How to summarize attributes from overlapping polygons in a raster?

[MirjamBruederle]

I have a feature class with overlapping polygons (and there are multiple overlaps, so on certain patches I have up to 20 “layers”). I want to generate a raster in which each cell contains a summary statistic (average or sum) of one
of the polygon attributes. This summary statistic should be computed from all the polygons that fall within / intersect with the cell, and take into account the proportion of cell area covered by the polygon.

Any ideas how this could be done? Any hint would be highly appreciated!

Anna

[FC_Basson]

For the statistics part you can convert the polygon datasets to rasters with the same cell resolution and use the Cell Statistics tool.  If you want to get the proportion of the raster cell that is covered, you will need to convert your raster to polygon blocks and run an intersect with the polygon layers. What do you want to achieve by getting the proportional raster cell coverage?

[DanPatterson_Retired]

There are some options in this thread in using zonal statistics with overlaps https://community.esri.com/message/595751#comment-595751

Cell statistics as FC Basson mentioned is definitely a viable option with the attentions raised.

Addressing you overlapping polygons in a  meaningful manner via the overlay tools, to maintain the overlap values, but provide a useful classification would be the preferred method prior to introducing the rasters

[MirjamBruederle]

Thank you very much for these suggestions. As far as I can see, converting polygons to raster does not give me the required
control over how cell values are calculated from the overlapping polygons (so the cell statistics tool is not applicable to my problem I think). Also, zonal statistics do not seem to be the solution, because I actually want the summary in the reverse way: not a raster value summary within a polygon, but a polygon attribute summary within a raster cell.

Meanwhile, I thought about the following procedure:

1. run the UNION tool with my single
   feature class of overlapping polygons to discover the overlap patches and separate them into new features;
2. use the DISSOLVE tool (dissolve based on the centroid of the overlapping patches) to generate the mean from all
   the layers for each area where there are overlaps
3. convert to raster.

Does that make sense?