Talk:Proposed features/Reservoir

From OpenStreetMap Wiki
Jump to navigation Jump to search

Arguments for/against Deprecation

General discussion should go in the section below. Top-level comments in this section should be used to pose arguments for/against deprecation of reservoir. This will allow voters to examine the various arguments in an organized way.

Arguments in favor of deprecating landuse=reservoir in favor of water=reservoir

Place arguments here in favor of affirming the 2011 water=* proposal and deprecating landuse=reservoir.

  • Main advantage of a new tagging scheme is that all water (except oceans) can be queried with natural=water tag, which simplifies using OpenStreetMap data in some cases (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
    • Note that synonyms are a minor problem while using OpenStreetMap data and, while irritating, are not a major barrier (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
    • This benefit will happen only after the alternative tagging is gone: as long as many reservoirs are tagged as landuse=reservoir without natural=water, data consumers must support also landuse=reservoir anyway (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
    • If someone wants to select only some specific types of water areas or have different rendering depending on a type of water area, this benefit completely disappears. (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
    • Questionable "advantage" will only happen if all waterbodies would fall into natural=water but besides lakes and reservoirs and oceans, there are a lot of other waterbody types like riverbanks which are nowhere close to being accepted into natural=water by mappers, so the whole premise of "unification of water" is wrong (Tstraupis (talk) 19:34, 12 January 2021 (UTC))
  • Hopefully, we end mostly pointless discussions over landuse=reservoir (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
  • landuse=reservoir forces mapper to distinguish man-made reservoirs and natural lakes while mapping, also when it is not possible to make such distinction (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
  • This will make a great precedence for deprecating duplicated tags that fragment the OSM data in favor of tagging schemes that are more systematic in their structure and better suited for the physical object/feature. --Mashin (talk) 14:53, 4 January 2021 (UTC)
  • The change supports a more logical hierarchy of tagging, where all water bodies share the same root tag, and then receive clarifying tags to further define the nature of the water body (which may or may not be known at the time of initial mapping). Of the two tagging schemes, using landuse is less intuitive to a new user, requires data consumers to process two types of top level tags just to see if water is present, and provides no benefits over the natural=water hierarchy. --Phidauex (talk) 18:08, 8 January 2021 (UTC)
  • Keeping all bodies of water under natural=water allows for smoother transitions when a water body changes use (which may be uncommon in some areas, but happens a lot in places with heavy development), for instance a an old reservoir getting built up on all four sides for roads and housing, then drained of water and recommissioned as a basin. The only tag change is water=reservoir to water=basin, retaining the same root tag. --Phidauex (talk) 18:08, 8 January 2021 (UTC)

Arguments for maintaining two tagging schemes for reservoirs

Place arguments here in favor of maintaining both schemes equally, with no preference for either.

  • No one is severely irritated that their preferred tagging scheme is deprecated (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))

Arguments in favor of deprecating water=reservoir in favor of landuse=reservoir

Place arguments here in favor of overturning the 2011 water=* proposal and deprecating water=reservoir.

  • landuse=reservoir forces mapper to distinguish man-made reservoirs and natural lakes while mapping (it is considered beneficial if one considers water area mapped with this distinction as much less valuable than water area with this distinction) (Mateusz Konieczny (talk) 07:05, 21 December 2020 (UTC))
  • natural=water claimed/disputed benefit of having one unified tagging only mentions oceans as an exception, but in reality there are many more waterbodies which are much less used in proposed natural=water (sub)schema, for example waterway=riverbank is almost totally overwhelming water=riverbank Tstraupis (talk) 18:38, 3 January 2021 (UTC)
  • reservoirs and lakes are different objects and people learning cartography should understand the difference, therefore it is essential from cartography/gis side to have this distinction on first level tags Tstraupis (talk) 18:38, 3 January 2021 (UTC)
  • landuse=reservoir is the original scheme, is used much longer, new scheme does not offer any serious advantages no matter how you look (classes of objects tagged are the same) so deprecating widespread tagging in favour of a similar one without a good reason would make a bad precedent for OpenStreetMap tagging Tstraupis (talk) 18:38, 3 January 2021 (UTC)
  • Hopefully, we end mostly pointless discussions over water=reservoir Mateusz Konieczny (talk) 08:36, 4 January 2021 (UTC)
  • I see reservoirs as a different class of water object because (in some cases rapidly) changing water levels are the "normal" behaviour. You should expect that most of the time only part of a reservoir actually contains water. --Lyx (talk) 20:23, 5 January 2021 (UTC)
  • Reservoirs are sometimes used for storing stuff that I would not feel comfortable calling "water", e.g. in sewage lagoons reservoir_type=sewage or tailings ponds reservoir_type=tailings. --Lyx (talk) 20:23, 5 January 2021 (UTC)


How is water=reservoir distinguished from water=basin and water=pond?

