Proposed features/Sinkholes refinement

From OpenStreetMap Wiki
Jump to navigation Jump to search
Status: Approved (active)
Proposed by: Penegal
Tagging: sinkhole=estavelle, ponor, doline, bluehole, pit
Applies to: node, area
Definition: Map the different types of sinkholes

Rendered as: mainly a black area when an area, and a single or double chevron or arc depending of sinkhole type when a node
Drafted on: 2017-01-29
RFC start: 2017-09-09
Vote start: 2017-09-26
Vote end: 2017-10-24


Give a way to distinguish the different types of sinkholes.


There are different types of sinkholes:

a sinkhole connected to a water stream or a water body, either intermittent or permanent, which drains the watercourse or body, or feeds it, according to the relative levels of the water stream or body and the aquifer; that would describe the cavities connected to a polje, in conjunction with
a sinkhole draining water, either temporarily or permanently, from a water body or a water stream; can be associated with polje as the draining orifice, and with rivers and streams, leading to their complete, permanent, or partial, temporary disappearance, and later resurgence by a
karst spring
a resurgence, either temporary or permanent, of water drained by other sinkholes; can be associated with polje, either as the outcome of its ponors or as its income;
blue hole
a vertical, underwater sinkhole, often used for scuba diving, named after the deep blue color they mainly show;
a sinkhole clogged up, sometimes only partly, with earth or debris, and which, when filled, generally looks like an old crater with flat, fertile ground; when fully clogged up, it often contains a pond from trapped rainwater, no longer able to escape by the clogged bottom; when only partly clogged up, the pit still reaches the surface; a doline can form around any of the above features;
a sinkhole without any of these characteristics; a simple hole, opened in the ground, which may present a waterbody at any depth; that includes cenotes, whose waterbody is often larger than the orifice in the ground.

There are several motives to map the different types of sinkholes:

  • the more evident: some are dangerous, especially ponors, but blue holes, dolines and estavelles can also be, depending of your activity or the layout;
  • some are renowned as touristic destinations, especially the blue holes which can be rather spectacular, but all of them can be a tourist attraction, if only a local one;
  • they can be a useful landmark, especially blue holes and dolines;
  • they can be looked for, apart from tourists, by scuba divers, speleologists, geology students, or simply geology enthusiasts;
  • they can explain the sudden disappearance of water streams or bodies: the latter are commonly expected to only feed larger streams or bodies, not disappearing in the middle of nowhere; mapping the sinkhole in which the loss occurs with its type would explain this disappearance. There are even some features which are only formed by sinkholes, like karst fensters, i.e. karst springs which immediately lead to a ponor, which would form on the map a little, cut-off stream, leaving this little blue dash unexplained if the karst spring and ponor aren't mapped.


Sinkholes examples
Example Proposed modelling
Estavelle-Trou des glanes.JPG Resurgence 2.jpg natural=sinkhole+sinkhole=estavelle+refitted=yes
Doline entree debain.jpg Doline entree crue 2013.jpg natural=sinkhole+sinkhole=ponor+refitted=yes
P1180254.jpg P1110580.jpg P1110578.jpg natural=spring+intermittent=yes+karstic=yes+refitted=yes
Doline Mardelle-Oberdorff.JPG natural=water+water=pond+sinkhole=doline
Jakovica, Logatec - požiralnik Pod stenami na Planinskem polju.jpg natural=sinkhole+sinkhole=ponor+refitted=yes
Golondrinas Entrance Rappelling.png natural=sinkhole+sinkhole=pit


