This raises an interesting question: if mappers aren’t generally able to improve individual property lines on their own, and there’s no need for explicit connections to OSM’s other data, does this read-only property boundary data belong in OSM? At that point, a separate layer might as well be a separate database, even if we make a mashup of OSM data and this other database to achieve the look you saw on Google. After all, some of osm.org’s featured tile layers already pull in other data sources, namely CyclOSM and OpenCycleMap with elevation contours and hill shading. It is technically feasible to distribute OSM alongside another database of things that don’t go in OSM. Call it Daylight or Overture if you like. 
2 Likes
Indeed. 99% of highways, railroads, rivers are already there in official
sources. Mappers could better spend their time not duplicating them.
Then everything could all be sandwiched together in apps like
https://www.offline-maps.net/ , including the 1% tweaks, and cafes, etc.
I can’t tell if this is sarcasm, but for what it’s worth, there’s a difference between the kind of data that can be improved by the general public and the kind that the general public has to take at face value. For the most part, you and I can probably shrug at the Board for Professional Engineers, Land Surveyors, and Geologists when armchair-mapping California’s streets and rivers, but do we really want to push the boundaries with bespoke property line mapping? Most of us don’t hold a surveyor’s license and can’t afford one.
4 Likes
Yes, if like in the Google example, the property boundaries were already
visible, then nobody would need to try to put their best guess as to
where they were on the map.
As far as I can tell in almost 14 years of OSM mapping (including, perhaps even especially boundaries), nobody IS “try(ing) to put their best guess as to where they (are) on the map.” Not down to the level of individual parcels.
If “the Google example” does that, OK, that’s Google (Maps). But I don’t expect to see property boundaries in OSM, especially down to the “parcel level.” At the larger scale of “zoning,” there might be what are effectively the same sorts of things (boundaries), say between a commercial landuse and a residential landuse, where the boundary between property zoning is effectively the same as the boundary between parcel (cadastral) boundaries — that’s a simple coincidence of what IS being mapped (landuse differences).
Let’s say there is a property line between a shopping center and a tract of houses, but no actual fence that delineates this. If someone were to map the landuse=retail surrounding the shopping center (nothing wrong with OSM mapping that), something approximating “property lines” would be mapped (maybe even WITH what might be called a “best guess”), but they’d be largely correct, and quite effective for delineating a landuse boundary (specifying a retail area, in this instance).
OSM does this all day, every day. What we don’t do is further delineate the parcel boundaries between, say, “the fuel station / convenience store on the corner” from “the anchor store in the middle of the shopping center.” Nor should we.
1 Like
So, if I understand you correctly, you want something like QGIS layers, right?
If so, what is stopping you from with using it (or similar solution) right now?
i.e. If your government publishes open GIS databases (via WMS, WFS, TMS, etc…) you just add them as layers in QGIS, and then display/query several such layers at will in addition to OSM data?
3 Likes
I think this discussion has come full circle, because that’s exactly how British cadastral parcels work in OSM. Not as an import but as a layer to be enabled in editors when mapping other things that are verifiable on the ground, such as leisure=garden garden:type=residential with a barrier=fence/wall/hedge in between them.
This discussion is also worth reading:
2 Likes
In Santa Clara County, California, around the corner from where @jidanni was looking on Google Maps, we’ve been referencing the county’s official GIS parcel layer when mapping shops and other POIs – not for the property lines per se, but rather to match addresses to the external dataset. Sometimes parcels have unintuitive shapes, so it isn’t enough to cross-reference a house number that’s floating about the parcel’s centroid.
Perhaps this was too subtle, so I’ll try again: the California state board responsible for issuing surveyor’s licenses has been fining unlicensed individuals for authoring and distributing maps that purport to depict property lines. In Crownholm v. Moore, an entrepreneur had made a business of drawing site plans for residents (to submit to their local authorities for permitting) by tracing the property lines off of Google Maps. The board ordered him to cease and desist and pay a $1,000 fine because he was unlicensed. With the help of First Amendment advocates working pro bono, he challenged the board’s ruling in federal court – and lost.
I don’t know where that case is headed next, or to what extent it applies to the kind of mapping being proposed here. But I personally wouldn’t want to jeopardize our scratch-made, free-range craft mapping for the sake of something that the public can already obtain elsewhere as open data. The most we’d do with the Santa Clara County parcel layer would be to make sure our landuse areas are consistent with the property lines, but we wouldn’t try to reproduce the property lines as such. We shouldn’t give users the impression that we’re trying to collect that kind of information.
Property lines are sensitive because they’re statements of ownership. People have gotten in trouble for misrepresenting ownership of land. That’s why the process for becoming a professional land surveyor is so involved. Even the county agencies responsible for maintaining these parcel layers and collecting taxes based on them insist that you respect their disclaimer of warranty, regardless of the data’s license.
