Tagging dead forest - Harz


I recently visited the Harz, a large forested highland area in Northern Germany.

Or so I thought.

As it turns out, the drought over the last five years or so gave the (mainly) spruce forest a hard fight, as the spruce is a shallow rooted tree. This weakened it enough to give over to a certain pest called the (spruce) bark beetle. The result looks like this. It looks as if a forest fire raged:

This is well visible on aerial imagery. Unfortunately, there is no date attached to the different layers in JOSM but it looks like the Esri one is the most current and the Bing one is the oldest. All which is brown is dead forest like shown above:


Tagging dead forest as dead forest somehow or maybe not even natural=wood anymore but something else has obvious applications like

  • being able to visualize the extent of the catastrophe
  • to know which hiking routes to avoid (dead forest doesn’t give shade)


So, how to tag that?

I could not find any discussion on the wiki nor any tag that seemed to be in use on taginfo. Maybe someone knows if this has been discussed elsewhere already?

A few things to consider after thinking about it a bit:

  1. Dead forest is dead forest. The cause that led to its demise (be it fire, pest, drought or a combination thereof) is secondary. After all, we map what is, not (primarily) why it is.

  2. It may be obvious to treat this with lifecycle tagging (destroyed:natural=wood) but this would only express that there used to be a forest and that it has been destroyed, not what is there now. The dead forest, however, is still standing. There is nothing else in its place, as would maybe be the case if the area was logged (grassland / scrubland)


OsmAnd actually renders an existing tag for this. It only seems to be documented in Russian though. Since I last looked it up there seems to have been a duplication in the tagging and FWIW I’m not sure which is the one that renders:




There are also tags (and OsmAnd rendering) of trees that are knocked over by wind or logged. They don’t render burnt trees but there is an option for that too.

In most instances I think the woods are likely to recover long term so with OSM’s cadence I don’t think it’s worth de-tagging the normal wood tag?


It is not only the Harz where this is happening … the spruce population in nearly every forest in Germany (and other European countries) is affected by the explosive reproduction of the bark beetle. In most areas the dead trees (mainly spruce, also pines here and there) will be removed as soon as possible to slow down further spreading of the pest resulting in clearcuts of a size never seen before. These areas can be tagged temporarily as man_made=clearcut without removing the main tag landuse=forest (as long as you trust the area will be reforested in due time).

To my knowledge it is only in the few National Parks where these areas covered with dead trees will not be cleared and the dead forest is allowd to be a dead forest. There have been some heavy public debates in regard of this policy (easy to be found in the web). Nevertheless I think there should be an explicite tag for such areas as this is not a short term phenomenon.

The tags proposed by @InsertUser seem fairly useful to me although I do not understand the duplication. In use approx:

wood:lost=windfall 800
wood:lost=burnt 200
wood:lost=dead_wood 60

wood:damage=windfall 300
wood:damage=burnt 200
wood:damage=dead_wood 150
wood:damage=deadwood 40

I would favour wood:damage because it covers a wider range than “lost” … and of course it would be helpful to have a wiki-page for that.

I would not put too much trust into that. Imho the problem of the dying forests has just startet, taking up speed and actually not much to be seen ahead to stop this process … :frowning_face:

I would dispute calling it dead. Diseased, yes. Many of the trees are dead, but the forest isn’t. There will be other bushes and trees growing. IMHO wood:damage=dead_wood is the best.


The affected areas are definitely not dead and will accomodate life even if not visible at the first look - even the sahara (which was forest covered in earlier times) is anything but dead … :wink: … but you cannot call it a forest or wood at this stage and nobody knows if the forest will really recover in due time.

I agree with you that wood:damage=dead_wood appears to be the most appropriate tag as it leaves all options for the further development open.

Der Anblick mag nicht dem entsprechen, was man sich unter “Wald” vorstellt. Trotzdem würde ich es als falsch empfinden, den Flächen die Waldeigenschaft abzusprechen. Die Ansichten sind nicht ungeteilt, Wie der Borkenkäfer hilft, den neuen Urwald aufzubauen - Wissen - SZ.de - Wald braucht eben Weile.

Schäden und Verluste erleiden in erster Linie die forstwirtschaftlichen Betriebe. Die zitierten Werte für die Unterschlüssel :lost und :damage passen beide gut, um Ursachen anzugeben, Sturm, Feuer, Käfer usw.

wood:lost|damage=dead_wood fällt aus dem Schema, gibt keine Ursache an, sondern einen Zustand. Da würde ich persönlich etwas wie forest:health=dead geeigneter finden. Beachte, bewusst forest, nicht wood. In Verlängerung der altbekannten künstlichen Unterscheidung. Man könnte dann auch excellent setzen. Aber ja, nur so als Vorschlag, sollte jemand sich die Mühe machen, ein proposal anzustrengen.

