Forest parcel with other landcover (scrub, scree…): how to map?

classic Classic list List threaded Threaded
77 messages Options
1234
Reply | Threaded
Open this post in threaded view
|

Forest parcel with other landcover (scrub, scree…): how to map?

David Marchal
Hello, there.

All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?

Awaiting your answers,

Regards.
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

marc marc
Le 20.01.19 à 19:32, David Marchal a écrit :
> The landcover tag?

you may of course, despite it's not used by osm-carto
(but we don't map for the render, isn't it ?)

> Another?

map with natural=wood for the area with treet
map the scrub area as usual

> boundary=forest_compartment?
i dislike it.

Regards,
Marc
_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Kevin Kenny-3
In reply to this post by David Marchal
On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?

This again.

There's a failed consensus here - and you risk reversion with either decision.

I tend to follow the principle that landuse=* denotes the land USE,
not the land COVER, so I don't demand that every square metre of
landuse=forest be covered by trees. But many do, and the renderer
follows their inclination.

natural=wood is a possibility to show tree cover - but that leads some
to argue that it has to be a 'natural' wood - whatever that means.
I've heard it argued that the 'old second growth' forest that's
increasingly common near me is still not 'natural' because a skilled
forester can still find the human impact. (Of course, that was true
even before the Europeans arrived - there was considerable
pre-Columbian human impact on these forests.)

landcover=trees doesn't render, but is at least unambiguous that it
means tree cover and nothing else.

landuse=forestry, for a managed forest, has been proposed but received
a lukewarm reception.

For the state forests and wildlife management areas around here, I tag
at least boundary=protected_area. (Tag with the right protect_class,
and add leisure=nature_reserve if it fits: 'nature reserve' covers a
lot of things.) If I'm mapping land cover (I seldom do), I will use
natural=wood to mean 'tree cover' and let others fight over it.

But that's just what I do, and I do not argue that it is right. With
the current state of the discussion, which has been in stalemate for a
few years, there simply is no correct tagging, and what I do appears
'least wrong' to me.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Joseph Eisenberg
> “an area considered as a forest by authorities”

If this is a protected or administrative “forest”, you can use boundary=protected_area with the proper class

But we usually try to map what is “real” and “current”. So if there is an area without formal protection, that people call “XXX Forest”, but it contains meadows, trees, rock and scrub, then map each area separately.

You might need to make a multipolygon if there are large areas of woodland with patches of other landcover inside. This lets data users know that the trees end there.
On Mon, Jan 21, 2019 at 3:54 AM Kevin Kenny <[hidden email]> wrote:
On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?

This again.

There's a failed consensus here - and you risk reversion with either decision.

I tend to follow the principle that landuse=* denotes the land USE,
not the land COVER, so I don't demand that every square metre of
landuse=forest be covered by trees. But many do, and the renderer
follows their inclination.

natural=wood is a possibility to show tree cover - but that leads some
to argue that it has to be a 'natural' wood - whatever that means.
I've heard it argued that the 'old second growth' forest that's
increasingly common near me is still not 'natural' because a skilled
forester can still find the human impact. (Of course, that was true
even before the Europeans arrived - there was considerable
pre-Columbian human impact on these forests.)

landcover=trees doesn't render, but is at least unambiguous that it
means tree cover and nothing else.

landuse=forestry, for a managed forest, has been proposed but received
a lukewarm reception.

For the state forests and wildlife management areas around here, I tag
at least boundary=protected_area. (Tag with the right protect_class,
and add leisure=nature_reserve if it fits: 'nature reserve' covers a
lot of things.) If I'm mapping land cover (I seldom do), I will use
natural=wood to mean 'tree cover' and let others fight over it.

But that's just what I do, and I do not argue that it is right. With
the current state of the discussion, which has been in stalemate for a
few years, there simply is no correct tagging, and what I do appears
'least wrong' to me.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Warin
In reply to this post by Kevin Kenny-3
On 21/01/19 05:52, Kevin Kenny wrote:

> On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
>> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?
> This again.

And it will continue to occur!

And reoccur, again and again.

>
> There's a failed consensus here - and you risk reversion with either decision.
>
> I tend to follow the principle that landuse=* denotes the land USE,
> not the land COVER, so I don't demand that every square metre of
> landuse=forest be covered by trees.

