2

I want to extract averages from raster data that fall within buffer zones around a village point.

The problem is that the villages are pretty close together in some places and I need to come up with a methodology to group villages and their respective buffers to create groups.

Then I want to look at averages in these zones using the zonal statistics tool.

I am open to another way of doing this - any suggestions welcome!

Improve this question

1 Answer 1

Reset to default
0

You need to decide how you are going to group the villages. Assuming you want to do a spatial grouping here is a work flow suggestion.

  1. To spatially group the villages use the Grouping Analysis tool. You can specify how many groups you want and how you want the grouping algorithm to work. The idea is that villages near each other will be grouped together. The output is a new feature layer with a new code showing the grouping. To get the tool to work with only a spatial dimension I added the XY coords to the point layers attribute data and then used those fields as the fields for the Grouping Analysis tool to use. I could then select No_Spatial_Constraint as the spatial element was already represented within the XY coordinates. You can experiment with number of groups to see what is best.

enter image description here

  1. When you are happy with the grouping you can then use the buffer tool. Here you can specify the buffer distance and most importantly can disssolve the buffers into groups using the newly created grouping code you produced in Step 1. Select "list" as "Dissolve Type" and check the field containing the new code.
  2. Finally you can use the Zonal Statistics as Table tool to summarise your underlying raster using the buffer zones as a template. If your polygon buffer areas overlap then you will need the Spatial Analyst Supplemental Tools extension which can be downloaded here http://www.arcgis.com/home/item.html?id=3528bd72847c439f88190a137a1d0e67. Included is a tool called Zonal Statistics as Table 2 which can cope with overlapping polygons.
Improve this answer
7
  • Thank you David! I tried grouping analysis, but I couldn't get the grouping analysis to work correctly. It was selecting villages that were not necessarily next to each other spatially. Could you elaborate on the specifics on the grouping analysis? Delaunay triangulation or K nearest neighbors? How many groups? What do I put for the analysis field if I just want them to be grouped spatially? Etc. Thank you!!!
    – user32636
    Commented Nov 13, 2015 at 15:14
  • I have edited my answer. The grouping is not easy to do and I had some issues too! In the end the only way I could see to do it automatically in ArcMap was to add the XY coords as shown above. If you cannot get a grouping you are happy with you can always just group manually by adding a field to the villages attribute table and typing in a group number. Then you move onto step 2.
    – David W
    Commented Nov 13, 2015 at 16:24
  • This is awesome! Thanks! My last question though is that when you get to step 3, some of the dissolved buffers overlap, and Im pretty sure zonal statistics cant handle that calculation, right? I've tried using this before with overlapping buffers and it just calculates portions rather than than the whole thing! Thank you! Almost there!
    – user32636
    Commented Nov 13, 2015 at 19:52
  • Yes, zonal statistics can't cope with overlapping zones. The solution suggested in the help is to process the zonal statistics iteratively for each of the polygon zones and collate the results. (There are python solutions but I assume that's not an option for you.) So it looks like you will have to split up your polygons into non-overlaping groups and do zonal stats on each group. The Multipart to singlepart tool can help with the split. Hope it works out - you are nearly there!
    – David W
    Commented Nov 13, 2015 at 20:34
  • What about this? arcgis.com/home/item.html?id=b859b33c616a47d2b99b5e133942db02
    – user32636
    Commented Nov 14, 2015 at 0:15

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.