I don’t really understand the duplication either, the headings for the two sections are “Повреждённые (но живые, пусть частично) древостои” and “Если перечисленные выше повреждения привели к гибели всего древостоя, то” which the translator build into my browser says are “Damaged (but alive, albeit partially) forest stands” and “If the damage listed above led to the death of the entire stand, then”, so maybe the latter is intended to be when the dead trees have left the whole area dead rather than just the taller trees? I don’t know if there’s a native speaker around here that might be able to clarify.

No worries, mate … you did well in digging out these existing keys … :+1: … and I do believe the translation is quite accurate. I’m just of the opinion that 2 keys for the various stages of decay are not necessary. Imho the key wood:damage covers a wide range from modereate damages to a stage very close to complete destruction whereas the key wood:loss would only be applicaple to the very final stage.

I simply doubt that many of us mappers are expert enough to evaluate if a forest is really completely lost or if it will still recover given enough time and that is why I do strongly favour wood:damage instead of wood:loss, although the latter has been used more often according to the taginfo.


“Damage” und “Loss” habe nicht nur mit monetären Aspekten zu tun. Wenn der Wald schwer beschädigt ist, ist das auch eine Zustandsbeschreibung und ein vollständiger Verlust ist auch u.a. ein ökologisches Problem.

Die Attribute “burnt”, “dead_wood”, windfall sind nicht nur Ursachen, sondern ebenfalls recht genaue Zustandsbeschreibungen und passen von daher m.E. durchaus in das übliche Taggingschema.

Hier geht es eben nicht um forest/bewirtschafteten Forst, sondern um natürlichen wald/wood. Im forest werden Totholzflächen schnellstmöglich abgeräumt, um eine weitere Ausbreitung der Schädlinge zu vermeiden, wie auch in dem von Dir zitierten Artikel detailliert beschrieben ist - dann entstehen Kahlschlagflächen, für die es ein etabliertes Tag gibt.

Lediglich im unbewirtschafteten “wood” lässt man das Totholz stehen und liegen in der Hoffnung, dass sich der Wald von alleine regeneriert, was er im Normalfall auch macht, wie in dem zitierten Artikel ebenfalls beschrieben. Allerdings gibt es auch dafür keine Garantie. Wenn man mal mitten in der Sahara vor einem versteinerten Baumstamm von mehreren Metern Umfang gestanden hat, kann man sich vorstellen, dass klimatische Veränderungen wesentlich gravierendere Folgen haben können als nur ein paar temporär abgestorbene Bäume.

I am a bit confused the the wood:damaged key doesn’t follow the normal life cycle ordering of cycle stage:object key, i.e. disuse:restaurant. Its seems like it should be damaged:woods instead.

It’s not a lifecycle tag, but an attribute of the wood. Note how it says damage, not damaged (i.e. it’s not a state, but a property) and it doesn’t replace the normal natural=wood tag like a lifecycle tag would.

Not sure, a single tag can apply to both the use cases mentioned:

Visualizing the extent of the catastrophe inherently requires specifying the cause of the view. Governments certainly already have quite accurate data on this. Will the openstreetmap community get on par with paid professionals educated in the subject matter - Should we try?

The amount of shade to be expected seems a much more easy task, perhaps a canopy:coverage=* tag exists. Something like that might make sense with healthy, but not dense woods too.

PS: Posting in english, posting in German seems to invite German style discussion.

As far as I understand we map what we see OTG. There is no need to specify if a tree died because of heat, draught oder bar beetles. What we can see OTG is dead_wood, burnt_wood, wood broken down by heavy winds and the listed tags describe these states of damage quite well. Also we can easily recognise a clearcut area and place the tag if we believe it makes sense.

I am aware some people are of the opinion these are all normal states in the lifecycle of a forest or wood and no need for a special tagging (there was some discussion about clearcut areas recently in the german forum including such positions) but well, no one is forced to make use of these specific tags.

For those who consider this to be an important information the available tags would serve well in my opionion.

?? … :face_with_monocle: … facts ar facts … no be so?

Thank you for the hint!

It looks like wood:lost=dead_wood is meant to be used as a subtag to natural=wood or landuse=forest. Which means, it is not similar to lifecycle tagging but kind of orthogonal to it.