+1

>   But many do, and the renderer
> follows their inclination.
>
> natural=wood is a possibility to show tree cover - but that leads some
> to argue that it has to be a 'natural' wood - whatever that means.
> I've heard it argued that the 'old second growth' forest that's
> increasingly common near me is still not 'natural' because a skilled
> forester can still find the human impact. (Of course, that was true
> even before the Europeans arrived - there was considerable
> pre-Columbian human impact on these forests.)

Those who argue this have no problem abusing the landuse tag, so I see no reason why the tag 'natural' cannot be similarly abused.
The OSMwiki for 'natural' even states that is can be used for human effected things.

>
> landcover=trees doesn't render, but is at least unambiguous that it
> means tree cover and nothing else.
>
> landuse=forestry, for a managed forest, has been proposed but received
> a lukewarm reception.

For forestry area I tag landuse=forest with produce=trees (or what ever is produced by the area for human use). This makes it clare that the area is for productive human use.

>
> For the state forests and wildlife management areas around here, I tag
> at least boundary=protected_area. (Tag with the right protect_class,
> and add leisure=nature_reserve if it fits: 'nature reserve' covers a
> lot of things.) If I'm mapping land cover (I seldom do), I will use
> natural=wood to mean 'tree cover' and let others fight over it.

I too use natural=wood with landcover=trees to map a tree area.

--------------------------
The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.
The essential difference between the two is that landuse must have some human benefit, a produce, and a clear way of doing that is to add the rendering of a axe to the tree.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Joseph Eisenberg
> The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.

Until several years ago the “standard” style (Openstreetmap-Carto) did show a difference between landuse=forest and natural=wood. However, mappers used these two tags interchangeably even then. The rendering was changed to match actually database usage on a global scale, which is that both tags are often used to tag any area covered with trees.

The current rendering follows tag usage and the current wiki page, which also discusses this issue in depth.

I wish it were possible to fix this, but the different meanings of “forest” and “wood” in various English dialects make it difficult, even before we add other languages and cultures to the mix.
On Mon, Jan 21, 2019 at 8:04 AM Warin <[hidden email]> wrote:
On 21/01/19 05:52, Kevin Kenny wrote:

> On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
>> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?
> This again.

And it will continue to occur!

And reoccur, again and again.

>
> There's a failed consensus here - and you risk reversion with either decision.
>
> I tend to follow the principle that landuse=* denotes the land USE,
> not the land COVER, so I don't demand that every square metre of
> landuse=forest be covered by trees.

+1

>   But many do, and the renderer
> follows their inclination.
>
> natural=wood is a possibility to show tree cover - but that leads some
> to argue that it has to be a 'natural' wood - whatever that means.
> I've heard it argued that the 'old second growth' forest that's
> increasingly common near me is still not 'natural' because a skilled
> forester can still find the human impact. (Of course, that was true
> even before the Europeans arrived - there was considerable
> pre-Columbian human impact on these forests.)

Those who argue this have no problem abusing the landuse tag, so I see no reason why the tag 'natural' cannot be similarly abused.
The OSMwiki for 'natural' even states that is can be used for human effected things.

>
> landcover=trees doesn't render, but is at least unambiguous that it
> means tree cover and nothing else.
>
> landuse=forestry, for a managed forest, has been proposed but received
> a lukewarm reception.

For forestry area I tag landuse=forest with produce=trees (or what ever is produced by the area for human use). This makes it clare that the area is for productive human use.

>
> For the state forests and wildlife management areas around here, I tag
> at least boundary=protected_area. (Tag with the right protect_class,
> and add leisure=nature_reserve if it fits: 'nature reserve' covers a
> lot of things.) If I'm mapping land cover (I seldom do), I will use
> natural=wood to mean 'tree cover' and let others fight over it.

I too use natural=wood with landcover=trees to map a tree area.