The tag water=reservoir is rather similar to water=basin (which comes from landuse=basin) and water=pond. It needs to be clear how these three tags are different. Are ponds, in fact, really just small basins or reservoirs or lakes, perhaps? --Jeisenbe (talk) 02:18, 17 December 2020 (UTC)

My opinion (and this is heavily colored by my US English view of things) is:
  • A reservoir is a dammed body of water specifically created to supply water for drinking, power generation, irrigation, or some combination of those three.
  • A basin (I usually refer to this as a catchment area) is a place for temporary overflow of stormwater (consistent with the picture in water=basin).
  • A pond is just a small lake (though, from recent tagging list dicussions -- clearly there are differences of opinions on this).
Are you thinking that I should cast a wider net here than simply the question of deprecation landuse=reservoir? --ZeLonewolf (talk) 03:13, 17 December 2020 (UTC)
A reservoir can be created for recreation or flood control or other reasons as well: as the page at Tag:landuse=reservoir says it is simple a "Man made body of stored water... Usually formed by a dam over a natural water course, water then backs up into a natural valley or depression." - so the defining characteristic is that it is a man-made water body which is formed by a dam at the head of a (semi)-natural valley, while in contrast a basin tends to have dykes (levees) on several sides, or has been dug out or otherwise re-graded.
Unfortunately a water=pond can be formed by either of these ways, with a dam or by grading, or it can even be a natural pond - though this does not match the old etymology, which came from water which was "empounded" -> "pound" -> "pond" and now does not imply artificial origin - see - so if we are going to approve the tag water=reservoir we need to decide when it should be used instead of water=pond for very small artificial lakes. --Jeisenbe (talk) 05:47, 17 December 2020 (UTC)
I made big improvements to water=reservoir which hopefully spells out what a reservoir is. Regarding reservoir vs pond, take a look at this example: way Flat River Reservoir/Johnson's Pond(!). IMO, "pond" is a colloquialism although I recall from the mailing list discussion that there was also a hydrological definition. I don't think that we're going to easily solve water=pond, but I think we can at least draw a bright line around what a reservoir is. Is what I have now at water=reservoir satisfactory for defining "reservoir"? --ZeLonewolf (talk) 13:58, 17 December 2020 (UTC)
You should include the definition and description on this proposal page, since this is what we would be discussing for approval. --Jeisenbe (talk) 16:40, 17 December 2020 (UTC)

After reading the proposal, this discussion and definition of landuse=basin/water=basin I still have no idea about...

  • How do water=reservoir and landuse=basin/water=basin differ?
    • basin is defined as "An area of land artificially graded to hold water."
    • The proposed definition for reservoir is "An artificially-created or enlarged lake, pond, or impoundment created using a dam to store water for human use "
Me: ¯\_(⊙︿⊙)_/¯
  • Why do both deal with water instead of liquids in a more general way?
It has been mentioned already that there are other large-scale containments for fluids, e.g., from mining operations or other industries, cf. reservoir_type=*. Apparently, the american(?) definitions of basin and reservoir in this context are different to how I understand them. For me the term reservoir would be very general while basin refers to either a small reservoir or at a very large scale to a depression formed by a former ocean. After looking at the possible meanings of basin in geography I would recommend not using that term at all with further clarification. It's obvious that just using "basin" as a key or value will create lots of confusion.
Similarly natural=water was always a bit of a misnomer (since not all natural=water is natural, rather the opposite in most populated areas) continuing that road and making not all water=* containing water seems consistent with the OSM mantra of creating overly peculiar mix-ups of definitions but this is of course not a good idea ;)
Nevertheless, I think the non-water bodies should be dealt with in the proposal in the sense that is becomes clear why this is only about water containment and give pointers on how to tag non-water containments, especially those using the previous reservoir_type=* keys and by clarifying the definition of and distinction to landuse=basin/water=basin.

