Select to view content in your preferred language

# Divide polygone by calculated area ratio

640
4
02-24-2014 07:35 AM
New Contributor
Hello there,

I am pretty new to ArcMap10.1 but so far I haven´t found a proper solution to my problem in the Forum.

The main topic is tree crown delineation. I am using airborne data to create a shape file for tree crowns. It works really well for single trees but if the corwns are touching I only get a big "chunky" polygon. What I did so far after I created the polygon: By calculating area, perimter and radius of each polygone I then calculated an index (Rindex=perimeter*radius/area). For a nomral circle this ratio is 2. So if I have a big polygone with Rindex = 6, I now know, that there are approx. 3 trees inside (--> Nr_Trees = Rindex/2).

Now my actual question: How can I divide my polygons by the calculated Nr_Trees (see attached screen shot)?

If you need more data to help me solve my problem please let me know! Thanks in advance for any suggestions.
Tags (2)
4 Replies
MVP Regular Contributor
Perhaps convert the polygons to raster, with the cellsize set to your normal circle radius. Convert this from raster to points, and you have points inside the polygon representing each tree crown.
Best Regards,
Jim
New Contributor
Hello Jim,

thanks for your fast reply! If I understand you correctly I can only set one number for the cell size, correct? The problem is, that the tree crown radius varies from tree to tree. That�??s why I tried to introduce the ratio between area, perimeter and radius. If I generalize the Raster with only one radius I either get too many or too few trees... Or did I misunderstand your suggestions?

Best regards,
Anne
MVP Regular Contributor
No misunderstanding. I wasn't aware you were working with multiple types.
Could you cluster by groves? Otherwise, it seems that you will have to divide by hand and visual interpretation.
Or am I missing something that allows you to determine tree type / crown radius in an automated fashion?
Regards,
Jim
New Contributor
Bummer 😞 Solving my problem manually won´t help since I am looking for a way to automate it. My test area at the moment is small but my main model is about 15km long. I cannot distinguish tree types from the given airborne data. My results will be approximated values anyway but I want to get as close to nature as possible.
I haven´t worked with clusters yet, thanks for suggesting it!