|The following discussion is primarily US-focused. Although it mostly holds good in other countries, in many places addresses are not assigned to a land parcel and there may be no clear 1:1 relationship between them.|
No widely accepted consensus exists on whether or not parcel boundaries and associated parcel attributes should be made available in OSM. The purpose of this page is to organize the discussion and hopefully establish a consensus, one way or the other.
- 1 Introduction
- 2 Parcel Data Questions
- 3 Parcel Data Tagging
- 4 Parcel data as a secondary source
- 5 References
A parcel, lot, or tract is a piece of real property. In the United States, legal property boundaries are specified on deeds and subdivision maps which are recorded at the recorder's office for the jurisdiction in which the land lies. Usually the jurisdiction is the county, but in many states it is the municipality. These records are public information and generally not subject to any restrictions on copying or redistribution. However, most agencies are empowered by their states' public records laws to charge reasonable fees for duplication of these documents. Recorded documents are often searchable online, and occasionally the scanned documents themselves are available. Recorders do not generally prepare geo-referenced versions of the parcel information. However, in principle, a motivated mapper could reasonably acquire a detailed description of legal parcel boundaries and verify them on the ground.
Local tax assessors in the United States use the recorded document stream to maintain tax maps. Because the assessors are the agencies that maintain actual maps, when people discuss parcel maps they are usually talking about assessor parcels, not legal parcels. This distinction is important because assessor parcels are not always exactly the same as legal parcels. For example, assessors may combine adjacent parcels owned by the same entity for administrative reasons. Finally, you will often find disclaimers in and around tax maps warning the parcels should not be used to determine legal boundaries. For that you must go to the recorder.
Assessors assign a unique Assessor Parcel Number (APN) to each parcel. They also maintain tax information associated with the parcels (such as assessed value, tax rate, exemptions, etc.) and contact information for the parcel owner. They also may track information regarding existing structures on the property such as the number of residential units, floor area, etc. Assessor maps are often available from the assessors' websites, and parcel information is often available indexed by address and APN. Some form of these maps are public information and available through mechanisms similar to those in the recorders' offices.
Assessor maps may or may not be readily available in a GIS format such as shapefiles. When they are available, they are often maintained by not by the assessor, but by some other agency such as a planning department. For example, in Marin County, CA, this is handled by the county's community development agency. In San Francisco, this is handled by the Department of Public Works. In any case, GIS parcel maps whether or not they are developed by government agencies may or may not be treated as public information. Many court cases have been fought on this matter.
Parcel Data Questions
A number of specific issues regarding parcel data have been debated on various OSM mailing lists. The specific correspondences are summarized on Talk:Parcel. The following specific questions are posed to organize the community thinking around the role of parcel data in OSM.
Does parcel data belong in OSM?
Parcel data is a bit of a general term. When we say parcel data, we mean the parcel boundaries and some associated data, including addresses, land use, zoning, data about the buildings, tax information (such as assessed value and assessor parcel number), and probably more. Nobody disputes that parcel data is useful, and many mappers have made use of parcel data to enhance OSM. These enhancements have included deriving address points from parcel data, inferring the boundaries of large land use areas from parcel boundaries, locating schools and parks, etc. Some mappers believe that adding parcel boundaries to rendered OSM data at the right zoom level would be a nice enhancement, and that the availability of the assessor parcel numbers permits the potential joining of the OSM dataset with other data sources. Some offer application-specific examples of parcel data being useful. Like in the context of disaster aftermath, for making accessibility maps, and for land-use modeling.
But just because parcel data is useful does not mean that it must be merged into the main OSM dataset. That is, we must differentiate between parcel data being useful to OSM and parcel data belonging in the main dataset. There is little dispute that picking and choosing various parcel attributes for inclusion in the main dataset (such as addresses) is probably acceptable. But many mappers believe that the parcel boundaries specifically do not belong in OSM. The core reasons for this position are discussed in subsequent sections. But the general idea is that OSM is not a general-purpose geo data repository. It is a system for crowd-sourcing a map. Also, parcel boundaries and some related data (e.g., zoning) are an administrative construct managed by a government authority. Accordingly, they change frequently in ways that can't be readily observed on the ground. Also, if a mapper edits the parcel data in a way that conflicts with the authoritative source, the conflict must be reconciled somehow, which would be difficult and maybe impossible.