--Stefanct (talk) 08:13, 9 January 2021 (UTC)

@Stefanct: thanks for the nudge. It seems there is already tagging for sewage ponds. I hate that they're under the natural=water scheme (as water=wastewater), though we should probably respect the significant usage there. I took a crack at the table below, what do you think? I'm thinking that this might need to be spun off as a new proposal to deal with these cases rather than jamming it in here. Appreciate any help you can give to get this right. --ZeLonewolf (talk) 17:14, 9 January 2021 (UTC)
Tag Replacement
reservoir_type=water_storage None; water=reservoir is presumed to be water storage
reservoir_type=sewage Replace with natural=water(??)+water=wastewater
reservoir_type=tailings Replace with a new tag (?) landuse=tailings

Recreational reservoirs

The proposal now says:

"However, recreation is a secondary use of reservoirs; dammed bodies of water that are dammed solely for the recreational use of the upstream water body should instead be tagged as a water=lake or water=pond"

I don't think this is how landuse=reservoir or water=reservoir are currently used. There are a number of reservoirs which are used extensively for recreation. If it's an artificial lake created by building a dam across a valley, then I don't see why it shouldn't be tagged as a reservoir even if the water is not being stored for drinking or recreation or power generation. --Jeisenbe (talk) 05:03, 19 December 2020 (UTC)

There are certainly many examples of (drinking water, hydroelectric) reservoirs which also allow recreation, but this is always a secondary usage from the examples I've found so far. I haven't found examples yet of reservoirs that are only used for recreation. Dams are expensive to build and usually involve taking away private land, so I'm skeptical that anyone is building a dam purely for the purpose of creating a body of water just for recreation. Perhaps there are reservoirs created for some purpose, but the original purpose is no longer needed? I could see a situation like that for sure. It would be great if there were some examples to look at of this case, particularly if you have some in mind. I have no strong opinions either way, just trying to understand/capture how people are actually using the tag in OSM and/or the term in real life. --ZeLonewolf (talk) 05:27, 19 December 2020 (UTC)
Also note this particular case:  Flathead Lake, which (1) has a dam which raised the lake level 10 feet, and (2) generates hydroelectric power, and (3) is considered a "lake". --ZeLonewolf (talk) 05:32, 19 December 2020 (UTC)
The Johnstown flood was caused by a dam failure on a lake which was used by a private club for fishing and recreation:
Here in the Portland, Oregon metro area we have "Lake Oswego" (formerly smaller "Sucker Lake" before the dam was built) which is a mostly artificial reservoir first built to facility navigation and provide hydroelectric power for an iron smelter, but now owned by the surrounding private residential community and used for recreation only: --Jeisenbe (talk) 23:16, 19 December 2020 (UTC)
Thanks for digging up this example! I've made the changes to reflect "recreational" reservoirs also. --ZeLonewolf (talk) 14:13, 20 December 2020 (UTC)


In Europe and Russia every artificial lake which is not elevated over the surrounding terrain through a dam is usually tagged as basin.

This applies to both rainwater retention/detention, but also firefighting, cooling (basin=cooling) and other technical basins. --Ankou (talk) 21:37, 19 December 2020 (UTC)

Could you give an example or two of this, so I can better describe the difference? --ZeLonewolf (talk) 21:50, 19 December 2020 (UTC)
Sounds pretty clear: =reservoir is for bodies of water created by a dam, and landuse/water=basin is for everything else? Also, I now understand why basin=cooling makes sense: it's different than basin=infiltration/dentention/retention/evaporation, since the water is being used to cool a power plant or other industrial facility, not being evaporated or draining away --Jeisenbe (talk) 06:28, 21 January 2021 (UTC)

Ponds and Pools

The proposal states "A lake or a pond is a naturally-occurring or artificial inland body of water that lacks continuous flow." Where there is continuous flow the body of water is called a pool, but there is nothing about pools in the proposal.

There are problems if you insist on the lack of continuous flow. I remember discussions (I don't remember where) on whether or not rivers/streams that lead into ponds/lakes should terminate at the pond/lake boundary then resume elsewhere on that boundary or if the river/stream should continue through the pond/lake. A break makes it harder for data consumers to deduce that they're dealing with a single waterway. Having the river/stream continue through the pond/lake can lead to renderers putting the waterway label in the pond/lake. There are also water bodies which are the result of outwelling from a spring into a depression in the land, which could be considered as having a continuous flow.