5 Likes
That is how I see the Cadastre used in Spain also. There is an ongoing process to import building outlines from Cadastral data. Separately from that, the map is available as a background layer in JOSM and ID, and the parcel outlines can be very useful for mapping the kind of physical objects you mentioned (I sometimes find myself flipping between aerial imagery and the Cadastre map). When used in that way, there is no suggestion that ownership limits are being defined. Indeed, a wall that separates two fields owned by the same farmer is tagged the same way in OSM as a wall that is a boundary between two farms.
1 Like
The history of cadastral parcels can be rather complex. My Uncle who died in November lived in a small hamlet in Finisterre. The house was awarded a single house number fairly recently, but is made up from 6 cadastral parcels. In this case we have a good idea of the history of these parcels: 3 were distinct properties in early 1960s, two represent extensions to two of those properties, the 6th is a former pig stye reused as a garage. All were acquired, and in one case built, by my Uncle and Aunt from 1980 onwards to convert a simple 1-up 1-down cottage used in the Summer into something more like a permanent home. This is one of the cadastral markers outside the pig stye:
I won’t share the location because I don’t know my cousin’s plans for the place now both his parents are dead. However, I suspect this sort of thing applies to many rural properties in France and Spain, and raises the issue of “what is a building?”
1 Like
Please don’t. It was a good example illustrating the messy nature of property lines and why they shouldn’t be included in OSM. Beyond that no one needs to know where you uncle’s property is.
3 Likes
|
|
- | - |
Perhaps this was too subtle, so I’ll try again: the California state board responsible for issuing surveyor’s licenses has been fining unlicensed individuals for authoring and distributing maps that purport to depict property lines. In Crownholm v. Moore, an entrepreneur had made a business of drawing site plans for residents (to submit to their local authorities for permitting) by tracing the property lines off of Google Maps. The board ordered him to cease and desist and pay a $1,000 fine because he was unlicensed. With the help of First Amendment advocates working pro bono, he challenged the board’s ruling in federal court – and lost.
is this something we have to fear as OSM(F)? We don’t care for Chinese legislation, or Russian, do we have to care for Californian? Actually, maybe yes 
I don’t know where that case is headed next, or to what extent it applies to the kind of mapping being proposed here. But I personally wouldn’t want to jeopardize our scratch-made, free-range craft mapping for the sake of something that the public can already obtain elsewhere as open data. The most we’d do with the Santa Clara County parcel layer would be to make sure our landuse areas are consistent with the property lines, but we wouldn’t try to reproduce the property lines as such. We shouldn’t give users the impression that we’re trying to collect that kind of information.
IMHO we are not aiming at collecting this kind of information (property register), but it happens frequently that landuses and features are located on plots/parcels, and thus the parcel boundaries are the feature limits, significant information we do collect.
1 Like
Local laws primarily matter for the contributors who reside in those localities. We as a project only care about California laws to the extent that we want to attract Californians to the project without asking them to face legal repercussions, or to the extent that we want the project to be used by individuals and companies based in this state. The same goes for any other state or country, but I focused on California because the original post used a neighborhood near San Francisco as an example. Anyways, if we’re asking people to take a potential personal risk, there has to be a good reason, but in my opinion this discussion hasn’t presented a compelling case yet.
Yes, I expect and certainly hope that this pragmatic approach will steer clear of the board’s concerns. I don’t mean to be alarmist, but I also don’t want our stylish OSM high-viz jackets and professional-grade GPS units to get to our heads without also considering the real-world implications.
Via this Slack thread, here’s an extended discussion between GIS professionals and the professional land surveyor who filed the original complaint against Crownholm. The discussion touches on some points that didn’t make the press coverage, including:
- The core of the complaint was that the site plans not only displayed property lines but also claimed them to be of sufficient accuracy for local pre-permitting. There’s some debate about whether that constitutes surveying, based on the idea of local authorities waiving the requirement for a professional survey.
- One of the lawsuit’s claims was that the state’s legal definition of a survey is overbroad, covering any “lines on a map”, thus infringing on free speech. (But the court was unconvinced by this argument.)
- The site plans explicitly annotate distances between structures and property lines, whereas Google Maps and county GIS layers do not, leaving it to users to measure distances on their own.
- The county parcel layers we’ve been talking about technically do not depict property lines, just tax lots. So it would be even riskier to claim them to be property lines.
- Crownholm did have some legal recourse at the state level – appealing to an administrative law judge – but allegedly declined in favor of this federal lawsuit on free speech grounds.
