Area Calculation using Summarize Within

[KristalWalsh]

I'm working in ArcGIS Pro v3.3, using Model Builder to create a workflow where one branch summarizes the acres of forested landcover classes within the AOI. The landcover feature class includes a definition query which is accurately reflected in the process and the count of polygons seem to be correct on a test AOI. But to check/compare acres (that my output was correct), I clipped the landcovers using the same boundary and came up with ~1400 more acres than the Summarize Within calculation. Was this the totally wrong move? I can't imagine there would be this big of a difference! Am I completely misunderstanding the summarize tool? Thank you for your help on this - need to move past this step. 

[DanPatterson]

What coordinate system are the data and map in?

Summarize Within (Analysis)—ArcGIS Pro | Documentation

The tool supports the Output Coordinate System environment on the Environments tab, so make sure you specify a planar one (such as UTM or State Plane etc) anything else but Web Mercator

[KristalWalsh]

Hi Dan, thanks for responding. My data is all in NAD 1983 (2011) Florida GDL Albers (Meters). Even so, I didn't have a target number for comparison to my result.  It is the variance between the tool outputs that is puzzling.  After some internal discussion I think that my problem stems from the difference in the way Clip and Summarize Within "selects" the resulting features.  Because the "summarize" tool uses intersect, it may be dropping parts of the features inside the AOI. I'm going to test this a little further I think. A difference of a 1000 acres could mean ineligibility for program applicants so I need to make sure I choose the right tool. 

[DanPatterson]

I would make sure that you are working with single part polygons and not multipart.  The tool will calculate a shape area of the overlap, but it would be prudent to check that the calculated value isn't being ascribed to its multipart original.  Comparing to a clip is certainly the way to compare

[KristalWalsh]

So, therein potentially lies the issue. In a previous step of the model workflow, the option to create multipart features (as part of the pairwise dissolve tool) is checked - the default. If unchecked, according to the information in the tool, individual features will be created. I wanted a one feature output so that I could have the model check that the entire AOI met the minimum area criteria in the next step. Thank you for this discussion! I am trying it outside the model just to see if that is the problem. 

[DanPatterson]

post on this thread if multiparts were indeed the issue

[KristalWalsh]

Back to the drawing board, Dan. I just ran another property using 'clip' and 'summarize within' on the single part feature. I got a difference of about 350 acres between the two, 'clip' had the larger output. I dissolved the feature and ran the summarize within tool again and got the same number as when using  'summarize within' on the single part. So it didn't make any difference with this example.  I understand the outputs are not derived in the same way but without being able to see what polygons are left out of the intersection, it's hard to make a decision on which tool is the correct tool to use. Will break from this tonight. Thank you for the suggestions today. 

[DanPatterson]

Ok... this is bugging me, so make your own sense out of the following example

• two simple polygon featureclasses, locally stored
• projected coordinate system (MTM zone 9 Canada), planar units
• I did summarize with and,
• a clip of the overlapping areas
• examine the tables at your leisure
• compare to a case in your files

[KristalWalsh]

I don't know what to say here. I found one small variation in datum in the coordinate system, projected to make sure they were the same and am coming up with the same result as before. There are many polygons within an AOI that need to be summed, but the feature count is consistently the same between tools. 

[DanPatterson]

Tech support would be your best bet since nothing I have suggest has narrowed down the issue