--------------------------
The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.
The essential difference between the two is that landuse must have some human benefit, a produce, and a clear way of doing that is to add the rendering of a axe to the tree.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Andy Townsend
On 20/01/2019 23:17, Joseph Eisenberg wrote:
> > The end to this madness is for renders to recognise that the
> landuse=forest needs to be rendered differently from natural=wood.
>
> Until several years ago the “standard” style (Openstreetmap-Carto) did
> show a difference between landuse=forest and natural=wood. However,
> mappers used these two tags interchangeably even then. The rendering
> was changed to match actually database usage on a global scale, which
> is that both tags are often used to tag any area covered with trees.
>
One suggestion that I've made here before is explicitly to use
"landuse=forestry" for areas that may or may not have trees on them, if
the areas with trees within have been mapped separately

An example is
https://www.openstreetmap.org/way/620189661#map=15/54.0158/-0.9549 , and
the result, in a renderer that deals with it, looks like
https://map.atownsend.org.uk/maps/map/map.html#zoom=16&lat=54.01539&lon=-0.95744 
.

That renderer also processes landuse=forest the same way - see
https://www.openstreetmap.org/way/44018882 and
https://map.atownsend.org.uk/maps/map/map.html#zoom=15&lat=53.21319&lon=-1.18217 
for an example of that.

So from both a tagging and a rendering perspective it's a solvable
problem, but the way I've done it does go against a couple of the "ways
of working" that OSM's standard style has (e.g. transparency).

Best Regards,

Andy



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Warin
In reply to this post by Joseph Eisenberg
On 21/01/19 10:17, Joseph Eisenberg wrote:
> The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.

Until several years ago the “standard” style (Openstreetmap-Carto) did show a difference between landuse=forest and natural=wood. However, mappers used these two tags interchangeably even then. The rendering was changed to match actually database usage on a global scale, which is that both tags are often used to tag any area covered with trees.

The current rendering follows tag usage and the current wiki page, which also discusses this issue in depth.

I wish it were possible to fix this, but the different meanings of “forest” and “wood” in various English dialects make it difficult,

The meaning of the key 'landuse' is fairly clear in any English dialect.

The problem of the key 'natural' remains.

Once you combine the OSM keys and values of landuse=forest and compare it to natural=wood I think most will agree there is a difference,
The former is for what the land is used for.
The latter is for the presence of plants, if you take any plant as being natural then natural=wood is 'acceptable' for ant tree area.' 

even before we add other languages and cultures to the mix.
On Mon, Jan 21, 2019 at 8:04 AM Warin <[hidden email]> wrote:
On 21/01/19 05:52, Kevin Kenny wrote:

> On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
>> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?
> This again.

And it will continue to occur!

And reoccur, again and again.

>
> There's a failed consensus here - and you risk reversion with either decision.
>
> I tend to follow the principle that landuse=* denotes the land USE,
> not the land COVER, so I don't demand that every square metre of
> landuse=forest be covered by trees.

+1

>   But many do, and the renderer
> follows their inclination.
>
> natural=wood is a possibility to show tree cover - but that leads some
> to argue that it has to be a 'natural' wood - whatever that means.
> I've heard it argued that the 'old second growth' forest that's
> increasingly common near me is still not 'natural' because a skilled
> forester can still find the human impact. (Of course, that was true
> even before the Europeans arrived - there was considerable
> pre-Columbian human impact on these forests.)

Those who argue this have no problem abusing the landuse tag, so I see no reason why the tag 'natural' cannot be similarly abused.
The OSMwiki for 'natural' even states that is can be used for human effected things.

>
> landcover=trees doesn't render, but is at least unambiguous that it
> means tree cover and nothing else.
>
> landuse=forestry, for a managed forest, has been proposed but received
> a lukewarm reception.

For forestry area I tag landuse=forest with produce=trees (or what ever is produced by the area for human use). This makes it clare that the area is for productive human use.

>
> For the state forests and wildlife management areas around here, I tag
> at least boundary=protected_area. (Tag with the right protect_class,
> and add leisure=nature_reserve if it fits: 'nature reserve' covers a
> lot of things.) If I'm mapping land cover (I seldom do), I will use
> natural=wood to mean 'tree cover' and let others fight over it.

I too use natural=wood with landcover=trees to map a tree area.

--------------------------
The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.
The essential difference between the two is that landuse must have some human benefit, a produce, and a clear way of doing that is to add the rendering of a axe to the tree.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging



_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Joseph Eisenberg
> “Once you combine the OSM keys and values of landuse=forest and compare it to natural=wood I think most will agree there is a difference,”

I certainly do, but I’m a native speaker of English (though not the British variety).

