From OpenStreetMap Wiki
Jump to navigation Jump to search

Individual Fields

With the announcement yesterday that SteveC has joined Bing Maps and that their imagery will become available for tracing, I started wondering about adding field boundaries near here when the imagery becomes available. These are my thoughts on mapping individual fields:

  • From the imagery, tag as many of the field boundary ways as possible with barrier=hedge/barrier=ditch/barrier=fence (or whatever, as appropriate).
  • Complete the field "boundary" with untagged ways, to form a closed area, with barrier=gate nodes if appropriate (possibly can't be determined from imagery).
  • Use a multipolygon relation tagged as landuse=farmland with the boundary ways as the members, role outer.
  • If the field has a name that you know (or have compatible source for) that tag can also go on the relation as name=*

Thoughts or comments? I'll test the above on a few fields once the imagery becomes available and see how it goes. --EdLoach 09:21, 24 November 2010 (UTC)

Further suggestions received - one large landuse=farmland area with barrier=* ways on top. May be simpler. --EdLoach 09:40, 24 November 2010 (UTC)

Redirected to landuse=farm

This page is redirected to landuse=farm. Maybe that's not the best thing, but in September somebody got rid of the redirect and copied the exact contents of Tag:landuse=farm here instead. That's pretty silly given that this content includes an explanation of why landuse=farmland is not the best tag to use.

For now I've put back the redirect, but maybe it should have explanatory content similar to the Farm page on here

-- Harry Wood 13:45, 22 February 2011 (UTC)

I'm amazed at how old this is. Either way, I'm copying the content Tag:landuse=farm over here, as it is already so edited as to only point to the usage of Tag:landuse=farmland. I'll try to edit the contents to deprecate Tag:landuse=farm. Messy Unicorn (talk) 08:54, 28 September 2014 (UTC)
Yes I see what you have done, but can explain why you are doing it?
Amazed at how old what is? Ye olde tale of the landuse=farmland tag? Here's another old discussion: What's wrong with landuse=farm?. ...still no answer to that question.
-- Harry Wood (talk) 00:53, 6 October 2014 (UTC)

Looking back on this it seems like my "pretty silly" comment was a little unreasonable, since the version Messy Unicorn put in place had been edited to point towards landuse=farmland and away from landuse=farm reasonably clearly (maybe I jumped to the conclusion that it hadn't been because it looked like an unedited copy). Apologies for that. Nobody ever did answer that question properly mind you! Anyway... ancient history. -- Harry Wood (talk) 01:24, 30 November 2021 (UTC)

Subkey farmland=

The key farmland=* was not previously documented, I've now created a stub page for it. But the common values are not very helpful. E.g. farmland=pasture has been used, but landuse=meadow is the standard tag for meadows / pastures. See e.g.

farmland=field does not seem to add any information, and farmland=arable adds little information (generally, farmland must be arable, since arable land is land that can be used for growing crops). There is also farmland=wetland which probably should be natural=wetland, and farmland=plant_nursery which could be landuse=plant_nursery in most cases. --Jeisenbe (talk) 01:20, 18 February 2021 (UTC)

This key ... exactly what is it? The word type tells me nothing about what the key is meant to map. Farmland may not be arable and yet used to raise livestock - eg cattle, goats. Another key is produce as this would map what is produced there. Warin61 (talk) 07:28, 18 February 2021 (UTC)
Areas used for livestock are mostly mapped as landuse=meadow (for meadows and pastures used for grazing) and landuse=farmyard (for paddocks and barns where livestock are kept, e.g. a feedlot or dairy), or less commonly landuse=livestock or landuse=animal_keeping. The tag landuse=farmland is generally used for land used to grow annual crops. Permanent shrubs, vines or trees grown for food are tagged landuse=orchard or landuse=vineyard, and meadow for hay or pasture is landuse=meadow as mentioned above, so landuse=farmland is mostly cropland - though occasionally new mappers use it for any kind of agriculture this is not the consensus view. --Jeisenbe (talk) 17:58, 18 February 2021 (UTC)
There are areas of natural vegetation that are used for grazing of cattle in Australia that I could not call a paddock nor a meadow in any stretch of the imagination. I think you will find similar areas in South America and Africa. The world is a large place. Warin61 (talk) 09:35, 19 February 2021 (UTC)
Yes, we have rangeland like that in the western USA as well. These areas are not tilled farmland nor managed pasture – which often has fences and irrigation in dry climates – and often it is not possible to tell if the land is grazed by wild animals such as elk, bison or kangaroo instead of domestic cattle or sheep, so I would map lightly managed rangeland with the appropriate natural vegetation tag, e.g. natural=heath, natural=scrub or natural=grassland usually. --Jeisenbe (talk) 01:59, 21 February 2021 (UTC)