I've mostly mapped such bodies of water as ponds (none of them were big enough to be lakes) for lack of any better tag. In a couple of cases I was mapping a river wide enough to make it sensible to define its boundary with water=river and was able to accommodate pools that way, but that did not allow me to name the pool (many pools around here do have names because they had some intrinsic significance).

So you need to drop the "no continuous flow" restriction or add water=pool to the proposal, or add the under-documented, and somewhat misnamed, existing water=stream_pool. --Brian de Ford (talk) 18:36, 20 December 2020 (UTC)

Thanks for this limnology lesson! I've removed the phrase "that lacks continuous flow", hopefully that addresses these concerns and keeps the purpose tightly focused. --ZeLonewolf (talk) 19:52, 20 December 2020 (UTC)
That works, and has the benefit of not invalidating the tagging of a lot of existing ponds. However, it's not quite as clean and may confuse limnologists. So maybe a few words to the effect that, for OSM purposes, pools are tagged as ponds/lakes. --Brian de Ford (talk) 20:23, 20 December 2020 (UTC)
Can you provide some examples? All the pools along the Klamath River which I know are currently just mapped as waterway=riverbank areas like the rest of the river. And in Indonesia I only use water=lake when the river goes into a lake with no consistent water flow, I wouldn't use it for a small pool along a stream or mountain river. --Jeisenbe (talk) 23:24, 20 December 2020 (UTC)
The pools that I'm familiar with are swimming pools, how could this be described without confusing people? --ZeLonewolf (talk) 20:31, 20 December 2020 (UTC)
Something along the lines of: "In hydrological terms, a pool (also known as a 'stream pool' in the US) is a body of water much like a lake or pond except that there is a continuous flow of water into and out of it. In OSM these are tagged as ponds or lakes." Example --Brian de Ford (talk) 21:24, 20 December 2020 (UTC)
Perhaps something like this (under the "Note that:" section):
  • Occasionally a river or stream will form a  stream pool or  plunge pool, which are bodies of water that naturally occur along the course of the waterway. These waterbodies may either be tagged as a lake or (usually) pond if they are named or significant in size, or else they can be simply conflated with the river.
--ZeLonewolf (talk) 23:30, 20 December 2020 (UTC)


Note that it also discourages waterway=riverbank by recommending to use natural=water for all water areas Mateusz Konieczny (talk) 18:39, 20 December 2020 (UTC)

Thanks Mateusz, I've updated the text to narrow the scope to tightly focus on reservoirs, lakes, and ponds. This proposal is not intended to modify river tagging or group all water features under natural=water. Please let me know if I've missed the mark in these edits! --ZeLonewolf (talk) 19:57, 20 December 2020 (UTC)
"Mappers should tag bodies of water as natural=water." still means that supporting this proposal supports adding natural=water to all waterway=riverbank Mateusz Konieczny (talk) 21:39, 20 December 2020 (UTC)
Good catch, I changed it to be more specific: "Mappers should tag lakes, ponds, and reservoirs (which often have similar appearance on imagery) as natural=water. If they are able to further determine the specific type of waterbody, water=* can also be applied." That should keep this laser-focused on just those three types. --ZeLonewolf (talk) 21:51, 20 December 2020 (UTC)


"Mappers have preferred the use of water=reservoir over landuse=reservoir in recent years." is a bit misleading - it is a result of deprecation rule in iD that is not really exposing tags, not explaining what happened, hiding tags by default and not rendering landuse=reservoir. While tags in iD are available, typical mapper will not see them and therefore we should not treat it as preference of mappers toward specific tagging schema.

It is typical that iD validator has bug and when mappers are asked why they blindly clicked "upgrade tags" they answer "because OpenStreetMap asked me to do this".

