12 replies to this topic

[ktulu]

I have a problem with the dissolve tool of Argcis.

I'm running Arcgis v 9.2. I have a shapefile with municipaltys, this shapefile have a field called province. I have stored the provincial code of each municipalty in this field and I want to make a new shapefile with provinces, so I run the dissolve tool with the field of provincial code checked in the box called "dissolve fields" and the multipart box checked, the new shapefile must contain the provinces. There is one that have a big island and some smaller islands. The problem is that I can't see the big island, it's like if it has disappeared.

Can someone explain me, ¿What has happened?

[MappyB]

I have a problem with the dissolve tool of Argcis.

I'm running Arcgis v 9.2. I have a shapefile with municipaltys, this shapefile have a field called province. I have stored the provincial code of each municipalty in this field and I want to make a new shapefile with provinces, so I run the dissolve tool with the field of provincial code checked in the box called "dissolve fields" and the multipart box checked, the new shapefile must contain the provinces. There is one that have a big island and some smaller islands. The problem is that I can't see the big island, it's like if it has disappeared.

Can someone explain me, ¿What has happened?

This might seem a bit basic, but have you checked the attribute table to make sure that the big island has that attribute filled in? Also, maybe you can check the connectivity of the big island to make sure the polygon is snapped at all vertexes(verti?)/nodes.

[ktulu]

Yes, I have checked the attribute table to make sure that the big island has that attribute filled in. I didn't say that the shapefile of municipalities has multipart entities (one record in the attribute table has some areas), I don't know if this changes something. The shapefile of municipalities is a polygon feature type, so I think that the polygon is snapped at all vertexes. It's not a polilyne.
I hope, I have explain me better.
Thank you for your replies.

[MappyB]

the shapefile of municipalities has multipart entities (one record in the attribute table has some areas), I don't know if this changes something.

This may be the issue; I've never tried to run Dissolve on a .shp that wasn't one area per record. My understanding of Dissolve is that it takes each record and dissolves them in to one based on a specific field. If one record contains more than one area, that might affect the output. Check on the ESRI forums too, although CartoTalk'ers responds much quicker! Sorry I'm not much help!

[ktulu]

Well, I have run the dissolve tool with this type of feature in the past and I had no problem.
Thank you for your replies.

[Hans van der Maarel]

Well, I have run the dissolve tool with this type of feature in the past and I had no problem.
Thank you for your replies.

Could you send me the file by email, so I can have a look?

-- insert standard FME notice here --

[ktulu]

Sorry I can't send you the files.

Thank you for your replies.

[Hans van der Maarel]

Sorry I can't send you the files.

Why not? (naturally all data files I receive are treated with confidentiality and will not be passed on to anybody else or used for anything else)

[DaveB]

Not sure if this will help, but have you tried running Check Geometry on the shapefile?

[ktulu]

Sorry Hans, but I'm trying to get a permission to send you the files. I have run the check geometry tool and the results are:
a lot of records with "incorrect ring ordering" and some records with "null geometry".

I understand that "null geometry" is a record in the atribute table with no area(s) related, isn't it? , but, what means "incorrect ring ordering"?

Thank you.

[Hans van der Maarel]

Sorry Hans, but I'm trying to get a permission to send you the files. I have run the check geometry tool and the results are:
a lot of records with "incorrect ring ordering" and some records with "null geometry".

I understand that "null geometry" is a record in the atribute table with no area(s) related, isn't it? , but, what means "incorrect ring ordering"?

I'm no ESRI user, but "incorrect ring ordering" sounds like there's something not in order with the donut polygons...

[ktulu]

I looked for "incorrect ring ordering" in google and I founded an esri web page with the solution.

Thank you DaveB, I have founded the solution with your post.

Thank you Hans, your replies were useful too.

Edited by ktulu, 26 September 2007 - 03:53 AM.

[Charlie Frye]

Just to explain why this might have happened:

The Shapefile format is public, which means many programmers have the information they need to write shapefiles. However it doesn't mean they will do so properly.

Ring order could mean one of two things. First, when a shape has more than one part, e.g., a donut hole or an island, it's shape will be comprised of multiple parts, or rings. Each ring defines the geometry (list of coordinates). The expected implementation is that the rings are listed in size order, with the largest ring (covering the most area) being first in the list. Another issue is whether the coordinates in each ring are listed in clockwise or counterclockwise order; clockwise means it is an exterior ring or an island, and counterclockwise is an interior ring or an island.

Back when we made the Shapefile format public we had a lot of problems with mal-formed shapes. Beyond better documentation/education we also enhanced the shapefile export tools to repair obvious errors. So for the most part all one needed to do to was write the shapefile again with ArcView GIS. In ArcGIS we have the Repair Geometry tool to serve the same purpose.