Many speakers of other languages just search for an English word in an online traslation service and then stick that into the editor to find a tag.
On Mon, Jan 21, 2019 at 1:34 PM Warin <[hidden email]> wrote:
On 21/01/19 10:17, Joseph Eisenberg wrote:
> The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.

Until several years ago the “standard” style (Openstreetmap-Carto) did show a difference between landuse=forest and natural=wood. However, mappers used these two tags interchangeably even then. The rendering was changed to match actually database usage on a global scale, which is that both tags are often used to tag any area covered with trees.

The current rendering follows tag usage and the current wiki page, which also discusses this issue in depth.

I wish it were possible to fix this, but the different meanings of “forest” and “wood” in various English dialects make it difficult,

The meaning of the key 'landuse' is fairly clear in any English dialect.

The problem of the key 'natural' remains.

Once you combine the OSM keys and values of landuse=forest and compare it to natural=wood I think most will agree there is a difference,
The former is for what the land is used for.
The latter is for the presence of plants, if you take any plant as being natural then natural=wood is 'acceptable' for ant tree area.' 

even before we add other languages and cultures to the mix.
On Mon, Jan 21, 2019 at 8:04 AM Warin <[hidden email]> wrote:
On 21/01/19 05:52, Kevin Kenny wrote:

> On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
>> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?
> This again.

And it will continue to occur!

And reoccur, again and again.

>
> There's a failed consensus here - and you risk reversion with either decision.
>
> I tend to follow the principle that landuse=* denotes the land USE,
> not the land COVER, so I don't demand that every square metre of
> landuse=forest be covered by trees.

+1

>   But many do, and the renderer
> follows their inclination.
>
> natural=wood is a possibility to show tree cover - but that leads some
> to argue that it has to be a 'natural' wood - whatever that means.
> I've heard it argued that the 'old second growth' forest that's
> increasingly common near me is still not 'natural' because a skilled
> forester can still find the human impact. (Of course, that was true
> even before the Europeans arrived - there was considerable
> pre-Columbian human impact on these forests.)

Those who argue this have no problem abusing the landuse tag, so I see no reason why the tag 'natural' cannot be similarly abused.
The OSMwiki for 'natural' even states that is can be used for human effected things.

>
> landcover=trees doesn't render, but is at least unambiguous that it
> means tree cover and nothing else.
>
> landuse=forestry, for a managed forest, has been proposed but received
> a lukewarm reception.

For forestry area I tag landuse=forest with produce=trees (or what ever is produced by the area for human use). This makes it clare that the area is for productive human use.

>
> For the state forests and wildlife management areas around here, I tag
> at least boundary=protected_area. (Tag with the right protect_class,
> and add leisure=nature_reserve if it fits: 'nature reserve' covers a
> lot of things.) If I'm mapping land cover (I seldom do), I will use
> natural=wood to mean 'tree cover' and let others fight over it.

I too use natural=wood with landcover=trees to map a tree area.

--------------------------
The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.
The essential difference between the two is that landuse must have some human benefit, a produce, and a clear way of doing that is to add the rendering of a axe to the tree.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Warin
On 21/01/19 16:47, Joseph Eisenberg wrote:
> “Once you combine the OSM keys and values of landuse=forest and compare it to natural=wood I think most will agree there is a difference,”

I certainly do, but I’m a native speaker of English (though not the British variety).

Many speakers of other languages just search for an English word in an online traslation service and then stick that into the editor to find a tag.

I would hope the OSM wiki would be better than that. It certainly could be a better method for mappers looking for some correct tags.

I recently came across 'sport=paddleboard' (non english speaking palce) and thought of table tennis .. contacted the mapper and no it is ....in my local version of English, a 'rescue board' used by surf life savers to rescue people.
Yes they do have completions for it, but no permanent infrastructure here so it does not get mapped. Only one instance in OSM so no wiki page for it.


On Mon, Jan 21, 2019 at 1:34 PM Warin <[hidden email]> wrote:
On 21/01/19 10:17, Joseph Eisenberg wrote:
> The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.

