Properly mapping dry washes

Yes and no :slightly_smiling_face:. Currently needs the waterway tag. But I can change that. I did that for canoe portages. We should map correctly, and data consumers like me with WWM must adapt. The practice of connecting waterways together has existed for years. There are already many things using the waterway key which aren’t ways that water flows (e.g. waterway=dam etc.).

1 Like

Yes please no trolltags! :slightly_smiling_face:


I feel like we’re already putting trolltags on dry lakes tagged with natural=water but that’s another topic.

1 Like

To hopefully bring the thread back on track, we have two separate issues here:

  1. How to map the linear course of water through a normally dry wash. The current practice is to use a linear way tagged with waterway=stream + intermittent=yes. It’s not my favorite, but I’m not bothered by it.

  2. How to map the broad area of a dry wash where it widens. That’s what the examples at the top of the thread refer to and what the original question was. There doesn’t seem to be an established practice for mapping these areas. I would oppose suggestions to use closed ways tagged with waterway=* or water=* to map these areas.

I understand these two topics are related, so I won’t ask to split the thread. But I will ask again:

Can we agree on tags that work for the wide areas of dry washes?

As mentioned earlier ^, what’s wrong with
natural=water + water=river + intermittent=yes?

1 Like

I covered that very specifically in the first post starting the topic.

If a feature is rarely wet, tagging it as “water” is incorrect and misleading. That just makes it seem like there are all these places that are wet when the reality is that they’re dry.

And, in practice, it would make our deserts look really blue. Even without this discussion, ordinary mappers have rejected that tagging because it simply looks wrong. That’s why we have so many alternate (and also incorrect) tags for these areas.


Revisiting the early versions of the thread…I tag waterway=stream|river; intermittent=yes for the linear pathways of the water AND add a 4WD track if visible or just rough it through the vague areas to show there is a waterway along with a routable path through it. Some of these are impassable across boulder fields and I try to tag that way, also there are some cliff faces which water can flow over but vehicles can’t safely.
Sidenote, no proposal here but we refer to the broad areas where the lakes form and quickly dry out as Etch-A-Sketch since any track gets obliterated from view when it fills up.

Don’t hate the player, hate the game…If there’s a vote and if mine counts using what is currently available I say:

  • Linear features - waterway=stream|river; intermittent=yes

  • Areas - natural=water; water=stream|river; intermittent=yes

If a proposal is to be made I revert to the suggestion of waterway=wadi
Alternate proposal if that was replaced with intermittent would be to subdivide/restate intermittent into something like:

  • intermittent=frequently
  • intermittent=seasonal
  • intermittent=rarely

If a flowing stream is common then mapping the active channel or channels helps further indicate the nature of the area. This would apply to an active meandering river with a recognizable floodplain including oxbow lakes and minor channels off the main channel. If there is an obvious channel cut in a dry braidplain the intermittent tag is useful for planning navigation. If the dry area is a broad non-descript area with no obvious main channel (wadi, wash, arroyo, glacial outflow) whether active or dry I would leave it as just an area.

Qatar has no flowing freshwater rivers|streams outside of major rainfall events and only limited tidal channels along the coast. Dry lake beds are common and when this all gets mapped it looks like a much wetter place than it usually is. Scrub works to make the areas appear on navigation maps and they often occur where the water would be but not always. The use case is when they are dry they are like concrete but when wet can swallow a Land Cruiser up to the rearview mirrors.

Is there a way to render an intermittent stream|river|lake as brown stippled? When I was mapping geology in the desert SW of the US the topo maps would have the dotted or dash-dotted ephemeral streams indicated. This was understood and very useful.


But does that mean that they’re frequently wet or frequently dry? :thinking:

Lakes / basins are currently rendered dashed blue areas in Carto, so that “should” be easy enough?

Off topic, but that’s totally true at ground level where navigating braided washes is definitely challenging!