Proposed tagging
Tag Applies to Comments
sinkhole=estavelle node When an estavelle is not in the bed of a stream but connected to it with a side stream, this side stream, as it can flow both ways according to the current levels of the aquifer and the stream, should be tagged with flow_direction=both, to highlight the unusual flow variations. As the stream connected to an estavelle is often intermittent, this stream should be tagged intermittent=yes accordingly.
sinkhole=bluehole node area This tag, when used on an area, does not need to be an inner polygon in a multipolygon relation, or otherwise disconnected or excluded of the surrounding waters, as it is included in them, owing to its nature.
sinkhole=doline node area
  • When the doline hosts a pond or another water body, which is common, tag it natural=water+water=+sinkhole=doline, to represent the fact that the sinkhole feature is a characteristic of the water body.
  • When the doline is also an estavelle, a ponor or a karst spring, tag the orifice as the sinkhole using according sinkhole=* or karstic=yes, as they better describe the effects of the doline on its surroundings than only saying it is a doline, and then use natural=cliff or related tags to describe the doline effects on its surroundings.
sinkhole=ponor node Ponors does not always have a well-defined orifice, and often are few square meters areas filled with permeable alluviums or simply punctured by dozens of holes, whose location will only be clear by the sucking noises they make while water drains through them, or by the orientation of aquatic plants, curved in their direction by the water flow. Such area being also a ponor, it should me mapped as one unoccupied orifice.
karst=yes node Used in conjunction with natural=spring, would extend the current modeling of springs instead of duplicating it; the wiki page Tag:natural=spring mentions using water_characteristic=resurgence, but there is only four springs tagged as such, and water_characteristic=* seems fitted to physical parameters of the water, not its origin; when the spring emerges from a pit, use also sinkhole=pit. This tag would warn about the following possibilities:
  • chaotic, dangerous surroundings, because of the karst effects on the soils;
  • quick and strong throughput variations, often related with surrounding rainfalls (if applicable, use intermittent=yes);
  • average water quality, as such springs are often contaminated with surface pollutants.
sinkhole=pit node area When the downward orifice is the only feature, i.e. there is no water flow and it is not filled with debris, even partly; if the sinkhole is known to be a cave entrance, map it as natural=cave_entrance in lieu of natural=sinkhole in order to reflect that the cave entrance is a pit, but keep it also tagged sinkhole=pit to reflect its dual nature. If the sinkhole is a cenote, map the underground waterbody as it is, with natural=water+layer=-1+whatever.



When several sinkholes are known to be linked, i.e. they belong to the same named ensemble, or mining sinkholes are caused by the same network of cavities, these elements should be grouped with relation:site. For instance, ponors and karst springs are known to form karst fensters, i.e. a karst spring whose emissary is quickly, often after a few meters, swallowed by a ponor. Such features, as the Trampot karst fensters (fr_FR), are known sites, which justifies using a relation:site.

This is also to be used on separate features belonging to the same karst system: even when spread over kilometers, such features still belong to the same system, which is what a relation:site is made for; in this case, use name=* on the site relation with the name of the karst system, if named. Once modelled, these relations between karst features can be used to explain, for instance, why a flood in a losing stream predates the activation of an intermittent karstic spring.

There can exist, along the "underground way of a river" (this kind of designation being merely a convenient, simplistic way of representing a karst system for mere mortals), shafts allowing one to observe the karstic system, either for monitoring water levels inside or to allow speleologists to visit caves belonging to the karst system; such shafts should be tagged according to their nature, and be included in the relation:site of the karst system to reflect their mutual connection.

Note that, as the effect of the members of the relation:site modelling the karst system directly stem from their nature, there is no need of using the relation roles; for instance, a natural=sinkhole+sinkhole=estavelle will inherently be both an inlet and an outlet of the karstic system described by the relation:site it belongs to, so giving it a role in this relation would be pointless.


When a karst spring is known to be linked to some ponors along another stream (c.f. the water of the spring of the Loue, in France, is known to come from another river, the Doubs), but the exact location of the ponor is unknown (for instance, they are too numerous and merely render the bed permeable across kilometers), then the spring, and only the spring (the resulting river being frequently accepted as a different river than the losing one), should be tagged karst=yes and included in the losing stream's relation:waterway with the resurgence role.

When the stream connected to an estavelle is merely a connection of the estavelle with a neighbor waterway, this side stream should be included in the linked stream relation:waterway with the side_stream role; if applicable, the estavelle itself is to be included to a relation:site describing the linked karst system.