The takeaway is that we aren’t publishing anything like these site plans right now, but we should be wary of any calls to do so, especially if the current legal definition of a survey remains. Incidentally, some of these site plans express distances down to a precision of a hundredth of a foot (3 mm), which is unrealistic in a land where tectonic plates can move many times that each year. It reminds me of my desire to be able to express elevations in feet and inches rather than overly precise or imprecise metric equivalents. (Now that the U.S. survey foot is finally deprecated, we have no excuse!)
Okay, I finally did it. I finally put myself in the average mapper’s shoes. The ones with only a cell phone.
And here is my masterpiece:
So how did I do it? Flipping between one screen and other and just trying to remember where the boundaries were.
I had the official public land stuff on one screen and open street map on the other.
So yes, even though there’s great California public lands maps websites etc. But if the best you can do is copy by memory between screens, well of course naturally you’re going to end up with some kind of a lopsided creation.
Sure you could say well you could still just take a screenshot and put it on Mapwarper, then you could stick it into Vespucci or whatever. But remember we’re just trying to do this on a cell phone. So Mapwarper would be hard to use.
Anyway, all the well-meaning citizen mapper wants to do is get that park on the map fast. So lopsided it will be. It’s better than nothing. And of course one day somebody will fix it hopefully.
Anyway if the needed layers were in the editors, that wouldn’t all have to be that way.
So I suppose it’s not a problem anymore in Europe.
Okay it’s true average mappers might not just have a cell phone, they might have two. But they don’t have a desktop or a laptop 
And I know, you’re saying, “Looks like a great park to me. I can’t wait to go there.”
But that’s before you compare the official boundaries.
Am I understanding this correctly that you wanted to map where the park really is, using only mobile phone?
If you’re using Vespucci that you mention, you could have started GPX track in Vespucci, walked around the park, and then use that same Vespucci to draw the park area along that GPX track (you could’ve done it even without enabling GPX track recording, but it would require more looking at the screen).
So I suppose it’s not a problem anymore in Europe.
Well, I don’t know, the method I mentioned seems to work worldwide when on the ground.
Or am I missing something?
I had the official public land stuff on one screen and open street map on the other.
Why? To me it seems counterproductive to be on the ground with only a mobile phone, and try to emulate JOSM experience on a mobile phone with tiny screen estate.
Anyway if the needed layers were in the editors, that wouldn’t all have to be that way.
So, what is the problem you’re trying to actually solve? I’m a little lost. It seem to me that you are trying to copy the map from some source that is legally allowed to be used with OSM, but does not provide WMS/TMS/WMTS nor any other common method of access? Am I close?
And I know, you’re saying, “Looks like a great park to me. I can’t wait to go there.”
Why? That buildup of suspension sounds ominous. Are there hidden crocodiles there or something? 
2 Likes
Also, that way says “source=https://open-data-portal-san-rafael.hub.arcgis.com/apps/wildland-urban-interface-map/explore” and visiting that URL says "No License Provided - Request permission to use".
Have we actually got their explicit permission that this data may be used for OSM tracing (and where is it documented, if we have)? Because if we haven’t, we are not allowed to copy data from that map to OSM - neither automatically, nor by manually switching apps and looking at it and redrawing it elsewhere.
1 Like
Fortunately, all California state and local government agencies are subject to a provision of state law that puts their work in the public domain as public records. (There are a few exceptions, like school districts, so that you can’t publish exam materials unhindered.) Only a couple states take such an enlightened view towards copyright. Most local agencies, including the City of San Rafael, host their datasets on ArcGIS Hub or an ArcGIS Online instance, which automatically provides a FeatureServer or MapServer with a WMS endpoint. Sometimes they’re even suitable for use as a background layer in any OSM editor.
Unfortunately, many local agencies claim copyright anyways. This particular case is benign – the city probably neglected to set a configuration option in ArcGIS Hub. If they had, Esri likely would’ve posted a boilerplate import proposal on the wiki by now. But the county I live in always claims “all rights reserved”, even though they lost before the state supreme court years ago over the very issue of copyrighting their GIS basemap layers, specifically the parcel layer I mentioned cross-referencing earlier.
Fortunately, the law is clearly on our side.
Unfortunately, this dataset is not an open space or parks layer at all, but rather an ArcGIS-based application displaying a dataset that indicates the wildland urban interface, a concept that relates to landcover but would not indicate the limits of any park or formal greenspace. (The application also displays the city boundary, which has already been imported from TIGER, and addresses, which Esri says they want RapiD users to add.)
None of these features provided by the City of San Rafael have anything to do with “San Rafael Open Space”. But the map you looked at does have a label saying “San Rafael Open Space” over a patch of green that sort of resembles this area you added. The label comes from Esri World Basemap v2, which is a fusion of many sources, including proprietary sources like Garmin and open sources like OSM. World Basemap is governed by terms of use that likely put it out our reach when editing. So what you did is probably not something to boast about.
