topic Re: buffering adjacent polygons in Geoprocessing Questions

buffering adjacent polygons

yanma — Mon, 10 Feb 2014 16:09:39 GMT

Hi, I am trying to buffer a shapefile feature comprised of many ploygons that represent different segments of water. What is the best way in ArcGIS to leave polygon interfaces un-buffered? Any input is appreciated.

Re: buffering adjacent polygons

JakeSkinner — Mon, 10 Feb 2014 17:38:02 GMT

You can select only the features that you would like buffered, and then execute the tool. Be sure to use the dropdown to select the input feature. The selection will be honored, and only these features will be buffered.

Re: buffering adjacent polygons

yanma — Mon, 10 Feb 2014 18:09:09 GMT

You can select only the features that you would like buffered, and then execute the tool. Be sure to use the dropdown to select the input feature. The selection will be honored, and only these features will be buffered.

thanks for replying,JS.

So when I select two adjacent polygons and buffer them with OUTSIDE ONLY. I end up with two buffer strips along the shared edge.

Is there a way to leave the shared edge un-buffered?

(To make this less confusing, what I want to achieve is to decide which of the many polygons contain a given location point, after the polygons are buffered. The problem is that after buffering, there is a overlapping region along the interfaces between adjacent polygons. This region is not desired.)

Hope I made myself clearer this time.

thanks again.

Re: buffering adjacent polygons

RichardFairhurst — Mon, 10 Feb 2014 18:48:03 GMT

thanks for replying,JS.

So when I select two adjacent polygons and buffer them with OUTSIDE ONLY. I end up with two buffer strips along the shared edge.

Is there a way to leave the shared edge un-buffered?

(To make this less confusing, what I want to achieve is to decide which of the many polygons contain a given location point, after the polygons are buffered. The problem is that after buffering, there is a overlapping region along the interfaces between adjacent polygons. This region is not desired.)

Hope I made myself clearer this time.

thanks again.

The answer is no. But if you have an Advanced license you could follow up the buffer with the Erase tool using the same input selected polygons and end up with what you want (since you only want outside buffers anyway).

Edit:

I realized you wont get what you want, since you want the shared boundary projected outside of the polygons in a tangent direction to cut off the outside buffer. Only way to do that is to isolate the shared edge somehow and extend it outward to cut with a knowledge of which side the polygons fell on. Not sure how to do that. I have wanted that ability too, but never figured out how to do it.

The Polygon to Line tool could conceivably do part of this and keep track of which polygon is on which side of the share boundary line, but I am not sure how to extend the line the buffer distance and then cut and remove the unwanted portions with the line.

Re: buffering adjacent polygons

yanma — Mon, 10 Feb 2014 20:10:30 GMT

The answer is no. But if you have an Advanced license you could follow up the buffer with the Erase tool using the same input selected polygons and end up with what you want (since you only want outside buffers anyway).

Edit:

I realized you wont get what you want, since you want the shared boundary projected outside of the polygons in a tangent direction to cut off the outside buffer. Only way to do that is to isolate the shared edge somehow and extend it outward to cut with a knowledge of which side the polygons fell on. Not sure how to do that. I have wanted that ability too, but never figured out how to do it.

The Polygon to Line tool could conceivably do part of this and keep track of which polygon is on which side of the share boundary line, but I am not sure how to extend the line the buffer distance and then cut and remove the unwanted portions with the line.

Thanks Richard,

I agree the challenge lies with the portion outside the original polygon. Therefore, I am trying to work out an algorithm that calculates the distance from these outlying points to the nearby polygons and assign to it the attributes of the nearest one.

How sensible does this sound to you?

my

Re: buffering adjacent polygons

KimOllivier — Wed, 12 Feb 2014 08:52:26 GMT

Am I missing the point? There was no diagram so maybe I am.

Why not dissolve on the selected features before buffering? Then the adjacent boundaries will not be buffered.
Are the adjacent polygons not touching? In that case there are other tools that can jump the gap in the Cartographic Toolset.
eg AggregatePolygons Tool

Re: buffering adjacent polygons

yanma — Wed, 12 Feb 2014 13:52:43 GMT

Am I missing the point? There was no diagram so maybe I am.

Why not dissolve on the selected features before buffering? Then the adjacent boundaries will not be buffered.
Are the adjacent polygons not touching? In that case there are other tools that can jump the gap in the Cartographic Toolset.
eg AggregatePolygons Tool

Hi Kim

Thanks for joining in.