Some mappers do believe that parcel boundaries and associated data should be imported into OSM. They conceded that the current tool chain presents some technical limitations [more on this below]. But they believe that importing parcel data now will drive the necessary improvements. They also believe that the parcel data adds a level of detail to OSM that is inevitably being pursued.
So no consensus on the question of whether or not parcel data belongs in OSM has been achieved as of this writing. But there is a consensus that having some sort of OSM-compatible source for parcel data would be a useful next step.
Can parcel data possibly be kept up to date?
The most efficient way to populate an area with parcel data would surely be to import a government dataset. The next natural thing to happen is a mapper (or preferably, the importer!) will tune parcel boundaries to match existing administrative boundaries and roads, and perhaps update some of the tags based on observations. Then, when the government dataset is updated, how can the conflicts be reconciled?
One idea that regularly surfaces in the community is that this sort of situation is to establish tools and processes for a two-way sync with external datasets. In this way, changes to OSM data that had been imported would be fed back to the manager of the source data. So future imports of updated data will in principle not blatantly over-write user changes. While some tools and processes such as these have been demonstrated , no best practices or common tools exist. Finally, agencies may not be interested in participating in two-way syncing processes.
Another perspective on this is that the Fresno parcel dataset was imported two or three years ago and has not inspired any updates. Considering the rapidity with which parcel data can change, it is not likely that the current parcel boundaries represent reality any more. Perhaps this is evidence that (under current circumstances at least) parcel data cannot be kept up to date.
A common corollary concern here is that parcel data management will require reliance on imports from official sources, so why not just go to those sources? Another is that imports are often performed in a way that destroys other user contributions, and that imports undermine OSM's purpose as a human-powered endeavor. Critics make the point that this brings up more general question about import practices, and is not just a question about parcel data.
There is widespread acceptance in the community that maintaining reasonably up-to-date parcel data in OSM or otherwise would require special tools and processes that do not currently exist. Notably, proprietary sources of parcel data layers do manage to offer regularly updated datasets. Of course they probably do not cope with the editing conflict problems that would be inherent in the OSM environment.
So managing parcel data in OSM or even in a fashion compatible with OSM will require clear processes regarding how parcel data updates are to be managed.
Does parcel data meet the "on the ground" verifiability criteria?
Some say yes, proposing that verification of parcel boundaries can be achieved using publicly recorded documents with suitable detail and a ground survey. Critics make the points that this is a very cumbersome process and not likely to result in widespread practice. They add that because parcel boundaries cannot necessarily be observed on the ground, they cannot be reasonably verified.
Can tools be adapted to accommodate this data density?
OSM editors are restricted by the number of objects or file size they can handle. Because parcel data is dense compared to some other OSM data, mappers will be forced to restrict their editing to smaller areas when parcel data density is high. This has been put forward as an argument against accepting parcel data. Critics of this argument point out that this is a limitation of the tools, not a problem with parcel data specifically. Indeed, building outlines present this same problem. Also, coming to a consensus on an optimal data density is unlikely.
Parcel Data Tagging
No detailed proposals for tagging parcel data appear to have been put forward. Some have suggested importing this data as landuse=residential. Others have pointed out that many parcels may be set within a single landuse=residential area and have called for a different tag.
Parcel data as a secondary source
Even though direct importation of parcel data is highly controversial, it can be of great value as a secondary source for other features of interest, particularly if the source includes information about land use or land ownership. Searching through parcels for ones owned by a government is likely to find features that deserve mapping (parks, schools, libraries, hospitals, prisons, government offices, sports facilities, cemeteries, ...) and give information about the boundaries of the associated lands.
In no case is it appropriate to use third-party electronic sources to overwrite data that a local mapper has produced from surveying in the field.
- County of Santa Clara v. California First Amendment Coalition
- Microdecisions, Inc. v. Skinner (2004) in Florida
- Sierra Club v. County of Orange 195 Cal.App.4th 1537
- County of Suffolk, New York v. First American Real Estate Solutions (2001)
- Portland Trimet personnel report attempts to do this http://lists.openstreetmap.org/pipermail/imports/2011-April/000881.html