On the other hand, I often find that the vehicle-width path through the scrub remains clearly visible in aerial imagery for a very long time. It’s even possible to spot it in Wilderness areas that have been established for more than 30 years. Where it’s possible, I use that to map the tracks through washes because no one wants to drive through scrub, and no one wants people plowing over the vegetation when there’s already an established track.

Since the thing being tagged is referred to as water then the intermittency tag would refer to the presence of water thus intermittent=frequently would de facto mean frequently water present…

1 Like

I like to style my hiking maps based on the older USGS topographic maps of the 1950s & 60s that I was raised with. At present, washes that have areas defined appear way to blue even though they are hashed out because they are tagged intermittent=yes. If I style them to look more like I expect for a desert wash, then the maps I make of another, wetter, area would appear too dry.

Extending intermittent=* like this suggestion would allow me to differentiate and properly style my maps for both situations. I like that. (Would that be tagging for the proposed renderer? :slight_smile: ) It would also allow the user to make a determination of how easy the area is to traverse most of the time.

I don’t have any issue with frequently, seasonal, or rarely (in addition to yes). Since the default for intermittent is either yes or missing (implied no) it is pretty obvious that frequently is a “usually in the yes state” and rarely is a “usually not in the yes state”.

1 Like

Using intermittent=rarely to mean rarely wet (thus usually dry), in combination with natural=water would be a classic case of a trolltag. natural=water means an area where water is present (or at least usually present). Adding a tag that makes this mean the opposite is a bad idea. Areas that are usually dry and very occasionally wet should use a different primary tag than natural=water.


I don’t know enough about this subject. However, if I Wiki Key:ephemeral think that should describe exactly that.


1 Like

I absolutely agree with @ezekielf that starting with natural=water as the primary tag is the problem and that adding secondary tags like intermittent=* or ephemeral=* to indicate that there is no water is trolltagging. That’s the main problem with the existing tagging and exactly what we’re trying to get away from.

I like the idea of using natural=dry_stream/dry_river or natural=wash as primary tags to show that these areas are principally dry. My only hold up over dry_stream/dry_river is that I’m not sure how I’d make the distinction between those two tags in practice. The width of the wash can’t be the deciding factor, and the volume of periodic flooding is variable and hard to verify.

Maybe we could agree on a single primary tag and use secondary tags to indicate the volume of water during floods, if that can be verified?


Defining a feature by the absence of the other feature isn’t how we make keys that are understandable. Any result here that leaves us marking areas that almost never have water in them as natural=water seems absolutely preposterous. It it exactly water=no_water! It’s basically where we’re at today and has caused this entire thread to exist.

There’s basically 2 questions that need to be answered and neither of them lead to natural=water. The first is “do dry washes across the globe have similar enough properties to have one key for them?” AND “what should that key/keys be and how to differentiate them?”.

  • natural=wash/dry_wash/wadi/arroyo - these all seem roughly interchangeable in usage. I am happy with any of them in practice but would love to know what, if any, subtleties are being washed (har har) over.
  • natural=dry_river/dry_stream - I get this has nice coupling in the mapped waterway= features but share your concern about how to just one vs the other.

If only such features were common in England, we’d already have our answer.


Guidance in the OSMWiki for geologic terms in the Glossary of landforms

Looks like arroyo and wadi made it in but not wash. They’re trying to clean up wadi and arroyo is waterway=stream; intermittent=yes

The linear feature waterway=stream|river; intermittent=yes works in my opinion. It’s a waterWAY, not water. Much as a carriageWAY doesn’t cease to exist and become a trolltag by the absence of carriages. Ephemeral is the term used in the literature but that does not appear to be a precedent followed within OSM so we’re left with intermittent=yes. I’ll close on that as it doesn’t seem controversial here. It does offer insight to the next resolution we’re all trying to achieve…

The issue we’re discussing here is how to tag the area which has the top level tag of natural=water for a wadi when the presence of water is both rare and the defining process for it’s existence.

