I am working in ArcGIS Desktop 10.5 and I am a beginner.

I have very basic, non-overlapping polygons over a raster that represents bathymetry data. I need to figure out various statistics about each polygon (like range) so I figured I would use the Zonal Statistics tool. I also needed the area of the features. So I did this and things were fine.

But then I happened to notice that the areas I got for my polygons were different from the "Shape_area" calculations that I see when I use the "identify" tool (the one in a blue box with an 'i' on it) and click on the polygon.

Which area is correct?

  • Zonal statistics convert polygons to raster. If you want to reduce difference between actual area (polygon) and output of zonal statistics, use smaller cell size. – FelixIP Apr 23 '19 at 20:12
  • It does not say that it does that. It says zonal statistics calculates statistics on values of a raster within the zones of another dataset. There is no "conversion" from what I can tell. How would I even do what you suggest (use smaller cell sizes)? What is correct, the "shape_area" that I calculated with the identify tool or what I get from the zonal statistic value? Can I trust the other zonal statistics that are provided (like mean)? – David Weinstein Apr 24 '19 at 17:09
  • Use identify tool. See environment settings, raster cell size if you want to reduce it. – FelixIP Apr 24 '19 at 19:06

The software isn't splitting the cells, it's counting whole cells whenever the majority of a cell falls within the zone defined by the polygon. So your areas are all going to be multiples of the cell area. It's the nature of raster analysis.

  • Thanks so much for your information. That makes a lot of sense. So then it is not nearly as accurate an interpretation of the area as the "identity" tool and I should just use that? I still need to use the zonal statistics to determine information about the bathymetry under the polygons (can not use the "identity" tool for that) though. Is there a problem with using the area calculation from one method and using the zonal statistics for another? – David Weinstein May 1 '19 at 4:29
  • I don't see a problem with that. Or if you have more than a few polygons, it will be less work to just work from the attribute table and export what you need. You should be able to join the attribute table of the zonal stats grid to the attribute table of the polygons. – Jersey Andy May 2 '19 at 13:59

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.