Note that you should not try to map the underground streams between ponors and resurgences, as

  • if the goal is to describe the connections between them, this aspect is to be taken into account by relation:site, so the modelling of these streams would merely be a duplicate;
  • they practically can't be accurately mapped, being underground, without physical access for speleologists on many kilometers, and often flowing through a permeable stratum, which doesn't make it a single channel flow to be mapped;
  • their paths, belonging to a karst system, are complex, ever changing, and probably not permanent enough to justify an inclusion in OSM.

Notes on tagging

No way

These tags should not be applied to ways; if the sinkhole borders are unclear, tag the sinkhole it self as a node, and the well-defined borders as they are, for example using natural=cliff.

Distinguish cause and consequences

A doline can contain any of the other features, like a pit or a ponor, as it frequently forms around them; if the doline is large enough to be considered a separate feature, map the other feature inside as the sinkhole and as the last or first node of the linked waterstream, in order to link it with the sinkhole in the OSM database. Then, map the surrounding of this sinkhole as it is, for instance using natural=cliff.

Water bodies

When a water-related sinkhole feature is inside a water body, map it at its precise location if possible with natural=sinkhole+sinkhole=; if the water body is itself the sinkhole, then tag it also with sinkhole= to highlight the fact that it is in the first place a water body, the sinkhole being a secondary characteristic; anyway, the sinkhole feature, either itself or the body containing it, should be properly linked with the related waterways.

Anthropogenic influence

Layout altered by human hand

When the sinkhole has been refitted, for example with a grid preventing unauthorized access, or a culvert to prevent its clogging with debris, it should be also tagged with refitted=yes to reflect these man-made changes.

This refitted=yes will of course be available for other refitted natural features, such as cave entrances, or springs, and will not be specific to sinkholes. This proposal suggests its introduction as it seems useful for sinkholes, but it can also be useful for many other natural features, so its inclusion in this proposal is merely opportunistic and, if approved, this tag will live its life apart from sinkhole features.

Anthropogenic sinkholes

Anthropogenic effects on the environment include sinkhole formations, for instance pinges which are caused by collapsing abandoned underground mines, or ruptured sewer pipes washing away soil until a sinkhole forms. These can be pretty spectacular, such as the Lake Peigneur or the 2007 and 2010 Guatemala City sinkholes, these two last being examples of piping pseudokarst. For these sinkholes, use anthropogenic=yes; if their source is known and modelled in OSM, link the sinkhole with its cause using relation:site.

Relation with landuses

Typically, as a sinkhole is neither a landuse=* nor a natural=* feature as natural=wood or natural=scree, it should be an inner member of the relation:multipolygon describing the surrounding area, which would also ease its rendering.

Applies to


  • Λ could be used for karst springs, v for ponors and a superposition of them for estavelles, all in light blue with a border in a deeper blue to distinguish stream-related sinkholes from others water-related features; a separate rendering between karst springs and other types of springs should not be relevant for the general purpose map, but rendering ponors and estavelles should be if their usage begins to be significant;
  • V should be used for pits and blue holes; the symbol should be brown for pits and, for blue holes, a deeper blue than the one used for water streams and bodies;
  • the dolines should be rendered with a brown downwards arc ⌣;
  • if the sinkhole is mapped as an area, the sinkhole icon could be rendered at the middle of it, and the area rendered black behind the icon, to highlight the nothingness of the sinkhole, excluding blueholes which could be then rendered as a deep blue area, to reflect their appearance and its difference with surrounding waters.

Features/Pages affected

See also

Proposed features/natural=cave


As karst systems are very complex, ever changing, underground natural features, it is highly likely that some features are not known and that trying to map them accurately would be futile and out of the scope of OSM. This proposal acknowledges this fact, by describing only the modelling of visible features and the proved connections between them. The features described in this proposal are within the reach of any reasonably educated person with enough common sense, at the very worst helped by asking on the OSM mailing lists, forums or on with some pictures or a description of the feature. Regarding the connections between these features, apart from accidental discoveries —like the link between the Loue and the Doubs in France, discovered when a fire in a spirit works alongs the Doubs forced the dump of coloured fluids in the river, coloured fluids which appeared some days later in the spring of the Loue, establishing the link between them—, they can only be known with geological or hydrological studies, often expensive and time-consuming; that means that the known connections betweens these features have been established at a degree of trust high enough to be reasonably included in OSM. Obviously, such OSM connections must not be made roughly, only when reasonably established.