Farmland, farmyards and meadows

Please discuss the topic of the above recent landuses changes here, not on a semi closed environment of tagging mailing lists (where only a few opinions can be expressed) so that everyone can read, disseminate and understand any changes proposed before altering the wiki pages for the tags landuse=farmland landuse=meadow and landuse=farmyard that has cited entries for its terminology. If there are regional differences in the understanding of the terminology then explain that under sub-headings, please don't just change for all regions. More information equals a better map and mapping experience for all. -- DerB (talk) 10:27, 16 December 2021 (UTC)

"where only a few opinions can be expressed" - [citation needed]. Have you actually discussed this changes with anyone? Which region is mapping in this unusual way? Note that as it stands now mailing list discussion unanimously (very rare for tagging mailing list) confirmed that your changes mismatch how this tag is used Mateusz Konieczny (talk) 10:36, 16 December 2021 (UTC)
The changes by DerBauer remind me of - the proposal failed, even in its modest form. Around the area of my local experience, meadow is used for both meadows and pastures, farmland is used for maize, wheat and vegetables. The distinction is easy to get right for people even without a background in agriculture and mapping from aerial. On the ground, the distinction is also very stable. Farmers usually manage their own properties or have long lasting leases. If I understand fully, the change would require either a massive effort to examine features or massive retagging due to the inability to decide. It would create curiosities, like alpine pastures turned farmland (although the change in appearance on OSM-Carto might make it resemble the aerial a bit more.) I do not see a better map coming to pass from the proposed change. --Hungerburg (talk) 11:34, 16 December 2021 (UTC)
Mateusz Konieczny As a quick response to your post above, 'how this tag is used' should not stop us from really defining what the tag means (actually and regionally, not colloquially and individually) and therefore allowing us (as a group) to define how it should be used.

So to start, It would be helpful if the discussion on the topic of farmland, farmyards and meadows was not in a semi closed environment of the tagging mailing list but on the discussion page of the item in question. I have now read the thread(s) you make reference to and have some queries over the items raised which would be good to discuss on an open forum, ie the discussion tab of the relevant tag. Firstly my primary concern is that the map is as accurate as possible - and in turn that the hierarchy of tagging landuses is as accurate as possible. The map data and its classification system is a system like many others that incorporates hierarchy if used correctly, with landuse sitting above other types as it infers use, ownership and in turn access and privacy. The ground cover or final ‘look’ of what is on that landuse might be classed as a child of that item or a secondary element like crop=grass or natural=grassland. For example a field might be farmland as a landuse, but its final groundcover may be grass or other types of crop. It is also important for me to note that the map data and classification is distinctly separate from how it may be rendered on the many variations of map renderers using OSM data, mappers (referencing Hungerburg'snote here as an example) should not be mapping areas as grass because it will render green on a map but should be more concerned by the accuracy of the operation/ ownership of that land if they want the dataset of OSM to be of any use on a large scale. The data of the map should be given priority over the rendered intent of the map, or the way that different users around the world interpret the sometimes ambiguous terminology of the tagging.

The specific meaning of a ‘meadow’ was cited from wikipedia (apologies for the chrome specific citing method also) as clear reference to what an ‘agricultural meadow’ actually means and not what it colloquially means - colloquial terms should (again, in my opinion when trying to be accurate) be avoided, or at least noted for the regional disparities on the wiki page of that item so that mappers know what is to be understood/used in their area. An agricultural meadow is typically (in the UK meaning of the term) used for hay /haylage, or set aside area, not for grazing or pasture which also brings into question the tag for pasture (unless there is first-hand knowledge of the field being used as pasture this should not be used). It can also mean a grassland area of environmental protection for the promotion of biodiversity (one which I believe is the best use for this tag). For the avoidance of doubt, and to help mappers understand and not infer their own understanding on terms used in tagging the wiki page must be as clear as possible and include for the regional disparities (with cited references). If those references are on wikipedia, all the better because (again my opinion only) OSM should not be hiving itself off in the corner of the web but drawing on the knowledge shared across ‘open’ reference repositories like wikipedia, wikidata etc.

All these terms are also made slightly tenuous when tagging from aerial imagery, where the actual usage of the land has not been noted over months/years of site specific knowledge. A grass field area could be operated by a farm (farmland) as pasture for animals, but it also could be part of an environmental grant scheme to promote biodiversity and not be grazed, it could also be temporary cover crop to promote nitrogen sequestration and not be permaculture. Transitional crop systems would not be evident in one picture of aerial imagery, nor would ownership. I use the example of the UK again which is made up of roughly 70% farm operated land which in turn is comprised of arable AND pastoral land. Where neither permaculture crop is known to the mapper, or is unclear because of the snapshot of aerial imagery then the correct landuse tag should not be meadow, or pasture, or grassland or just grass. Where we can be specific about land operated by farms AND that the ground cover is grass, then landuse=farmland, crop=grass are the most accurate and descriptive tags to use to say that the land is owned/operated by a farm and that the groundcover is grass regardless of whether it is cut cropped, eaten in the field by livestock or is left untouched as set aside. These all need to be made reference to in the wiki page of farmland, not just reverted back to an old out of date version of the tag because a limited group of people on a mailing list don’t agree, don’t have experience/knowledge of, or haven’t defined what the actual meaning of that terminology is in their area or how it is to be used, NOT from how users choose to interpret and use that tag. The two are distinctly different.

