landuse/amenity suggestions needed for private estate/facility area

Looking for suggestions on how to properly tag an area for the use-cases below.

The goal is to “protect” known areas where roads cannot be residential but only service/unclassified.
The same way residential areas tagged as landuse=residential make it clear you expect only residential/unclassified and driveways/alleys.

amenity=university ?
landuse=religious ?


  1. For the authority’s office area, office=government
  2. amenity=research_institute
  3. amenity=university
  4. It’s complicated, wiki seems to suggest landuse=religious for the temple ground and amenity=place_of_worship for the temple building. However, it appears to be more common in Asia (including Thailand) to tag amenity=place_of_worship to the Buddhist temple ground, because the Buddhist temple is intended to be perceived as a group of buildings, as opposed to a church or mosque where we can tell what is the main building. Also most temple buildings have no specific name and are referred to by their function.
  5. maybe landuse=farmland and landuse=farmyard. AFAIK, there is no tag for a single owner’s entire land plot.
  6. office=government to the building and boundary=national_park for a whole National Park area. There is no specific tag for that area.


Thanks for the overview :slight_smile:

One common issue I face is temples are often found within a large forest/hill area (sometimes surrounded by walls, and gates) but the actual temple buildings are within a small compact area in the middle/top. Typically roads leading to the temple buildings grounds inside the large forest/hill area are tagged as service.

I wanted to find a way to tag the large area to inform mappers. Would it be ok in that case to tag the entire area (like in the example I gave) as amenity=place_of_worship ?

I’d tag this entire area as amenity=place_of_worship.

The place_of_worship identifies the landuse. A natural=wood can be mapped as an area to identify the landcover. For OSM-carto rendering, if the natural=wood is smaller than place_of_worship, the green area will appear in front of the temple area.

I’d probably use landuse=religious rather than tag the whole area as a place of worship. Religious landuse is meant at least in a western context to deal with things like: graveyards associated with the place of worship; monasteries which will have a distinct church and the rest of the monastery may have other purposes (for instance a school, brewery, farm etc).

Don’t get too hung up about multiple overlapping landuse as these can usually be resolved either directly by need, or during necessary post-processing for a specific use case. In other words if the place is a monastery we can use religious for the whole area, farmland for areas within which are farmed & so-on.


I would agree. Like for villages/towns, I would have expected a similar tagging method:

  • area with landuse=religious to define the whole temple grounds (indicate highway=service should be used)
  • a point with amenity=place_of_worship to define the “center” of the temple grounds.
  • main buildings tagged as building=yes

I am pretty sure a temple name “Wat --------” defines the entire area and not the main praying building.
Often you will find pagodas, residence quarters, farming areas, water tanks spread out around the land.

This is a difficult issue that may need to be addressed on an international level, as this style of Buddhist temple complex may be found in many countries.

If we stick to the definition in wiki, it’s obvious that we should tag the worship building with amenity=place_of_worship and a temple grounds with landuse=religious, but we should be concerned about this issue because OSM was developed from a more Christian than Buddhist perspective. One thing, OSM mappers and users value an amenity=place_of_worship more than a landuse=religious. This is not a problem if we are thinking about a church or cathedral (as well as a mosque and even some Buddhist religious monuments), we are thinking of the building rather than the land, in this case a landuse=religious is useful for tagging the area around the church to define their boundary. However, if we strictly adhered to this standard, there would be a cluster of amenity=place_of_worship in the small area. Most of them would have the same name (Wihan, Ubosot, etc.) and would clog the map.

I looked through a taginfo for some information, and this is what I discovered. (Only way and relation)

religion=buddhist + landuse=religious = 1,457
religion=buddhist + amenity=place_of_worship = 27,064
religion=buddhist + building=* = 17,743

If we assume that all building=* are tagged together with amenity=place of worship (which is impossible), there would be at least 9,321 religion=buddhist + amenity=place_of_worship that are not buildings (There should be more, but I can’t find a way to combine 3 tags to see statistics.). As a result, the Buddhist temple grounds are currently tagged with amenity=place_of_worship at least 6.4 times greater than landuse=religious. Some of them may be tagged with both amenity=place_of_worship and landuse=religious as I’ve seen in Japan.

