Jump to content

 
Photo

Consolidating two roads layers from different sources

- - - - -

  • Please log in to reply
5 replies to this topic

#1
johnnyh

johnnyh

    Contributor

  • Validated Member
  • PipPip
  • 43 posts
  • United States

I've got 2 roads layers from 2 different levels of govt that do not match at all.

*The state layer seems to be more accurate and has much more complete ROAD NAMES attributes.
*The county layer looks to be inferior, BUT has a more complete ROAD NUMBERs attributes. The county roads also has many numbered roads the state does not.

At least half of the same features (by NAME or NUMBER) seem to not be identical. I can find no clues in the attributes or metadata to help me.

I've got over a million acres to do, and would really dislike having to do this 100% manually... Any ideas?! :)

In this image, you can see the State roads (blue w/ black label) are much more accurate and have complete road names.

Posted Image

In this sample you can see that the forest roads are mostly identical, but the County roads (red label) have complete road numbers.

Posted Image

#2
Charles Syrett

Charles Syrett

    Ultimate Contributor

  • Validated Member
  • PipPipPipPipPipPip
  • 536 posts
  • Canada

I'm not sure what the problem is here. Data sets from different sources are usually different, in my experience. People like us (cartos, that is) are hired to sort through it all and come up with reliable mapping. If anything, I would encourage you to find yet another data set or two, do more comparisons -- and then the fun begins! By all means glance at the metadata and attributes, but most of the work is just comparison, looking at photography (be mindful of the date!), solving all the puzzles, and then creating your own map. B)

Charles Syrett
Map Graphics
http://www.mapgraphics.com

#3
johnnyh

johnnyh

    Contributor

  • Validated Member
  • PipPip
  • 43 posts
  • United States

True Charles and thanks... I guess I'm thinking about getting a good workable layer first and a map(s) second. This might be more appropriate for the ESRI forum now that I think about it. Mostly I was hoping somebody had some smart GIS shortcuts.

For example:
select "shares line segment" in County (this will give me most of those ROAD NUMBERS)
>select from selection "ROAD NUM <> 0" (this is all the info I want from the County layer really)

selected in State "shares line segment" with selected from County and DELETE

Export selection from County into new layer and UNION with State. Then clean up attributes and meta...

I wonder if I should explode multi-part features in both layers and use the feature "is identical to" rather than the "shares line segment"

#4
jrat

jrat

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 143 posts
  • Gender:Male
  • Location:Hagerstown, MD
  • United States

Even if you explode to multi-part you might still have issues using "identical to". The center lines might be broken at different intervals, hence not identical. If you like the state geometry but want the county attributes I would reccomend a spatial join. Most are close enough that they will match right. This will give you state geometry and attributes with county attributes added. compare the road names in the two different colums to find areas that have the wrong join. Then start getting rid of what you don't want.

#5
ceicher

ceicher

    Contributor

  • Validated Member
  • PipPip
  • 44 posts
  • Gender:Male
  • Location:Charlottesville, Virginia
  • United States

Conflation is Not for Wimps

re: "I'm thinking about getting a good workable layer first and a map(s) second. "

This is _definitely_ the right approach in my opinion. "Correct data make correct maps", or something like that.

The general problem you have here is (or is solved by?) "conflation". In GIS it applies very often to the exact problem you have. How to combine attributes from one set of features and geometries from another set of features.

This is not an easy problem to solve. Quite a few people have put quite a lot of time and money into building intelligent solutions. Google "gis data conflation" to see for yourself. I found this one interesting, at least the title: http://proceedings.e.../conflation.pdf (from http://proceedings.e...005/index.html)

Hope this helps,

-Cory

#6
Charles Syrett

Charles Syrett

    Ultimate Contributor

  • Validated Member
  • PipPipPipPipPipPip
  • 536 posts
  • Canada

Conflation is Not for Wimps
re: "I'm thinking about getting a good workable layer first and a map(s) second. "
This is _definitely_ the right approach in my opinion. "Correct data make correct maps", or something like that.


Well put. Something like "good ingredients make good recipes". This is the first time I've heard of "conflation", and it's good to know that some GIS people do this to create better data sets. So many times, I've seen features duplicated on things like Google Maps and Tiger. You know -- a building, or some streets, actually appearing twice and slightly displaced from each other. :unsure:

Also noteworthy is that that the kind of conflation described in that document takes "only six months" for certain parts of the process. In the rough-and-ready graphics-oriented commercial mapping my company does, we do a kind of "lite" version of conflation, simply by importing shapefiles from various sources, and then scavenging the best parts of each.

Charles Syrett
Map Graphics
http://www.mapgraphics.com




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->