Fortunately, I was able to track down the original source of the feature you traced. It comes from a “Parks” layer in Marin County’s open data portal. The county did properly configure the license to state their disclaimer of warranty, so license-wise you’re good to go.
Here’s the FeatureServer. You won’t be able to load it as a background layer in iD, because it doesn’t support FeatureServers yet:
In JOSM, you also won’t be able to load it as a background layer in imagery preferences, because they didn’t enable GetCapabilities. But if you download the dataset as a shapefile, you can install the OpenData plugin in JOSM and open it as a normal layer, which you could then merge into an OSM layer, feature by feature.
The metadata associated with this particular feature indicates that it’s just open space, not necessarily a formal park deserving of the leisure=park tag. The FeatureServer description explains what the dataset counts as “open space”:
“Open space” is defined as property protected from development for the purposes of preserving the upland greenbelt, habitat, native species, or special environments. Recreation is usually allowed in these open spaces, but it is generally limited to passive recreation such walking, hiking, bicycling, etc.
Maybe this is a park, but I’m getting pretty strong leisure=nature_reserve vibes from aerial and street-level imagery:
Beware of crocodiles!
The name “San Rafael Open Space” occurs on 16 disconnected features all across the city. So this is just a generic name the city applies to any open space it owns. name=San Rafael Open Space is probably inappropriate, but owner=City of San Rafael would be appropriate.
So yes, it is more complicated than you asked for, but hopefully peeling back the layers of this onion proved informative.
4 Likes
@jIdanni, I appreciate your efforts, as they were on a cell phone (with a tiny screen, compared to the relatively luxurious 27" Retina I used to do what I describe below).
I read up on what you call “the official boundaries” at California/Using CPAD data - OpenStreetMap Wiki (full disclosure, I am the author of this wiki). I used the cpad2022b.zip data (the latest) that are available from the data’s publisher GreenInfo and “pulled out from the whole state” (via the “Holdings” file) exactly this polygon. I adjusted the tags according to the suggestions in the wiki, migrated what your tags were to the one I’m entering (as you have what appears to be a name=* tag, which might be better stated as loc_name=*, I included what’s in CPAD as official_name=* and otherwise fiddled the tags to “better conform” to OSM for “singly-curated data” (like these — again, CPAD isn’t meant to be wholesale-imported into OSM, but to grab a single “park” it’s fine). And because (no amenities), this doesn’t really meet the definition of what OSM defines as a leisure=park (full disclosure, I’ve had a lot to say in that wiki, too), I changed its “physical tag” to leisure=nature_reserve.
I deleted yours as a “first draft,” as I replaced it with “official data.” (And yours “wasn’t too bad,” but it was, let’s say “a bit crude” — no slight meant, I hope none taken).
This “new one” is Way: San Rafael Open Space (1165313068) | OpenStreetMap.
Upshots? Mappers have a wide variety of devices with which to map, including cell phones, laptops and desktops (I have all three, and more). Cell phones CAN BE and ARE used to map “first drafts” of such “things,” and personally (I think many other OSMers agree), I have no problem with this. Heck, desktop machines can be used to enter first drafts of things, and that’s OK, too. I wholeheartedly agree with you “it’s better than nothing” (and thank you for your entry, it inspired me!) and “one day somebody DID fix it.” (Me, here, now). This all took me all of <10 minutes.
Thanks for your many contributions. OSM: what a great project!
(BTW, you seem familiar to me somehow from long ago. Did you work at Apple in the '80s or '90s?)
2 Likes
Of course I am familiar with you. I’m that lasagna you had for dinner last night.
Anyways, I cannot tell a (second) lie.
I also did Changeset: 135252963 | OpenStreetMap at which time I was getting rather frustrated so I did a real Hasty job, Good thing somebody’s going to fix it
And I just couldn’t bear, I just couldn’t couldn’t hold myself back. This boundary was running all over people’s property. I’m talking splitting homeowners’ houses into half in one city and half another. So I had to straighten it out:
Changeset: 135252750 | OpenStreetMap .
Good thing I only "straightened out’ about 100 m of it.
Maybe it was from official Tiger sources…
Anyway I wasn’t actually at the scene of the crime. I’m rather kind of like in a different time zone or something.
Anyway I swear I was just looking at the maps that I sort of clicked into by looking at Steve’s California free data website or something.
Anyway now I know why people just use points instead of attempting to draw any boundaries for important items no matter how big they are.
We see in the reply to Copying exact positions from background layer · Issue #9595 · openstreetmap/iD · GitHub that future generations (OK, maybe in our lifetime ) might have better tools…