To keep things simple, perhaps we don’t need to identify a distinct place for amenity=place_of_worship and landuse=religious, because they can be the same! If we don’t tag the building as an amenity=place_of_worship and instead use the area, there shouldn’t be one area inside another, as that would be too complicated and difficult to find a boundary for each.

The suggestion to tag amenity=place_of_worship as a node solves one problem while creating another. It’s unhealthy in some ways since it doesn’t adhere to One feature, One OSM element principle. Furthermore, the temple area should be useful for some user’s purpose; separating them from their boundary may result in data loss. Also, mapping it as a node appers to be a kind of quickly mapped that will need to be cleaned up later.

If we wish to designate the complex’s center (which is not yet described in the wiki), we can create a proposal for a relation for Buddhist temple complex, which can provide a role for temple center (main building?). However, because the center is difficult to locate and somewhat subjective, and there is no physical evidence on the ground, simply not mapping them may be the best solution. OSM users can still use the temple mapped as an area for navigation. A GPS software should be able to take you to the temple’s gate and that’s it. If users want to go to a specific building, they should type the name of that building into the software to ensure it will take them there.

P.S. While the landuse=religious wiki page uses Asakusa Temple Complex in Japan as an example, ironically it is still tagged with amenity=place_of_worship in OSM (without landuse tag). :slight_smile:

@nitinatsangsit Thanks for the thorough explanations,

I agree it would make sense to look at the most common usage to decide how a certain object/area should be mapped.

However, most (new) mappers will not compute statistics but simply look at how it’s been done in their area of mapping.
When there are inconsistencies in the way it’s done, mappers will either not map, or follow one method, or come up with a new one, leading to more inconsistencies.

I understand we cannot cover all cases in the guidelines, but I think temples (and villages/towns) are a core element of maps in Thailand and should deserve a small section in the official wiki.

I agree, and I hope we can discuss it and reach an agreement soon.

Could we please reach a general consensus on how to tag temples in Thailand?

I start seeing older changes reverted because of different mappers conflicting approaches (tags values remains identical), which I find very unproductive.

The two most common approaches for tagging temples have been so far:

A) node inside temple area tagged as religion=buddhist + amenity=place_of_worship + name=… + name:en=…

B) main building tagged as building=yes + religion=buddhist + amenity=place_of_worship + name=… + name:en=…

To me, (A) approach is the most natural.

  • a temple name does not necessarily represent a single building (many large complexes have multiple praying halls)
  • node could be freely placed on the main building, parking area, or center (if main area unknown when mapping with imagery)
  • match approach on how most places are tagged (e.g. town/villages)

What is your preference?

I use version A because it’s fast and easy, and contains all relevant information.
Better would be to draw the temple area and then add the tags as with A; also draw the buildings (just tagged as buildings). But that takes more time, and offers hardly any extra value.

@nitinatsangsit: going back to this old comment of yours, could you please clarify your reasoning? Most users including you map villages/towns using 1 or multiple areas (landuse=residential), a node placed in the “center” of the town to document the name and type of town, and further buildings can be added inside the larger area. I really do not see any difference with temple complexes, or any other type of places, including large areas like hotels, schools, or golf courses.

In general, most tags can be used for either a node or an area, and both are acceptable; however, tagging both with the same tag is considered duplication.

BTW, I think my previous comment was misconstrued, and I’d like to clarify. I have no objections to tagging a temple as a node, and as Bernhard said, it is a quick method to map it. My concern is that the same information should not be tagged on both the node and the area at the same time, and that the tag should not be removed from the area for placing on the node instead.

place=* is a specific case that is generally used for only a node, whereas an area should be tagged with boundary=administrative. Though some place=* is also used for an area, it is much less prevalent. For routing, place=* is used to tag the settlement center, which is sometimes debatable, whereas boundary=administrative is used for addressing of places, such as for Nominatim search. As a result, the name=* value tagged on the node and area differs slightly. This seems to be a duplicate, however it is a widespread practice, and I have found no other tags where a similar method is utilized.

Finally, I favor your A approach, but if the boundary can be traced, mapping it as an area is preferable.


It is not required to map everything in extreme detail as individual areas. In many cases, a simple node provides all the required details.

The wiki has some hints in case you want to also map the landuse (like the area of a temple with parking lot and multiple buildings):

But there is no requirement to map all these details. In many cases simply a node with the name of the temple is providing already a good amount of information to the map.