Voting closed

Voting on this proposal has been closed.

It was approved with 16 votes for, 0 votes against and 0 abstentions.

The vote of Erkin Alp Güney, despite favorable, was not yes, so, as its counting would not change the vote result I prefer treating it as spoilt.

  • I approve this proposal I approve this proposal. usefull. --DF45 (talk) 23:28, 13 October 2017 (UTC)
  • I am in favor of this proposal.. Erkin Alp Güney (talk) 19:05, 26 September 2017 (UTC)
  • I approve this proposal I approve this proposal. --Waldhans (talk) 20:19, 26 September 2017 (UTC)
  • I approve this proposal I approve this proposal. --Javiersanp (talk) 10:53, 27 September 2017 (UTC)
  • I approve this proposal I approve this proposal. --HillWithSmallFields (talk) 13:08, 27 September 2017 (UTC)
  • I approve this proposal I approve this proposal. Very nice one :) -- Fanfouer (talk) 16:47, 28 September 2017 (UTC)
  • I approve this proposal I approve this proposal. --Hufgardm (talk) 09:21, 29 September 2017 (UTC)
  • I approve this proposal I approve this proposal. Nice work! --Dr Centerline (talk) 02:03, 3 October 2017 (UTC)
  • I approve this proposal I approve this proposal. --Michalfabik (talk) 08:06, 8 October 2017 (UTC)
  • I approve this proposal I approve this proposal. I’ve never mapped a sinkhole, but seems ok, and if you need these tags, fine —Dieterdreist (talk) 09:56, 8 October 2017 (UTC)
  • I approve this proposal I approve this proposal. There are lots of sinkholes in my region (karstic area). Only the naming of tag karst=* or karstic=* should be agreed upon. --Bielebog (talk) 22:18, 9 October 2017 (UTC)
  • I approve this proposal I approve this proposal. I appreciate the refinement of the existing key. Thanks! --Vademecum (talk) 22:37, 12 October 2017 (UTC)
  • I approve this proposal I approve this proposal. Well thought through proposal for a scheme that would allow me to tag features I wasn't able to tag so far. Thank you --Claudius (talk) 11:21, 13 October 2017 (UTC)
  • I approve this proposal I approve this proposal. --JyriK (talk) 13:51, 13 October 2017 (UTC)
  • I approve this proposal I approve this proposal. --DenisCarriere (talk) 13:54, 13 October 2017 (UTC)
  • I approve this proposal I approve this proposal. --Thetornado76 (talk) 20:15, 13 October 2017 (UTC)
  • I approve this proposal I approve this proposal. --Nospam2005 (talk) 20:41, 14 October 2017 (UTC)

Secondary vote: karst=yes or karstic=yes?

As highlighted by some voters, even if the proposal carries karst=yes, an example carries karstic=yes, a typo from an old draft version which passed through my checks. Consequently, you are welcomed to tell below, by editing this page, if, according to the voters comments, the talk page comments and the proposal content, which tagging seems the best to you, amongst karst=yes and karstic=yes.

This secondary vote only had 4 voters, with 1 abstention and 3 karst=yes votes. As it was in the proposal core, and the other one was in some erroneous examples, karst=yes has be retained. --Penegal (talk) 13:38, 12 November 2017 (UTC)

  • I would let British native speakers decide. I abstain. --Nospam2005 (talk) 19:31, 24 October 2017 (UTC)
  • I've seen the word "karst," but not "karstic." "Karst" seems right to me. --Dr Centerline
  • "karst" is the noun, "karstic" is the adjective. It may answer my question, keys are usually nouns, not adjectives. --Nospam2005 (talk) 18:21, 25 October 2017 (UTC)
  • karst=yes I'd vote for karst=yes then, because as Nospam2005 mentioned, keys are usually nouns --Bielebog (talk) 12:15, 27 October 2017 (UTC)