See also

Mateusz Konieczny (talk) 06:58, 21 December 2020 (UTC)

I've added an additional monthly graph that shows mapper preference over time, and points out where the iD editor change happens. I do agree that it is likely that the iD change pushed the preference more in the direction of water=*, although there was a clear 2+ year preference that had already happened before that. With the new graphic, this should clearly detail exactly what the mapper preference has been over time and where iD had its influence. --ZeLonewolf (talk) 17:47, 21 December 2020 (UTC)

Disused open air mines

In many places where there were open air mines, the hole in the ground became a body of water. Would those be tagged with water=reservoir? --AntMadeira (talk) 18:27, 21 December 2020 (UTC)

How are they normally tagged now? --ZeLonewolf (talk) 03:46, 22 December 2020 (UTC)
There isn't any consensus, really. Sometimes they're tagged simply as natural=water, other times as natural=water+water=reservoir, lakes or ponds... there's no consensus at all. Probably because these tags were never consensual or maybe because no one ever thought about asking this. I know a 3 km radius zone where you can find 4 different tags for the same feature. --AntMadeira (talk) 04:51, 22 December 2020 (UTC)
Is there a dam in these cases? Or are they just holes in the ground that filled up with water over time? If you applied the definitions in the proposal, how would they be tagged? Sorry I've got more questions than answers, but this sounds like a very interesting case! --ZeLonewolf (talk) 05:05, 22 December 2020 (UTC)
It is a hole, below ground water level that during mine operation was actively drained. After mine operation stops it fills with water. See for example Mateusz Konieczny (talk) 08:43, 22 December 2020 (UTC)
Yes, it's just like Mateusz described! --AntMadeira (talk) 18:17, 23 December 2020 (UTC)
I don't think that water in an abandoned quarry is a reservoir, since there is rarely a dam. It's more like a basin, since the grade was changed by digging out the rock, but a basin is usually built intentionally, and usualy is graded from soil rather than dug straight down into rock like a quarry or open pit mine. Perhaps a dedicated value like water=quarry (based on landuse=quarry) would be best? 99% will be abandoned rock or gravel quarries - a few will be mines for other kinds of minerals, but these look similar to quarries in aerial imagery and we use landuse=quarry for all types of open pit mines, so water=quarry would be reasonable. --Jeisenbe (talk) 06:44, 24 December 2020 (UTC)
Apparently there is even a term for this:  Quarry lake. I agree that this ought to have a separate tag from water=lake. --ZeLonewolf (talk) 15:13, 24 December 2020 (UTC)
I agree. water=quarry would be a good adition to this scheme. In my region, there are many limestone quarries, but also many sand extraction quarries. It would be best if the wiki focused more on the extraction operations that took place and left an artificial hole in the ground and not so much the kind of soil that exists there. --AntMadeira (talk) 16:48, 24 December 2020 (UTC)
To play devil's advocate -- why not lake=quarry as a sub-key of water=lake? Don't we still think that quarry lakes are a subset of all lakes? Also, are there any other types of man-made holes that are not quarries that might fill up as a lake? --ZeLonewolf (talk) 16:54, 24 December 2020 (UTC)
Re: "why not lake=quarry" - Because a water=lake is natural (or semi-natural), while a lake inside of a disused or abandoned quarry is often clearly artificial. Encouraging tagging with water=lake would blur the meaning of that tag. Reservoirs, basins, salt ponds, wastewater treatment features, canals, ditches, drains, locks and swimming pools all have unique tags, so it makes sense that this artificial water body would get it's own tag too. --

stream pools

> Occasionally a river or stream will form a stream pool or plunge pool, which are bodies of water that naturally occur along the course of the waterway. These waterbodies may either be tagged as a lake or (usually) pond if they are named or significant in size, or else they can be simply included as part of a river polygon.

I don't see any reason to change the way these are currently tagged, which is water=stream_pool documented at They are not big or significant enough to be called lakes and aren't man made so aren't ponds. --Aharvey (talk) 05:51, 22 December 2020 (UTC)

After seeing your examples on the tagging list, I now agree that mapping these as water=stream_pool is fine when they are along a waterway=stream feature. For larger "pools" along rivers, I would just map these with waterway=riverbank (or water=river if you must) like other parts of a river. --Jeisenbe (talk) 06:46, 24 December 2020 (UTC)
I tend to agree that either way is fine. In either case, I've left it squarely out of the scope of this proposal which will hopefully keep this squarely focused on reservoirs, lakes, and ponds. Please let me know if I haven't drawn that line well enough! --ZeLonewolf (talk) 15:06, 24 December 2020 (UTC)
Yes, it's fine not to mention it, because it's only for semi-natural and natural pools along small streams, so should not be confused with reservoirs, basins or lakes. The tag water=stream_pool is now documented at Tag:water=stream_pool. --Jeisenbe (talk)

This proposal needs to be split into two

As this stands I'd be voting no. Not because I think landuse=reservoir is any good, rather I don't agree with the proposed definitions for the water=* values. Other people may also vote no for the opposite reason. Better to split this into two proposals and the second one (clarifying definitions of some values for water=*) can be written on the basis of the result of the voting on the first. Adavidson (talk) 05:41, 1 January 2021 (UTC)

What I'm trying to say is that it may be easier to get a vote on the deprecation first and then put the rest of it up as a separate proposal. Otherwise you are not going to know if a 'no' vote is because of that or the clarification part of the proposal.

