I want to use a high-resolution classification (3cm) as training data for a classification using Sentinel 2 data (10m).

To do so, I want find which high-res value is most common within each Sentinel "cell". I tried to do this by using Zonal Statistics with a Fishnet Grid based on the Sentinel data (Zonal Statistics has an option called Majority to find this most common value) but this did not work satisfactorily since I got some wrong results (see my other question at Zonal Statistics and output cell size)

What is another way to go about this?

1 Answer 1


You should be able to do this by resampling the 3cm raster to 20m using the "majority" method in ArcGIS Resample. See https://pro.arcgis.com/en/pro-app/2.7/tool-reference/data-management/resample.htm.

Then your can extract Sentinel-2 and the resampled pixels to generate your model predictor and response data.

  • Thanks for this tip! However, I tried using Resample and got incorrect results. As in the question that I linked, some cells which only have a small percentage of a certain value are still assigned that value as the most common value. Could it have something to do with using Polygon to Raster first? Is this known to somehow corrupt data or something like that?
    – stijnionio
    Apr 25, 2022 at 7:46

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.