There is an issue with this sort of tagging, and that is the question whether a dead forest still constitutes as a forest. To illustrate:

Why lifecycle-prefix tagging is superior

The ruins of a church can be tagged in different ways:

  1. With lifecycle prefix: ruins:building=church
  2. With subtag: building=church + ruins=yes
  3. As tag value: building=ruins (+ ruins=church)

Option 1 is better because option 2 and 3 require additional software support to recognize an object tagged as such as not actually still a church building. A ruin of a church is simply not a church (anymore).

Implication on dead forest

Similarily, natural=wood + wood:lost=dead_wood, as a subtag of natural=wood requires additional software support to recognize that there is no more forest, as if it is a property of the forest that it is gone.

From the viewpoint of someone having went hiking there, it kind of strikes me as wrong to treat it as a property of forest, because everything that makes it a forest, is gone. Sure, life finds a way, and it’s not as if the area will be devoid of life from now on, but for all intents and purposes, it’s not a forest anymore, instead, something else. Maybe in 5 years, it will be natural=scrub or natural=heath, but not now. Now, it is more like man_made=clearcut only that it hasn’t been clear-cut.

On the other hand there are some forests that are reliant on e.g. fire for the next generation of seeds to germinate so rather dramatic damage may be far more temporary than it initially appears. If this is an entirely expected part of a forests ongoing lifecycle would we really want to de-tag it for transient damage that may be long gone before the next mapper arrives?

1 Like

would we really want to de-tag it for transient damage that may be long gone before the next mapper arrives?

Yes! There should be at least the possibility to tag it for those who want to depict the current situation.

Even eucalyptus takes ~10 years to be called something of a (new) forest again after a wildfire raged that left nothing behind. It is the same thing as with man_made=clearcut really - eventually, new trees will probably be planted, but trees don’t grow from one day to the next.

A time period of even 5-10 years does not strike me as so transient that it is not worth even having a proper tagging for it.

I guess I am starting to see, what is meant. Instead of lifecycle, why not just natural=deadwood, plain and simple?

You can see it this way but you could also understand the period of damage by wind, fire or pest as part of the normal lifecycle of a forest. Your example of the church ruins does not really fit because once the church is in ruins it will never recover again (unless by chance someone may come, remove the debris and build a new one). It is not a living organism like a forest/wood.

That is why I would not favour wood:loss … hard to estimate if it is really lost. If in a couple of years it may be ovbious there will be no forest any more, all forest related tag can still be removed but then you also would not need any wood:loss tags any more. Until then it could still be seen as a forest in a poor state and I would not see a reason why to remove the forest or wood tag. The same applies to man_made clearcut which also can be tagged additionally to the existing landuse=forest.

wood:damage leaves the end open and recognizes that this is a forest area even if badly damaged at the time being.

Not at all. I have been living in Australia in 2003 when bad fires destroyed lots of forested areas at the east coast. I visited some of these areas just a couple of weeks later and saw this:

All over the black skeletons of the entirely burnt trees fresh sprouts had popped out … just weeks after the fires had raged in these areas. That is why I believe not many of us would be able to determine how serious a bad looking forest damage really is and even less if it would be a complete loss.

1 Like

Just out of curiosity, I mapped two clear-cuts that I happened to pass by this weekend. Of course, OSM-Carto does not show the extent of the catastrophe. Reading up a bit on the subject, OSM-Carto has indeed valid reasons, to not show clear-cuts. I might use osmand, to get to see it. Maybe, I still had needed to have punched holes into the mulitpolygons, the OSM entity that woods are mapped here?

The reasoning goes: We do not show actions, we show state, and clear-cut is not a state. I do not think, natural=deadwood might fare better, although lots of area in the Harz may look like the pictures above for several years, never mind it is teeming with life of various kind. Still, even if adopted, why should natural=deadwood, even when it does not conflict, merely adds detail, trump landuse=forest|fuelwood? I’d still have to swiss cheese those multipolygons. And as long as there is no rendering of the mapping intent, I’d just get blank empty space, same as with lifecycle prefix, not a visualization of the extent of the catastrophe.

Ich möchte noch einen weiteren Aspekt für Totholz hinzufügen, der hier noch nicht besprochen wurde: Ursache ist kein Waldbrand, kein Sturm, kein Borkenkäfer, keine Dürre. Die Ursache ist vielmehr zu viel Wasser.

Auf Flächen knapp unterhalb des Meeresspiegels sind teilweise Wälder entstanden. 1990 wurde ein Nationalpark eingerichtet, einige Jahre später wurde die Entwässerung der Flächen beendet und der Grundwasserspiegel ist gestiegen. Die Bäume vertragen so viel Wasser nicht. Inzwischen sind große Flächen abgestorben. Die Bäume stehen noch, sind aber kahl und tot. Der Mensch greift nicht ein, der Wald wird wahrscheinlich nicht neu entstehen.