Question 1 is an emphatic answer=yes! Dry washes as used by the OP are pervasive and arguably a subset of braided rivers.

Question 2 is where we should focus.
Currently, I use River Area; intermittent=yes to map wadis in Qatar, Oman, Georgia(country) and other places I work and travel.

This is the definition for a wadi and how the process is the same but the name is different for reference. Wadis are almost always dry, sometimes for decades at a time.
So, why care and even why map them? As the photos posted by the OP @Kai_Johnson these features are distinct and the areas are very easily mappable in most cases. They are also differentiated from the host area in that they will usually bear the marks of intermittent water flow, debris flow, changes in vegetation and material assemblage (sand, boulders, rounded cobbles, washing machines, small cars, beer cans etc). I saw a Volkswagen go by during a rainstorm in New Mexico in an otherwise dry arroyo.

A wash is a wadi is an arroyo. Independent of local name (learned yesterday they use vadi in Turkey) they are all subsets of what I would call a braided river. Braided river is the top level term which describes everywhere the feature occurs be that a glacial outflow, fluvial or dry wash. A braided river|stream can have an active (water flow) channel, multiple channels, one giant channel (flood stage) or completely dry. In Oman near Salalah, Wadi Darbat is infuriating since it can have flowing water continuous along the main channel, be at flood stage and cover the entire area and during dry periods have pools that are spring fed and go a distance and disappear into karst as a lost river/losing stream…smh…

You do! You just label a component of them natural=shingle. Again, a local subvariant of a braided river|stream

Depending on gradient and water flow rate you can transition between braided (high flow and|or steep gradient) and meandering (low flow and|or shallow gradient). An occurrence I located in the UK is covered in vegetation and has natural=shingle tag.

I suggest these areas are all Area=River Area although the documentation is weighted towards it being always or more often water covered. Expanding the definition/guidance to encompass the recognizable bank full main channel area meets the need. During a true flood stage the water has escaped the confinement and occupies the floodplain although the main channel is still flowing faster and transporting material. This can be discussed in the other active discussion mentioned earlier about mapping waterways through standing water areas. As a thought process let me mention all the subvariants covered by the (proposed) modified definition of Area=River Area term and see if they fit your need:

  • Meandering streams and rivers
    Components under this include active channel, secondary channels (waterways), meander scars, point bars, levees, banks, shingles and all the associated natural things affected by the flow of water in a low gradient naturally flowing fluvial system.
  • Braided streams and rivers
    Main channel, meandering low flow channel, secondary channels all either continuously water present, intermittent or dry. Shingles, braidplain, gravel, sand, boulders.

This is reminiscent of construction area where construction is occurring but not everywhere all the time and recreational area or parking area and their respective activities. River area is where river stuff happens and that can be when there’s a little or a lot of water moving through it. The river area is defined by the effects of the river processes and is distinct from adjacent areas. So, a dry wash is a river area where river things happen during flow and the effects of the river stuff are obvious when it is dry.

1 Like

A couple of links that might help:

Values used with the waterway key in the USA
An overpass query for closed ways with a particular waterway key (here, wadi)

Obviously you can move the overpass query around and search for slightly different things.

Edit: Actually, you can search more widely for water tagging if you look where there is little water like here in the USA or here in North Africa. Quite a few

intermittent = yes
natural = water
water = wadi



Yes, using natural=water as top level tag for wadi/wash/arroyo is the status quo (with the addition of the intermittent=yes attribute). That doesn’t mean the status quo is optimal or correct. It’s just a tag combination that is somewhat close to describing the features and renders reasonably. A river is a stream of flowing water. A wash/wadi/arroyo is not a stream of flowing water, it is the landform created by an occasional stream of flowing water. There is a reason people came up with different words for these features rather than just calling them rivers that happen to be dry most of the time.


Taginfo natural=wadi 22x

1 Like