that solves my particular problem is Cell size (10m) in Raster properties differs from measured cell size (20m) in multiband (Sentinel) raster?

• Like you said, they are just 8x8 blocks of 1m cells. There isn't really a way to "smooth" a raster to a finer resolution. – Jacob F Mar 20 '18 at 18:35
• @JacobF Other rasters I am working with are made up completely of 1 m cells, giving it that smooth look, instead of grouping the cells into 8x8. Not sure why this raster is grouping them. – Kristen Mar 20 '18 at 18:46
• I should not have said there is not a way to smooth a raster. There are plenty of ways using focal statistics and other methods. But when resampling to a finer resolution, ArcMap defaults to nearest neighbor. See the second note here: desktop.arcgis.com/en/arcmap/10.5/extensions/spatial-analyst/… If you want the resampled raster to look smoother you can try the cubic convolution method in the resample tool. – Jacob F Mar 20 '18 at 19:33
• It's not "grouping" values. It's just larger pixels of a value that have been cut up into smaller pixels of the same value. Changing the pixel size doesn't magically make higher resolution data out of lower resolution. You need to resample or interpolate. – user2856 Mar 25 '18 at 7:11