My apologies Mateusz for the length of this reply, as you might be able to tell the subject is important to me. So finally how might the tagging mailing list users be informed to take part in the discussion on the discussion pages of the tagged items so that the range of viewpoints can be discussed with input from others with more specific knowledge on the subject matter please? I would really appreciate it if you were to respond linking in others on the tagging mailing list there please so that the subject can be discussed more fully. DerB (talk) 16:53, 16 December 2021 (UTC)

"'how this tag is used' should not stop us from really defining what the tag means" - "how this tag is used" in cases of clear situation is defining what this tag means - even if it conflicts with meaning of word. See Counterintuitive key names for some irritating cases Mateusz Konieczny (talk) 16:47, 16 December 2021 (UTC)
"I would really appreciate it if you were to respond linking in others on the tagging mailing list there please so that the subject can be discussed more fully" - I posted link Mateusz Konieczny (talk) 16:57, 16 December 2021 (UTC)
I appreciate your interest in tagging and schema design around land-use, land-cover, and other-attributes. The "Tagging" is an open forum that anyone can subscribe to, post to, and view archives of at Any email account can be used to participate and to date this has been the primary global forum to discuss and announce tagging usage and changes. For changes to tagging structure, a good practice is to follow the Proposal_process, which includes documenting the desired changes, getting feedback, and a determination of consensus by a voting period. While not perfect by any means, this proposal process opens the opportunity to gather feedback and understanding in a way that unilateral changes to wiki pages don't. I've been a long-time supporter of shifting toward a distinction between land-use and land-cover, but have learned through participating in the discussions on proposed tagging changes that arranging the world into neat hierarchical categories can sometimes be more difficult than I had appreciated. I do hope we can move to improved tagging hierarchies that don't internally conflict in the way that some current practices do, but we need to do that as a global community and bring the people, the data, and the tools along with us. --Adamfranco (talk) 17:04, 16 December 2021 (UTC)
DerB, I posted this question to the tagging list rather than this talk page because the changes were not limited just to this page and I wanted opinions from a wider audience. As Adam states above, the mailing list is open to all. So far there is very little support for these changes so it seems they will need to be rolled back. The conversation can of course continue and perhaps this will lead to some appropriate changes that can be made to these pages once a consensus is found. -- Ezekielf (talk) 17:46, 16 December 2021 (UTC)
Ezekielf, Apologies but I think the question you posed to the group missed the main point. What is the groups' consensus on 'farmland', land owned and operated by farms - does it constitute both arable and pastoral land as per the real world, or does it not? How pastoral, grazed, non-grazed grass and any other land covered by grass is tagged is a separate debate. -- DerB (talk) 18:12, 16 December 2021 (UTC)
For how OSM Wiki describes landuse=farmland the dominant tagging practice is more important than typical meaning or interpretation of word "farmland". Yes, it is irritating when this things diverge, but it happens: see Counterintuitive key names Mateusz Konieczny (talk) 18:46, 16 December 2021 (UTC)

A formal proposal, as suggested by Adamfranco would be the best place to discuss the proposed changes indeed. After all, they set out to change the meanings of well established tags to make them contradrict widely exercised mapping practice. As Mateusz says, a term gets its meaning from its use. No matter what Wikipedia claims. This is openstreetmap after all.

In a way, this is similar to the forestry debate. If a property is agricultural ground is decided by local authorities, it is not something to be easily observed at all times, neither from the air nor from the ground. BTW, growing grass for hay in my area is a very typical use of farmland (in the sense of land used for agriculture, in contrast to construction/transport/forest), so the UK meaning of a meadow not being farmland can not be maintained here.

So something along the lines of a boundary might be an even better choice than redefining landuse keys. Maybe the formal proposal should include the deprecation of "landuse=meadow", make "natural=grass" a suitable replacement, and move the administrative aspects to a completely separate domain. Consequently, "landuse=farmland" would have to be deprecated just the same; possible replacements might be "landuse=field". --Hungerburg (talk) 21:00, 16 December 2021 (UTC)

Not that I would bet even a dime on the chances of such a proposal to get accepted. But thinking about it: DerBauer would prefer to have landuse=meadow only applied to areas, that are not economically exploited - I think, I have already used "natural=grassland" for those. And, one consultation of the dictionary later, "farmland" and "field" seem to be synonyms. So the only thing missing, is a way to map land used for agriculture, regardless of crops, being that grass, corn, fruits, wine, etc. A boundary=agriculture? --Hungerburg (talk) 21:18, 18 December 2021 (UTC)

@DerBauer: The tagging mailing list isn't a "semi closed environment [...] (where only a few opinions can be expressed)". You just need to sign up, like here. See also proposal and proposal process. --Hufkratzer (talk) 17:35, 17 December 2021 (UTC)