I want to use zonal statistics on Chinese counties (which can be very small) from a fairly coarse (0.5 degree) raster grid. The problem with zonal statistics is that it returns a much smaller number of counties after the feature-to-raster transformation. This means a lot of my counties won't be assigned any data.

Can this be overcome somehow?

For example, I would like for example to have a list of raster cells that lie within each county, along with the cell's value and the share of the cell that lies within the county. This way, I would be able to construct a nice measure.

  • Would you be able to do a raster to polygon conversion? If so then you could run an overlay of your county polygons with the raster's polygons.
    – John
    Feb 5, 2013 at 13:38

1 Answer 1


Sounds like the problem is in the Feature to Raster transformation. Try changing the output cell size from the transformation to a finer resolution than the default. For example, the below image shows two outputs from the Feature to Raster tool (The input feature is displayed as hollow red). The default cell size is shown on the left and a fine resolution raster is shown on the right. This way, all of your counties should be accounted for, rather than being pixelated-out.

I see two alternatives:

  1. Do a vector-based analysis using any of the overlay tools.
  2. Use Geospatial Modelling Environments (GME) isectpolypoly (Intersect Polygons With Polygons) which is essentially zonal statistics for feature data (Free from GME).

enter image description here

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.