My original version of this proposal was a simple deprecation of landuse=reservoir. However, there is a concern that if we are going to approve water=reservoir that it should have a clear definition, and that definition needs to differentiate it from water=lake and water=pond (see @Jeisenbe: comments above). I'm not opposed to either combination or separation, just trying to figure out how to thread the needle on community support and find the points of consensus.
going to approve water=reservoir ? approved in 2011 --Marc marc (talk) 22:21, 31 January 2021 (UTC)
Believe me, I find it as ridiculous as you do that the community has not accepted that 2011 vote as a deprecation of landuse=reservoir. See this discussion along with the page history of landuse=reservoir. Multiple attempts to mark the tag as deprecated in the wiki have been reverted due to disagreement over whether landuse=reservoir was actually deprecated. I crafted this proposal in order to put that question to rest, once and for all. --ZeLonewolf (talk) 23:34, 31 January 2021 (UTC)
Main problem with that deprecation is that proposal itself redefined what "deprecated" means - see '"Deprecates" means "is equivalent for all purposes to"' Mateusz Konieczny (talk) 06:43, 1 February 2021 (UTC)
yes the depreciation of tags is a sometimes absurd process (I think especially of the chicken and egg syndrome: some people do not want to depreciate a tag because it is used because it is not depreciated)
But the tag water=reservoir being approved, you don't need to come back to it, on the contrary you should even lean on it for example with "depreciate landuse=reservoir in favor of the tag water=reservoir approved in 2011".
your proposal is multiple :
- depreciate landuse=reservoir (this alone will make the conservatives scream)
- depreciate reservoir_type (see this issue)
Thanks for the suggestion to increase the emphasis on the prior approved proposal, I've edited the lede to include this. Also, I will re-raise the question of separation on the tagging list once the ongoing proposals for tailings ponds and evaporation basins are complete.
- improve the wording of what a lake or pond is (and in my opinion, this has nothing to do with a proposal, it's a thread for tagging mailing and if appreciated, modify the wiki... on the other hand the proposed definitions are in my opinion too long and the circular references "a lake is not a pond that is not a lake") should be avoided --Marc marc (talk) 08:40, 1 February 2021 (UTC)
There is no circular definition between lake and pond in this proposal that I am aware of, and I challenge you to point out exactly where such circular definition may be present. I did find a circular definition loop between lake/pond and reservoir, which I've corrected by changing the lede in the reservoir definition section.
With regard to length, the definitions are as long as they are in order to address the extensive community feedback that exists on this subject, both on this talk page and in the many discussions on this topic that have occurred in the mailing list. Most recently, I would point you to this discussion where the differences between "lake" and "pond" are discussed in detail. Unfortunately these classes of object are necessarily fuzzy when it comes to their definition and the distinction between them. The "long" definitions aren't me spouting off, they're addressing the aggregation of community concerns that need to be addressed in order to achieve consensus.
I gather from your comment that you favor deprecation of landuse=reservoir but oppose the currently-proposed definitions of water=pond and water=lake. Could you elaborate on the objections that you have to these definitions so that I might try to address them? (edit: also, if you could clarify whether your objections extend also to the definition of "reservoir"? --ZeLonewolf (talk) 15:27, 1 January 2021 (UTC)
proposals in the last 2-3 years with a size comparable to this one, have all been rejected. i think you should strongly cut, maybe in 3-4, as you have done about tailings --Marc marc (talk) 22:21, 31 January 2021 (UTC)
Do you have specific objections to what is proposed here, or are you merely commenting on the likelihood of success? I do not think that this is a "large" proposal and I suspect the community would find it tedious to have individual votes on lake and pond. --ZeLonewolf (talk) 23:34, 31 January 2021 (UTC)
both : objections on part of what is proposed here + too big = too many reasons that could lead a contributor to vote against, including simply because it is too big
my suggestions :
1) cut/remove any vote about the depreciation of an already depreciated tag
2) depreciated reservoir_type : I'm in favor but
I would prefer to see the proposal on evaporators accepted first, eliminating an additional value.
there is a lack of consistency depending on whether the reservoir is open (water=*) or closed (content=* is used for man_made=tank and storage_tank).
there is also an inconsistency between water=* describing the importance of the water flow (river, ...) and its quality (sewage).
This double issues need to be fixed
3) changing existing tag definitions with definitions that are a bit heavy and that refer to each other (A is not B which is not A).
In the same way to say "A concerns objects often having A in its name" is not really a definition. At best it is a hint, which should support another criterion.
split the proposal with only the point 3, look with the tagging list to make a less circular definition
--Marc marc (talk) 07:34, 11 February 2021 (UTC)
The trigger of that proposal is that Proposed_features/Water_details#Deprecation was bungled. Note 'Deprecates" means "is equivalent for all purposes to"' that redefined deprecation in that proposal resulting in a widespread confusion (word "deprecation" means something different). This proposal wants to confirm that deprecation and deprecate as in "encourage to replace and remove", not something else but confusingly named Mateusz Konieczny (talk) 07:38, 11 February 2021 (UTC)
With regard to #2, this is already my strategy, and I'm sorry if that approach was not clear. I am planning to move forward with this proposal only after all subordinate proposals have been resolved. --ZeLonewolf (talk) 13:45, 11 February 2021 (UTC)

