Parcel (cadastral) boundaries are great! Yup, *private* property

Just the other day I was looking at this map chock full of parcel lines.
I thought to myself, “Self, what a rich map!” Not only can one use it to avoid venturing onto private lands, boundaries of parks in forests no longer need to be guessed. Anyway, even if private property is borgoreious, I hope nobody is against showing the divisions of private property here on OSM.
I mean I’m not saying we should be like some county assessors’ GIS, and even have MR JOHN C AND MRS MARTHA B NEBBLESFORD owners’ names plastered on top of them. Well, not at least for now :-).

Hmm, it seems I mentioned the concept way back in '13.

1 Like

In the UK we actually have the land registry (government) cadastral parcels available to use in mapping in OSM.

They’re particularly useful in determining imagery offsets (as one can align hedges/fences etc with the boundaries).


Cadastral boundaries are extremely useful, but as someone who manages them for a county as my day job, I can tell you that they are very approximate at best. In some areas, they are off quite a bit, and it’s a massive task to actually get these things located correctly. Bringing parcel data into OSM would almost certainly be imports from “authoritative” sources like the one I manage. I’ve worked with surrounding counties as well (admittedly a sample size of a half dozen here), but all of them are pretty iffy in places, and I would be wary of importing any of it wholesale.

Also, in Illinois, maybe other locales, only a licensed surveyor is allowed to legally say where a boundary actually is, and showing the information anywhere else has to come with big disclaimers on it.

But anyway. I think cadastral data from counties and similar should be treated the same way that we do TIGER data: maybe an okay starting point, but requiring a lot of checking and cleaning before it can be considered reliable, current, and accurate.

For OSM specifically, I think we very quickly run into the problem of verifiability. The average person is not equipped to go out and locate the boundary markers, so how can we add these features to OSM with any confidence? Or edit them after they are imported? At least with TIGER, you can see where roads are in an aerial, and an offset by a couple feet isn’t huge.

With property boundaries, though, even a few feet can make a huge difference, and there’s no clear distinction between well-located boundaries and approximate ones in your typical cadastre. Also, they are always changing. We split hundreds of parcels a year, and are always working on refining the placement of existing boundaries. You couldn’t do a one-and-done import, and changes to boundaries would not be obvious in aerials or on the ground. Keeping this kind of data up to date would be a massive undertaking.

I’m not against adding this data to OSM, but I think it’s such a monumental undertaking to actually do it well, and maybe impossible / illegal to do with any claim of accuracy. But hey, more power to the one who wants to make the attempt!


There’s a relatively broad consensus not to map cadastral parcels in general. You can map walls and fences of course, but you generally cannot import cadastral parcels when you have zero clue of how verifiable they are on the ground. See also the warning boxes on Tag:place=plot - OpenStreetMap Wiki

Of course, there’s no rule without exceptions, and a US mapper has once only half-jokingly written that for a hiker, knowing where the public land ends can mean the difference between getting shot or not. Unsure if I’d like to burden OSM with that responsibility though!


I hope that you don’t copy from “that map” though! :face_with_monocle:

1 Like

OK, OK, boy do I envy those Brits with their perfect cadastre. OK, I got a fresh idea that… too bad the Brits don’t have…

Man, I hate the PLSS. Here in Kendall County, IL, we have what’s known as the “Indian Boundary Line”. South of the line was surveyed earlier for the I & M Canal, north of the line was surveyed years later. The two surveying crews did not end up in the same spots at this line, so all the township and section lines have a horrible zig zag in it, and boundaries along here are off sometimes over 100’! It’s so bad, some surveyors around here refuse to work on properties along the IBL.

They’re even harder to locate than cadastral boundaries, but you could at least get the townships themselves into OSM, for places where admin_level=7 is a real thing.



Unfortunately, a critical reference had been written as “miles and chains” in the Minnesota office, when it should have been written as “chains and links”. Consequently, the Seventh Standard Parallel across the reservation was drawn about two and a quarter miles south of where it should have been.

Not perfect at all: the first house I bought the cadastral parcel lines went right down the middle of our garage, so officially we only owned half of it, and half of the one next door. One frequently sees oddities with the UK cadastral data: very obvious when the parcel lines do not split a semi-detached property into two equal halves.

ALL data of this kind of volume will have issues, so needs to be used with some awareness of such a possibility. Also the cadastral boundary data we have is only indicative of the legal boundaries, in part because they will have been created from detail map data and tend to be aligned with visible features such as hedges and fences.


Agreed, property maps contain inaccurate information and would add a lot noise if added to OSM. Though it would useful having access to the data as an overlay when determining the relative center of a parcel. It would definitely make it easier to get the relative size and shape of each parcels. Making it much easier to figure out how handle their addresses. It is especially helpful when there are little to no obvious separations between properties.

1 Like

I only recently saw on a map that Washington DC is a perfect diamond shape!

