Remove sections of polygons that do not meet minimum contiguous area requirement?

[ColinMurtoff1]

I am performing parcel analysis that requires the removal of parcel sections that do not meet a minimum contiguous area requirement of 300 square feet (able to have a 300x300ft square pass through it). Is there a way to identify and remove these parcel sections, or "thin strips", that do not meet this requirement? See attached photo for clarification.

[DanPatterson_Retired]

You have described the ….

Inscribed square problem - Wikipedia 

there is no arc* implementation, nor one that could readily be used but

Xander Bakker‌ had a novel approach that you might want to try

[XanderBakker]

Could you share a sample of your data? I wonder if you would do a buffer with a negative distance, followed by a positive buffer and clip the polygon, if this would solve the problem...

[ColinMurtoff1]

Hi, Xander.

This was my initial thought. In an attempt to start this process, I ran the negative buffer on the data layer, but what followed was ERROR 999999.

The data layer does have 1843 features.

[DanPatterson_Retired]

what size of a negative buffer did you run?  You may have collapsed the geometries in on themselves (ie black hole)

Xander is suggesting a relatively small negative buffer to clip off those long skinny bits from your polygons to leave the larger chunks.  So you might want to get an estimate freehand how wide the skinny strips are and negatively buffer by that amount-ish to 'clip' them or 'remove' them from the main chunkl.  Then buffer outward, the file that you buffered inward

[ColinMurtoff1]

Since I am looking to identify contiguous areas of at least 300 square feet, I am negative buffering the polygons by 212 feet (half of the distance from corner to corner of 300x300ft square). I have also tried incrementally buffering (30ft at a time), but get the error 999999 message as well.

[DanPatterson_Retired]

Colin... see the "remove" part of you pic1.jpg.... halfish that width 

to clean off/up those pieces where there isn't a chance in … of a rectangle fitting.  It just makes the process go smoother if the 'shape' is fairly regular and not looking like a crab out of water

[ColinMurtoff1]

Unfortunately I still am getting an error 999999 message when doing the negative buffer of halfish the width of the slim pieces (-25ft). Not too sure where to go from here.

[DanPatterson_Retired]

That error is generic, so that means your inputs. ie the file name and location... could be that it can't create the outputs or your buffer size doesn't match your featureclass units.

screen grab and details time.

And try a negative buffer on a totally different file... say 1 foot on a State or County if the units are in feet

[MarianneRohrbach]

Regarding ERROR 999999: Did you try to run "Repair Geometry" before applying the buffer? This could also be necessary again after each buffer operation especially for inset buffers, since there are often a lot of intersects and multipart geometries.