Water behind a weir

waterway=weir suggests one use reservoir to tag water behind a weir. This appears to contradict this proposal. Invidious (talk) 19:57, 3 January 2021 (UTC)

I didn't read it the same way. Reservoir was just in the "see also" section. I tweaked that to make it a bit more clear on that page. The proposal that created waterway=weir has no mention of reservoirs. --ZeLonewolf (talk) 05:52, 5 January 2021 (UTC)

Basin vs. pond

  • landuse=basin|pond is a totally different object which cartographers would never want to select together, they are always selected for different ranges of scales and usually have different symbolisation therefore by selecting "natural=water" cartographer will always get not what is expected/needed Tstraupis (talk) 18:38, 3 January 2021 (UTC)
    • "is a totally different object which cartographers would never want to select together" - in many maps based on OSM data it is actually selected together, plenty of map makers are not making this distinction Mateusz Konieczny (talk) 08:35, 4 January 2021 (UTC)
      • Note: this is an opinion of Mateusz who is not familiar with basic concepts of Cartography/GIS. Those familiar with basics of cartography would want them separate. There are a lot of low quality data visualisations created using OSM data (which is noted in a lot of scientific papers on cartography), keeping cartography in mind is aiming to increase at least basic cartography skills of all interested. Cartography is an art/science of communication, not vectorisation/re-creation of ortophotographic maps (Tstraupis (talk) 11:20, 4 January 2021 (UTC))
        • You missed completely that I explicitly stated "plenty of map makers are not making this distinction" that includes all people making maps, not only cartographers as defined by Tstraupis. For example orienteering maps are not making such distinction at all, because they show a different properties more relevant in a given context. Mateusz Konieczny (talk) 11:36, 4 January 2021 (UTC)
        • "who is not familiar with basic concepts of Cartography/GIS" is a lie. Not sure whatever it is deliberate slander or just random baseless insult. And note that GIS and Cartography is not limited to Tstraupis-approved maps - it includes also things that are not even producing maps Mateusz Konieczny (talk) 11:36, 4 January 2021 (UTC)

        @Tstraupis: Your comment could easily be interpreted as a form of gatekeeping or personal insult, either of which would undermine your argument. I hope it is clear that your frequent ad hominem comments and appeals to authority here and on the associated tagging mailing list thread have been trying peoples' patience.

        For better or worse, it will always require some attention to tagging nuances to produce a high-quality rendering based on OSM data, regardless of this proposal and regardless of what one considers to be the standard for quality. To the point made above, openstreetmap-carto renders landuse=basin basin=retention the same as landuse=basin natural=water and basin=detention/infiltration the same as natural=water intermittent=yes. [1] That's not to say that it's the only way to render a basin. But it does match the USGS practice of symbolizing perennial versus intermittent bodies of water, regardless of the method of construction. [2] Perhaps there are more respectable cartographic products than USGS topo maps, but it's one useful point of reference for a result that an OSM data consumer would find acceptable.

