Intersect, Union... or? (now: how to combine stacked polygons)

[DavidMedeiros]

I have a geo-processing question I need some help with.

I have a student (I'm a GIS assistant/intern at local library) who has a single feature class with multiple overlapping polygons representing distributions by species (so each poly has only one attribute, a single species type). The student would like to output a new layer where each polygon overlap becomes a single individual polygon with attributes reflecting all of the species present in the overlap area. The image below may help conceptualize this (mostly for me!).

What I need are the overlap areas as single new polygons showing the species present from the input polygons (so one new poly each for AC, ABC and BC). Not stacked polygons where the area for say ABC would actually end up as several new polys (AB, BC, AC). From the documentation I can't tell if Union will do this in a single step or if I need to run another process after that to merge the resulting output stacks... or if I need another tool all together?

Any help is appreciated,
Dave

[DanPatterson_Retired]

what did you get running the various options on your sample file?

[DavidMedeiros]

what did you get running the various options on your sample file?

I haven't had a chance to try these in Arc yet, I was given the question just before leaving the library today and don't have Arc at home. I'll do some experimentation tomorrow but thought I would do some research outside of Arc in the mean time.

[DanPatterson_Retired]

start with the online help
http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/An_overview_of_the_Overlay_toolset/000...

[DavidMedeiros]

start with the online help
http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/An_overview_of_the_Overlay_toolset/000...

Yes, that is where I started Dan. And my reading is that either Union or Intersect are the correct tools for me to use here. I originally thought Intersect would not handle more than two polys in an overlap but read in the description that it can do any number of overlaps. My concern is how what happens when you intersect a large number of polygons with varying levels of overlaps within the same feature dataset (all species). What will the output table look like, will I get multiples of the species attribute field for each species at the overlap location.

I just need to run the data to figure this out... was just looking for some dialogue on what tool to use while I'm at home thinking about this.

[MathewSchmidtlein]

If I understand your question correctly, then yes, I believe a single union should do it.  Well, it will generate AC, BC, and ABC, anyway.  You will also have the remainder of A, B, and C, as separate polygons.  But you can go back in and delete those from the resulting output.

If you were to run an intersect, I believe the only output you would get would be ABC.

ms

[DavidMedeiros]

I ran some test data this morning at work, similar looking to the image attached in the first post here. Neither result is what I'm after.

Intersect gives only the overlap areas as output and ignores the remainder of the polygon where there is no overlap. Overlaps are stacked and separate (as Mathew mentioned above).

Union gave me separate polygons for each overlap as well as the non overlap remainders. So where poly A, B and C overlap there are 3 identical stacked polygons clipped to the overlap area: A, B and C. What I'm after is a tool or sequence of tools that will give me a single polygon at each overlap that contains the attribute info for all overlap areas in that stack. So at the A,B & C overlap I would get a single clipped poly that was now ABC. I get the basic functions of intersect and union, but I need to go one step further and am not sure what approach to take.

[DonovanCameron]

How do you merge 2 identical polygons and combine / concatenate attributes

use a relate after a dissolve (before dissolve, attribute table is exported for the relate later).

[DavidMedeiros]

How do you merge 2 identical polygons and combine / concatenate attributes

use a relate after a dissolve (before dissolve, attribute table is exported for the relate later).

I may not be understanding the example clearly but the Dissolve aggregates my separate polygons (if done after a Union) or simply does nothing when applied to my initial sample file and run on the Species field.

Seems like such a simple action, to merge identical stacked polygons, or at least concatenate the species data into the top polygon in the stack. One key piece of info I am still missing (until I interview the student myself) is the exact need for this data. I'm assuming a simple need to visualize the overlap extents.

[DonovanCameron]

I may not be understanding the example clearly but the Dissolve aggregates my separate polygons (if done after a Union) or simply does nothing when applied to my initial sample file and run on the Species field

The dissolve will need to be run so that you are left with a single polygon for each overlap, not stacked polygons. This is only possible if a UID has been generated for the overlapping polygons. One way to do this, is to generate a pt from the Unioned overlay, and dissolve those points based on their X,Y coordinates (matching UIDs in this case) and then create a new field and populate a UID using any method you are familiar with.

in field calculator:

"ABC" & "-" & [FID] + 1

or whatever


Then perform a spatial join to transfer attributes from the centroid pt, to the unioned polygon.

Your GDB should then be ready for the relate.

That archived post does leave out a lot of intermediate steps because the assumption is that the user has some basic working knowledge of ArcGIS as seen in the initial post. Its assumed you know how to eliminate the stacked polygons (identical spatially) into a single polygon. And also how to populate a field to concatenate the values in the related table.


In the last post of the archived forum, download the Zipped mxd and gdb to explore how it was done (hint: explore the GDB when in ArcCatalog to observe the relates, tables, structure, etc...)