Measuring the distance between multipart elements of a feature

[SimonCrutchley]

Is there a way of measuring the distance between multipart elements of a feature? I ask because I've been provided with some dissolved datasets where I've found some parts of what are supposed to be the same feature are kilometres apart. They share most of the same attributes, but there has been a typo in the unique identifier, which has caused the mismatch. I need to find a way to discover the outliers so I can explode, edit and redissolve. If there's an alternative way to find these, that would be equally good, but this was my first thought.

Thanks

[DanPatterson]

Multipart To Singlepart (Data Management)—ArcGIS Pro | Documentation

but that would only allow you to detect those that are indeed completely isolated prior to the dissolve.

re-assembling would be required after with Dissolve once you have fixed the attribute

[SimonCrutchley]

Hi Dan, that's fine. I'm happy to do the dissolve again knowing that at least they're correct. Is there a way to find them automatically? I have tens of thousands, so simply exploding and querying visually won't work, or have I missed something in the link you sent?

[DanPatterson]

That is what I was afraid of.

Feature To Point (Data Management)—ArcGIS Pro | Documentation  

will allow you to get the centroid of multipart polygons, and it will probably not be within the extent of an existing polygon part.  The nice thing is that everything done has an ID value, so what I am thinking is

Generate Origin-Destination Links (Analysis)—ArcGIS Pro | Documentation

with the centroid of the multipart polygon as the "origin" and the singlepart representation of the destinations.  The distance LINK_DIS field could be queried by some threshold distance to identify outliers

A spatial join isn't recommended since you could inadvertantly "assign" one of the singleparts to another multipart polygon because of the distance.

I don't have a dataset to test this on, so I would definitely subsample your thousands to an area that you know you have an issue.  Perhaps remove sensitive data from this subsample and post a zipped gdb (  Feature Class(es) To Geodatabase—  )

[SimonCrutchley]

Thanks for that Dan. I'm trying processing on a subset, and if I have issues I'll copy it to you and you can have a go, but I thought I'd try myself first, as I tend to learn best through actions.

Cheers

[DanPatterson · Accepted solution]

slow day

[SimonCrutchley]

Hi Dan,

Thanks for that. I missed the dissolve step and so joined everything to everything else, which wasn't very helpful 🤔. I'll try this again.

[SimonCrutchley]

Sorry for delayed reply, but I was in a course all yesterday.

This is brilliant and does exactly what I needed 😀.

Thank you.