For my purpose it is essential to maintain the defining boundary for each polygon in the feature. As every polygon has a unique ID in the attribute table, I feel dissolve will compromise this unique identity, as it is a tool to 'aggregate' features. Am I right?

my

Re: buffering adjacent polygons

RichardFairhurst — Wed, 12 Feb 2014 14:35:41 GMT

Hi Kim

Thanks for joining in.

For my purpose it is essential to maintain the defining boundary for each polygon in the feature. As every polygon has a unique ID in the attribute table, I feel dissolve will compromise this unique identity, as it is a tool to 'aggregate' features. Am I right?

my

Initially I was going to make Kim's suggestion, but realized that Dissolve would compromise the Unique ID, so I did not suggest it.

This problem is the projected jurisdictional coastline problem. Multiple jurisdictions along a coast line want to project their jurisdictional boundaries some distance offshore without overlapping each other. The onshore lands would have the jurisdictional boundaries already defined, so the mainland would never be included in the buffer. Effectively only the coastline is really supposed to be buffered (ideally a continuous line defining the entire coastline buffered using a single side buffer extending offshore only). The division of that buffer would extend the onshore shared jurisdictional boundaries at the coastline offshore in a tangent straight line based on how the onshore boundaries break up the coastline to define the portion of the coastal zone that each jurisdiction controls.

Re: buffering adjacent polygons

RichardFairhurst — Wed, 12 Feb 2014 19:07:46 GMT

I have attached a python file that was exported from Model Builder that created the output shown for a set of Southern California Counties. It creates a non-overlapping external buffer that is 1 mile wide. It requires an Advanced license to use all of the tools I applied in the model.

One picture shows the rather complex coastline where Los Angeles and Orange Counties meet which illustrates how well the model worked.

While Yuma and Imperial Counties do not really have a coastline, I used them to illustrate one problem I resolved. The extension of the internal boundary between these two Counties ran parallel to the buffer boundary. If I had done a normal line extension Yuma County would contain most of the zone that ran in front of Imperial County. To resolve that I contained the interior boundary extension to fit within a buffer only surrounding the interior boundary lines that was at a 3 to 1 ratio with the full outer polygon boundaries. Not the perfect division, but it was acceptable.

The second issue shown that I did not resolve involved an Island of Ventura County where the Centroid of the Island Buffer did not fall within the actual Island polygon. This made it not take on the island's data when I ran the Spatial Join with that join criteria. A special process would have to be added to handle this special situation, but I think with a select by Attribute and a secondary process it could be resolved.

A final problem not shown and not resolved is the issue of two parallel original polygon boundaries that do not touch and that are close enough together that their buffers would merge. This is similar to the problem with internal boundaries being parallel to the buffer boundary, but in reality this does not involve the internal overlapping boundaries. It involves creation of a Centerline between the two parallel boundaries to create a division in the buffers. I don't know if the tools in ArcGIS are good enough to handle that situation and it did not come up with this particular data.

Anyway, I hope this gets you closer to what you want.

Re: buffering adjacent polygons

RichardFairhurst — Wed, 12 Feb 2014 19:29:31 GMT

Here is a screen shot of a more typical buffer division where the internal boundaries are actually perpendicular to the external buffer. In this situation the model worked perfectly. The other post emphasized more problematic situations, but in reality the majority of the buffer divisions the model created actually looked like this illustration.

Re: buffering adjacent polygons

yanma — Thu, 13 Feb 2014 13:09:39 GMT

The second issue shown that I did not resolve involved an Island of Ventura County where the Centroid of the Island Buffer did not fall within the actual Island polygon. This made it not take on the island's data when I ran the Spatial Join with that join criteria. A special process would have to be added to handle this special situation, but I think with a select by Attribute and a secondary process it could be resolved.

hi Richard

I just saw this and am yet to digest your first illustration. But some quick thought on the second ventura island problem. Now that you have the buffer and the island, have you tried other Match_Option in Spatial Join?(I actually do not know how the centroid comes into play. What if you select Completely_Contain? ) Or even use Identity tool to impress the attribute onto the buffer?

It would be great if you could explain the first problem in an easier language. non-native speaker here. Sorry.

Thanks
my

Re: buffering adjacent polygons

RichardFairhurst — Thu, 13 Feb 2014 13:55:11 GMT

hi Richard

I just saw this and am yet to digest your first illustration. But some quick thought on the second ventura island problem. Now that you have the buffer and the island, have you tried other Match_Option in Spatial Join?(I actually do not know how the centroid comes into play. What if you select Completely_Contain? ) Or even use Identity tool to impress the attribute onto the buffer?

It would be great if you could explain the first problem in an easier language. non-native speaker here. Sorry.

