Properly mapping dry washes

A dry_lake would also fit into the scheme that uses dry_river and dry_stream.


There is a dry lake proposal that probably just needs to be finalized and voted on.

But what’s the distinction between natural=dry_river and natural=dry_stream?

I was thinking dry_stream would be for a wash that is waterway=stream sized when water is flowing and dry_river would be for a larger one. Basically an easy transition for small and large washes that are currently tagged as waterway=stream|river + intermittent=yes.

1 Like

I’m not super familiar with desert environments, but do washes ever have what would be considered “normal” flow for a non-intermittent stream? Or do they alternate between dry and flood-stage? If the latter, a waterway=dry_stream wouldn’t necessarily be jump-able when water is flowing.

This would be my interpretation. Though I’d probably classify a wash less than maybe 5 meters wide as dry_stream which is a bit wider than I think a stream would be tagged if perennial.

The ones where I was raised basically had two states: Dry and full. Though “full” did vary a bit but usually had water across from bank to bank the variation was how high up the banks the water was when full.

@cactolith said it well in the thread on mapping virtual waterways.

There may be seasonal (or less frequent) events where the flow of water is small, in which case it is likely constrained to the active channel. In the case of a significant flood, it’s basically sheet flow across the entire area of the wash.

I guess there’s some variation based on the local topography and weather. As @n76 says, in some areas when the wash floods, it’s entirely full. But otherwise it’s dry.

(Edit) I should add that I’ve discarded the measure of “jump-ability” to determine whether something is a waterway=stream or waterway=river. For me, the distinction is whether it’s wide enough to be mapped as an area in OSM.


When is it a wash and when is it an intermittent stream? When is it an area and when is it a line? Since the distinction is very vague, I’d suggest the same tag for both. The NHD washes are often just a vaguely wide spot in a gulch.

It’s possible that Colorado is over represented due to the nature of the NHD data or the import.
Currently there are a huge number of dry gulches tagged as waterway=stream and intermittent=yes. It seems like reasonable tagging for the most part, but sometimes I question whether these features are relevant enuf to even be in the data base. Most are not seasonal, but would have flow once or twice a year or once or twice a decade. Most are narrow gullies. Most have no name, but if they have a name it’s Something Gulch. Ephemeral seems like a good tag for many of these dry gulches.

In AZ it’s flatter, the intermittent water flow areas tend to be wider and are usually called washes. They also do not seem to be as thoroughly tagged.

For a taste of how thorough the database is near me (for dry gulches):

We make a clear distinction between the course of a river mapped as a linear way tagged with waterway=river and the broad water area of a river, mapped as an area tagged with water=river. The same distinction applies to the course of water through a wash and the broad area between the well defined banks of the wash. I think it makes sense to distinguish between the linear water course and the broader area of the feature.

I don’t personally have much interest in the detailed tagging that goes on in urban areas, but I’m glad that some people do. OSM is richer for it. There’s room for complex tagging schemes for crosswalks and benches, so there should be room to tag natural features as they are. I think OSM data for natural features should be just as rich as the data for urban features. Getting by with tags that don’t quite work is like fitting square pegs into round holes.


When hiking in the desert, a wash can be a pretty good landmark for navigation.

For urban vs rural tagging, I seldom bother with power lines in a suburban or urban area. There are just too many and provide little information for my suburban/urban map use (mostly car navigation). But in the desert they are a feature that can be visible for miles and good for keeping oriented while hiking so I am likely to map power lines there. I guess it comes down to what you plan on using the map for and how you personally use a map.

That said, almost any slight terrain feature could carry water during a downpour and we probably shouldn’t clutter the map too much with every wrinkle in the landscape being tagged as an ephemeral stream. For me, a wash has to be long enough, maybe 1/2 mile, and/or wide enough, maybe 3 meters, to count.

I look forward to a consensus on how to tag these desert washes so that I can improve my mapping and update the scripts that generate my hiking maps from OSM data.


I’m 100% with this. We need to be able to distinguish a significant wash from a very small gulch, one of many.

This is pretty much what we have some places, every wrinkle. We probably should keep the data that’s there, but the maps don’t need to show all of them. Perhaps there does need to be a distinction between intermittent and ephemeral.

1 Like

Sadly, the proposal for ephemeral=* failed and everything has been pushed into intermittent=* and seasonal=*. Those are very broad brushes with which to paint the range from wet to dry.

I recall being disappointed that the ephemeral tag failed to get support.

But I also think a lot more states could be defined for when water is present: Perennial, seasonal (or perhaps list which season, i.e. “summer”), vernal (at least for pools/ponds), intermittent and ephemeral. Rather than adding a bunch of *=yes tags it seems to me that something like water_present=perennial | seasonal | vernal | intermittent | ephemeral would be better than what we have. Though vernal is probably just a subset of seasonal.

That would be for tagging the main waterway centerline. I am beginning to like the water = dry_stream | dry_river suggestion for the area tagging between the banks.


I’m not quite sure that a single tag wouldn’t be better than natural=dry_stream/dry_river. (Let’s not do water=no_water.) But I could be convinced.

There are several examples of questionably tagged washes at the top of the thread. How would you tag these areas?

1 Like

I agree. I’d also prefer to avoid the pattern water=dry_* since the vast majority of the time there is no water. For the linear ways, using the waterway= key also feels a little weird, but more ok because it is already used for various water adjacent features (like dams and docks) that aren’t actually water themselves.