        – Minh Nguyễn 💬 09:38, 5 January 2021 (UTC)


Resolved: Separate proposal spun off for tailings pond tagging

There should be some path for proper tagging of reservoir_type=tailings. Tagging that as natural=water is a poor idea Mateusz Konieczny (talk) 22:26, 5 January 2021 (UTC)

meaning of water=* : about the object (reservoir) or the content (water, sewage) ?

water=* is mixing info about the object (lake pond reservoir) and the content (clean <> sewage/wastewater). for man_made=storage_tank, content=* is used to talk about the content. it seems more coherent to me to also use this key when it concerns a reservoir in stead of the current mistake water=wastewater.

Sewage ponds

What about man_made=sewage? Something B (talk) 09:02, 26 February 2021 (UTC)

See also man_made=wastewater_basin Something B (talk) 13:37, 26 February 2021 (UTC)


This is only a minor detail but I think you should mention man_made=reservoir_covered under "Additional clarification". I would add it myself but it feels like a proposal is almost a personal page. :) --Stefanct (talk) 11:12, 14 December 2021 (UTC)

Not wanting to hijack the topic, I think man_made=reservoir_covered is an unnecessary tag, because you have the key covered=yes. --AntMadeira (talk) 15:42, 14 December 2021 (UTC)
And what would the main tag be? Certainly not water=* since these reservoirs do not show any water surface at all but are completely underground. --Stefanct (talk) 15:48, 14 December 2021 (UTC)
The main tag would be man_made=storage_tank + contents=water. But where does it say that they're completely underground? --AntMadeira (talk) 15:56, 14 December 2021 (UTC)
That's implicit but very clear in all examples I know of. The German wiki page states it explicitly (but I consider such deviation as an error in the translation). Using man_made=storage_tank alternatively for those has been introduced many years later. But that's not too relevant here anyway: The important point is that the proposal should mention how it is related to the real-world objects in question. --Stefanct (talk) 04:58, 27 December 2021 (UTC)

Some general comments on the proposal

While i appreciate the effort of trying to improve the tagging of waterbodies this proposal seems to try to conflate at least three different endeavors into one, which i don't think is a good idea. These three things are:

  • suggesting a new tagging/sorting out different overlapping taggings for artificial waterbodies in OSM,
  • redefining taggings with significant use (water=lake and water=pond) (a) in a predominantly negative fashion, i.e. based on what it is not, (b) in significant conflict with existing tag use and (c) in an English language culture specific way based on the assumption that all languages of the world have exact equivalents to the English language terms 'lake' and 'pond',
  • re-framing the proposal process as an authoritative instrument speaking for the whole OSM community ("The OSM community recommends").

The first point is one that makes sense in a proposal, the other ones less so. I would suggest to split these things and discuss them separately. Working on more clearly documenting what water=lake and water=pond mean respectively is a good idea but this would need to be based on existing use of the tags and follow the requirements of good tagging of physical geography features, i.e. having a positive definition, being globally uniform based on verifiable observations of the physical geography and being largely compatible with existing use of the tags. This would not really really require a proposal, improving the tag pages for these tags would already go a long way.

--Imagico (talk) 13:21, 14 December 2021 (UTC)

Definition is too restrictive

Not a huge issue, but I suggest broadening the definition of reservoir somewhat.

> A reservoir is an artificially-created or enlarged body of water created using a dam to store water for human use. Dam, in this context, specifically refers to a man-made structure which allows for human control of the reservoir water level and downstream water flow.

This focus on dams rules out a lot of reservoirs. I know of a fair few reservoirs where the high water lign is the same as when the water was a lake, but a new low water lign has been introduced by a tunnel that can drain the lake. One example from Norway: The reservoir Ljosavatnet is being regulated by this drainage tunnel:

Suggestion for rewording:

> A reservoir is a body of water utilised to store water for human use. For a body of water to be considered a reservoir, there needs to be a man-made structure which allows for human control of the reservoir water level and downstream water flow. This structure can be a dam, a drainage tunnel, or both.

--Harahu (talk) 12:24, 22 January 2022 (UTC)