Thanks
my

The Spatial Join in my model only used the Contains Centroid match criteria, which is the fastest and produced the best results of the 3 options I tried. Intersect with a negative buffer and Closest both produced bad results by assigning incorrect polygon attributes to the buffer, which was unacceptable. Contains Centriod assigned the correct attributes and only missed the one unusual Island, which was acceptable since it can be followed up with another Spatial Join for just the special cases. I did not feel it necessary to actually do this additional Spatial Join to release the model, but it would be added to my model if I refined it. You would have to add these steps if you encounter the problem of no attributes being assigned using the script I provided.

For the first issue just accept that there was a problem when an interior boundary would extend parallel to the buffer boundary rather than perpendicular to the buffer boundary. My solution, contained in the script, is to create a secondary buffer just for the interior boundary lines to contain this problem to a reasonable result. If I showed the result without this correction the boundary of Yuma County would have been extended as a long narrow strip along the Imperial County frontage until it hit the San Diego County line, which is not at all what I wanted. Follow the script and get it to work as written. After that if you want to better understand the problem I had to deal with you can experiment by removing certain steps which should reveal why I added them.

Anyway, I am happy with the result of my model and other than adding a few more steps to deal with the one failed attribute transfer, I probably would just get the model result and make minor manual corrections rather than try to perfect it through a model. For example, there are some places where snapping the interior lines to the buffer boundary would have been better than extending the interior boundary line. However, it is extremely hard to design a model that can determine which way is better to use at each location and I could visually locate and manually correct these issues far easier than I could write a model to do that for me.

Re: buffering adjacent polygons

yanma — Fri, 07 Mar 2014 13:04:34 GMT

The Spatial Join in my model only used the Contains Centroid match criteria, which is the fastest and produced the best results of the 3 options I tried. Intersect with a negative buffer and Closest both produced bad results by assigning incorrect polygon attributes to the buffer, which was unacceptable. Contains Centriod assigned the correct attributes and only missed the one unusual Island, which was acceptable since it can be followed up with another Spatial Join for just the special cases. I did not feel it necessary to actually do this additional Spatial Join to release the model, but it would be added to my model if I refined it. You would have to add these steps if you encounter the problem of no attributes being assigned using the script I provided.

For the first issue just accept that there was a problem when an interior boundary would extend parallel to the buffer boundary rather than perpendicular to the buffer boundary. My solution, contained in the script, is to create a secondary buffer just for the interior boundary lines to contain this problem to a reasonable result. If I showed the result without this correction the boundary of Yuma County would have been extended as a long narrow strip along the Imperial County frontage until it hit the San Diego County line, which is not at all what I wanted. Follow the script and get it to work as written. After that if you want to better understand the problem I had to deal with you can experiment by removing certain steps which should reveal why I added them.

Anyway, I am happy with the result of my model and other than adding a few more steps to deal with the one failed attribute transfer, I probably would just get the model result and make minor manual corrections rather than try to perfect it through a model. For example, there are some places where snapping the interior lines to the buffer boundary would have been better than extending the interior boundary line. However, it is extremely hard to design a model that can determine which way is better to use at each location and I could visually locate and manually correct these issues far easier than I could write a model to do that for me.

Hi Richard

I finally got chance to carefully look at your python code. And I am trying to reproduce the model based on it. While I keep trying, is it possible for you to upload a snapshot of the model structure in model builder?

Appreciate it.

my

Re: buffering adjacent polygons

RichardFairhurst — Sat, 08 Mar 2014 08:27:45 GMT

Hi Richard

I finally got chance to carefully look at your python code. And I am trying to reproduce the model based on it. While I keep trying, is it possible for you to upload a snapshot of the model structure in model builder?

Appreciate it.

my

Attached are 4 screen shots of the Model builder model structure that was exported to the script. The first or last connected tool or output on each picture overlaps with the last tool/output of the picture that precedes it or the first tool/output of the picture that follows it. Look at the script to see each tool's settings. If you are uncertain about the tool settings and would like a screen shot of the input dialog for any of the tools I used, provide me with the number of the attached picture it appears on and the tool name. I will post a screen shot of each requested tool dialog in response.

Re: buffering adjacent polygons

yanma — Mon, 10 Mar 2014 16:00:41 GMT

Attached are 4 screen shots of the Model builder model structure that was exported to the script. The first or last connected tool or output on each picture overlaps with the last tool/output of the picture that precedes it or the first tool/output of the picture that follows it. Look at the script to see each tool's settings. If you are uncertain about the tool settings and would like a screen shot of the input dialog for any of the tools I used, provide me with the number of the attached picture it appears on and the tool name. I will post a screen shot of each requested tool dialog in response.

Thanks so much.

my