Summary of Mailing List Correspondence
Thread: Parcel Data in OSM?
The original OSM-talk thread started by Christopher Schmidt on Feb 17, 2009 can be viewed here. In short, Chris proposed the possibility of uploading parcel boundaries and attributes from [MassGIS]. The thread contains enthusiasm for the idea along with concerns about the quantity of data and how updates would be managed. A question about which zoom levels (if any) would be appropriate for rendering parcel boundaries. From the [MassGIS] page, it appears that building outlines and open space ended up making it into the map, but there is no indication that parcel boundaries made it.
Thread: Parcel data
The original OSM-talk thread started by Anthony on Sept. 21, 2009 can be viewed here. Anthony proposes uploading the parcel boundaries from a public dataset that he acquired from his local assessor in Florida along with some discussion about a tagging scheme. He mentions particular interest in the address information. Frederick Ramm encouraged proper resolution of overlapping boundaries. Pieren mentioned that he and fellow users in France decided not to upload the parcel boundaries because they change frequently, are not verifiable, and may not be valuable to the OSM community. A discussion proceeded around other import best practices, tagging practices, updating pitfalls, and an offer for collaboration.
Thread: Importing Arkansas data
The original imports thread started by Al Pascual on Mar. 21, 2011 can be viewed here. Al announced his intention to use a custom ArcGIS tool to import state-wide parcel data for Arkansas, and then proceeded with the automated import. Others expressed concerns that this import had not been properly discussed. After a few days of importing, Al's account was blocked to halt the import. He subsequently deleted the parcel boundaries. In the aftermath of this, the discussion broadened to question the importing techniques used by Al and a debate regarding the value of parcel data to the OSM community. The main points regarding the latter included concerns that:
- Are parcel boundaries useful to OSM?
- Can parcel data possibly be kept up to date considering how frequently it changes?
- Does parcel data meet the verifiability criteria?
- Parcel data is too big and clogs up editors
- Continual imports of parcel data (and other data) undermine OSM as a human-editable map
Thread: Fresno castradal [sic] imports
The original talk-us and import threads initiated by Paul Norman on April 26 2012 can be viewed here and here respectively. Paul mentions a number of problems with the Fresno cadastral import that was performed in 2010 by Nathan Mixter. He cites a consensus against "dumping" cadastral data in OSM and proposes to delete all of the parcel polygons. Martijn van Exel and Nathan Mixter dispute the existence of a consensus against cadastral data, but agree that "dumping" (i.e., imports that violate best practices) are unacceptable. Toby Murray cites the Arkansas case as support against parcel boundaries. Paul Norman mentions a Spanish cadastre import that ultimately decided not to import parcel boundaries. Nathan Edgars II and Alan Mintz make the point that the parcel data density restricts the size of the maximum area that can be edited. Ian Dees points out problems with "on the ground" verifiability of parcel boundaries, and that parcel data is difficult to keep up to date because it is created administratively by official entities outside of OSM. Paul Johnson and Nathan Mills point out that parcel boundaries are verifiable (albeit with notable difficulty). Paul Johnson and Brett Lord-Castillo report that parcel boundaries are useful in the context of natural disaster recovery. Toby Murray disputes natural disaster recovery as being within OSM's scope. Gregory Arenius defends parcel data as a nice touch in rendered maps, and a solid base to work with for other mapping tasks, like updating landuse. He also responds to the matter of parcel data density as a tool problem, not a data problem with parcel data specifically. As of this writing, the Fresno parcel boundaries still appear in the data base.
Thread: parcel data in OSM
The original talk-us thread initiated by Jason Remillard can be viewed here.
Fresno and data quality
I don't have a problem with importing parcel data if it of high quality and can actually be useful. But sometimes the data will have only the shape of the parcel and nothing else included. If they have address data included with the parcel or zoning designation then they become more helpful. Many counties offer parcel data or zoning data.
I was able to get permission from Fresno County to add their parcel data to OSM. The data set was extensive. It included exactly what the parcel was used for, many of which could be translated into OSM. It even included down to what individual crop was grown. It did not include addresses, but these could easily be added later in areas that are surveyed. Some of the parcels were empty or did not translate into an OSM value and need to be surveyed later on to see what the land use is. Overall it was a good way to turn a blank white map into a map full of color and meaning.
http://www.openstreetmap.org/?lat=36.779&lon=-119.835&zoom=11&layers=M -- Srmixter 22:38, 23 March 2011 (UTC)
- Thanks for the feedback. I'd agree that just parcel geometry is not worthy of import, but address or landuse information is. - Joshdoe 12:02, 24 March 2011 (UTC)
Please do not import as landuse=residential
Please do not import this as landuse=residential. It'd make things only worse. We need a new tag for a parcel, since usually it is contained in a landuse=residential. Let it not be the same. It contradicts years of work in which human mappers handcrafted landuse=residential data from survey or aerial data. Mixing in official data does not help keeping things right. --Cmuelle8 21:19, 6 September 2011 (BST)
In developed countries at least the parcel corners are in fact marked on the ground by the state surveyors. Or were when the parcel was defined. Difficult and extremely tedious to verify, but they exist. Alv 09:00, 8 September 2011 (BST)
- It's easier than that – fences tend to follow parcel boundaries reasonably well. Ploppy 17:02, 19 December 2011 (UTC)