Glancing at it, & thinking that it must make for some “interesting” features, I spotted this place: OpenStreetMap

“No, I did not commit that crime under your State laws, as I was in my back yard at the time”! :crazy_face:

1 Like

The boundary you show is between 2 Virginia counties. It hasn’t been the boundary of the District of Columbia for a long time (since 1846). There are numerous examples of buildings straddling country boundaries, often allegedly exploiting differences in the laws between the two.

1 Like

There you go!

Only found that out a few weeks ago, & now I learn that it’s wrong!

1 Like

I am against this.

That’s why it’s useful to map the limits of public lands, which is universally accepted (parks, nature reserves and that kind of thing). The danger of getting shot while accidentally wandering over the boundary of a nearby nature reserve is of course overstated (though maybe not zero), but it’s certainly useful to be able to see on a map that a trail leaves public land.


Yes, with all the cadastral boundaries available like the British have, one could instantly map park boundaries, no matter how hilly or woodsy they were.

Except, using cadastral boundaries wouldn’t be the right way to map park boundaries. (And I know of what I speak, having touched this third rail, and being seriously electrocuted by the controversy this created several years ago, worldwide, regarding the definition of leisure=park, noting that the US English definition of “park” is very much more widely encompassing than OSM’s wiki entry, which I’ve helped tune up, and the consternation level is way down).

@ZeLonewolf is correct that public lands (and what both states and the federal government publish into PADs — Protected Area Databases of geo data which delineate where park and “protected area” boundaries are) is the right way to do this. In the USA, our wiki are fairly well-decorated with such documentation, such as United States/Public lands - OpenStreetMap Wiki and California/Using CPAD data - OpenStreetMap Wiki. The way that this is done (and works, to a very large degree) is that the agencies or non-profits who publish such data take on the burden of chasing down what ambiguities there might be, so you don’t have to. This is not the same thing as saying that these data are definitive, authoritative, or even correct, but they are “the best data for this particular purpose.” As I document in our wiki, these aren’t really meant to be imported wholesale into OSM, I find it useful to curate one or so at a time into OSM where absolutely necessary, for example to clarify a park boundary.

So, @jidanni, at least in the USA, don’t map park boundaries (or nature reserves, or other “public lands”) using cadastral boundaries, do so (if you must) using PAD (for your state or federal data, linked above).

I’ve spent the last 13 years or so cleaning up a Zoning → landuse import by an infamous “spiller of buckets of paint” (back in the late-aughts and early teens when OSM importing was much more wild and carefree). The data have been through a dozen or so iterations, and countywide (it’s my county, and it’s relatively average-sized for the USA and very small for California) doing the necessary early corrections and about-annual updates was tedious (but worth it) to have improved these data (Santa Cruz County, California - OpenStreetMap Wiki). This gets easier as “most of the mountain has been climbed,” and now it’s essentially keeping up with updates (we lag by a couple years, but not terribly).


County, city, or state governments often make their parcel data available as public records and map viewers and they have enough trouble with accuracy of geography and ownership. OSM has enough troubles as it is than to chase that dragon to the ground. Commercial services like Regrid which market access to nation-wide extents of this public data are available anyhow.

1 Like

I don’t know about Regrid, but it is obviously a thing, so there you go. Thanks, @Skunkman56.

Let’s all pause for a moment and realize that even for people who have been in OSM for a long, long, LONG time, there are use cases and downstream applications of OSM’s data which we have not yet even imagined, yet somebody else did. And they are picking fruit from fertile soil. We only imagined that the seeds we sprinkled about and planted with good intentions, good local data and some wiki entries sprinkled about here and there would turn into a whole community of excellent mappers.

1 Like

Ah yes, the parcel lines that make it painfully obvious that Google Maps lacks cul-de-sac rendering. :grin: Google Maps evidently divides its database into traditional GIS layers. This gives them the ability to integrate massive datasets without worrying too much about conflation or connectivity, as long as no one is too worried about seams or inconsistencies caused by different vintages of data – the kinds of issues we’d tend to fuss about. Many querying tasks we can perform by walking OSM’s object graph in Overpass probably require spatial joins on their end.

Yes, probably nonzero in areas where nature reserves tend to be surrounded by hunting grounds, gun clubs, or military firing ranges. I’m not really into these things, but I do my best to map their limits accurately, if imperfectly, for awareness’ sake. (Fortunately, we do have a disclaimer.)

1 Like

Ah, maybe that’s the key, to a healthy OpenStreetMap future.

Instead of expecting everything to happily all connect together,
have several independent layers.

All the government pipeline, parcel, etc. information could still be
right there on the map. But users could no longer connect their go-cart
trails etc. to its nodes.

And let’s say when the Canada Pipelines or whatever layer needs to be
adjusted a few meters, there would no longer be a worry about dragging
some junk someone has attached to it along with it.