Since there’s resistance to calling these ephemeral streams, I’m coming back around to what @watmildon did, call these washes. natural=wash or waterway=wash. I don’t see why we need a different tag for a linear feature and an area (with a really fuzzy, hard to map boundary). Perhaps someone can convince me otherwise.

It would be nice to use the same tag for dry watercourse’s whether they’re desert washes or mountain gulches (which I usually think of as having a steeper drop and steeper sides). Hmm, how about dry_watercourse? I don’t favor dry_stream or dry_river, it just doesn’t fit with many of these features.
I don’t have a problem calling them waterway’s though, they are features where water flows and were created by water.

The pics that @Kai_Johnson posted above look distinct and mappable. However, we need to consider that many of these imported features do not match current reality or are very inconsistent.

This one: Way: 47356327 | OpenStreetMap , and the one just SW of it are tagged areas in a wash which is otherwise not in the database. To my eye those areas not not very distinct from the rest of the wash. Perhaps it was when it was put into NHD. The whole wash could be stream, intermittent, or whatever we come up with for a wash

Way: 46227123 | OpenStreetMap: It looks like the dunes have taken over (it’s in Great Sand Dunes Nat Park). The area doesn’t match well with any imagery, there may be some wet spots, which are different in ESRI and Bing. Is there any reason it should be kept?

Way: 47425436 | OpenStreetMap: doesn’t match imagery. Yes, there’s a wash there, but the wash doesn’t line up well with the polygon. The area doesn’t look any different than what’s upstream or downstream.

Whatever we end up wanting to try for tagging, I’m sure there will be an infinite sea (dry_sea?) of cleanup!

Maybe @watmildon has some comments about this after working on cleaning up circular waterways.

And I totally sympathize with the inaccuracies of NHD data! It’s apparently a very good data set if you are looking for the topology and connectedness of watersheds. But if you import it into OSM, it’s easy to see that it doesn’t line up well with aerial imagery or USGS 3DEP elevation data. There are lots of places where the water goes improbably uphill or where the waterways have meandered out of alignment with the data. I suppose that doesn’t matter so much if what you really want to know is how a change to one waterway will affect others downstream. But it sucks when you import it into OSM and it doesn’t fit with what’s on the ground.

I think there’s value in having a tagging scheme for various area features (landcover, surfaces, etc) and another potentially complimentary tagging scheme for the things that run through those features. It’s 2 abstractions on top of one another. In the same way we have area:highway (to denote the dimensionality of a roadway) and highway= (to denote the flow patterns).

I personally like the natural= key for the areas we’re discussing because it’s a very common primary key for area features already AND I think co-opting a water= subtag key looks kinda goofy when we’re trying to work out how best to represent not water. That said, so long as we have SOMETHING I’ll be so so happy.


The challenge with tags that can apply to linear or area features is that every data consumer has to figure out which is which on an object by object basis. leisure=track is the classic example of that - in cases without an explicit area tag, you genuinely cannot tell without looking at the aerial imagery of each!

Obviously washes aren’t quite that challenging - a closed way is “very likely” to be an area, not a ephemeral waterway going around in a circle. Still, it absolutely helps from a QA point of view to have different tags for linear and area features.

The key is “something not already used for something else” - if there’s nothing suitable suggested by taginfo already**, just make something up, and tag a bit of desert with it?

** not just in the USA, but in places like Saudi, north Africa, etc…


Regarding lines versus areas, an interesting use case I came across was when the geography and national borders between Croatia and Serbia was discussed. The river (linear flowing feature) is used as the international boundary. The river doesn’t have a passport and doesn’t really care so it meanders throughout the braidplain/floodplain (area) and creates small enclaves of deposits that are solid material above waterline. A few enterprising individuals took the opportunity to declare these as independent states resulting in tanks (Croatian) moving in on one to settle the dispute.

Much of the approach relies on the use case. I will echo the observations that in a landscape with few features, anything you can reference is critical, sometimes life-saving. I have personal experience with this. The deserts I have lived, worked and mapped in knowing the existence and extent of a wash can make a difference in where you plan to hike, drive, camp. The location also provides secondary information to users such as location of gravels, mud, sand, vegetation, fossils, sinkholes, caves and other access points or barriers to navigation. Most importantly, whether a trickle or dry, these features can suddenly turn into a true force of nature as a flash flood under a clear blue sky.

I generally use a loose adaptation of the Strahler number for deciding which streams to map and which ones to leave unmapped. If its a deep cut and goes for a mappable distance I map it. If it has a few feeders/branches I map it. Satellite resolution improvements also allow a better map, especially to those who can put the features into a broader context such as catchment areas, braidplains, floodplains, washes, arroyos, alluvial fans, fluvial deltas etc. A good example of this would be here in Oman near Salalah. Streams coming out of the mountains suddenly have little to no confinement and even the areas interweave! Dry and static for most of the time but critically dynamic when weather systems put enough water in to create these features.

Wandering back to the specific topic here…mapping these systems are critically important for the use case of people venturing into areas where they exist. The current tools available to tag are limited but if done consistently can be very useful. waterway=stream/river/gully; intermittent=yes provides the necessary information for someone in the area under low/confined flow status. Coupled with area: water=river_area; intermittent=yes helps for flood stage/high flow times, a de facto hazard map for flash floods and navigating by vehicle/4WD etc.