Until several years ago the “standard” style (Openstreetmap-Carto) did show a difference between landuse=forest and natural=wood. However, mappers used these two tags interchangeably even then. The rendering was changed to match actually database usage on a global scale, which is that both tags are often used to tag any area covered with trees.

The current rendering follows tag usage and the current wiki page, which also discusses this issue in depth.

I wish it were possible to fix this, but the different meanings of “forest” and “wood” in various English dialects make it difficult,

The meaning of the key 'landuse' is fairly clear in any English dialect.

The problem of the key 'natural' remains.

Once you combine the OSM keys and values of landuse=forest and compare it to natural=wood I think most will agree there is a difference,
The former is for what the land is used for.
The latter is for the presence of plants, if you take any plant as being natural then natural=wood is 'acceptable' for ant tree area.' 

even before we add other languages and cultures to the mix.
On Mon, Jan 21, 2019 at 8:04 AM Warin <[hidden email]> wrote:
On 21/01/19 05:52, Kevin Kenny wrote:

> On Sun, Jan 20, 2019 at 1:33 PM David Marchal <[hidden email]> wrote:
>> All is in the title: when hiking in a forest (I mean, an area considered as a forest by authorities), I often encounter other landcovers, like scrubs in recently teared down parcels, or scree in the mountains. These area, although, clearly and morphologically, not a forest, are still mapped as such, as they are considered to be part of the forest and are treated this may, but they are morphologically not the forest: the forest is the area administratively regarded as such, but it is not always the case; if I want, for instance, to map them as a scrub area of the forest, as the polygons overlapped, they are rendered in a mixed way. Is there a recommended way of handling such cases without broking display? If so, what are they? The landcover tag? boundary=forest_compartment? Another?
> This again.

And it will continue to occur!

And reoccur, again and again.

>
> There's a failed consensus here - and you risk reversion with either decision.
>
> I tend to follow the principle that landuse=* denotes the land USE,
> not the land COVER, so I don't demand that every square metre of
> landuse=forest be covered by trees.

+1

>   But many do, and the renderer
> follows their inclination.
>
> natural=wood is a possibility to show tree cover - but that leads some
> to argue that it has to be a 'natural' wood - whatever that means.
> I've heard it argued that the 'old second growth' forest that's
> increasingly common near me is still not 'natural' because a skilled
> forester can still find the human impact. (Of course, that was true
> even before the Europeans arrived - there was considerable
> pre-Columbian human impact on these forests.)

Those who argue this have no problem abusing the landuse tag, so I see no reason why the tag 'natural' cannot be similarly abused.
The OSMwiki for 'natural' even states that is can be used for human effected things.

>
> landcover=trees doesn't render, but is at least unambiguous that it
> means tree cover and nothing else.
>
> landuse=forestry, for a managed forest, has been proposed but received
> a lukewarm reception.

For forestry area I tag landuse=forest with produce=trees (or what ever is produced by the area for human use). This makes it clare that the area is for productive human use.

>
> For the state forests and wildlife management areas around here, I tag
> at least boundary=protected_area. (Tag with the right protect_class,
> and add leisure=nature_reserve if it fits: 'nature reserve' covers a
> lot of things.) If I'm mapping land cover (I seldom do), I will use
> natural=wood to mean 'tree cover' and let others fight over it.

I too use natural=wood with landcover=trees to map a tree area.

--------------------------
The end to this madness is for renders to recognise that the landuse=forest needs to be rendered differently from natural=wood.
The essential difference between the two is that landuse must have some human benefit, a produce, and a clear way of doing that is to add the rendering of a axe to the tree.


______________________________


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Paul Allen
In reply to this post by Andy Townsend
On Mon, 21 Jan 2019 at 04:20, Andy Townsend <[hidden email]> wrote:

One suggestion that I've made here before is explicitly to use
"landuse=forestry" for areas that may or may not have trees on them, if
the areas with trees within have been mapped separately
 
You're not the only one to have made that suggestion.  It makes a lot of sense, since
the original intent for landuse=forest was for forestry and the natural language/OSM
mismatch is one reason the tag is often used for a different purpose than intended.

I've mapped several areas of trees where the OS_OpenData_StreetView layer shows a
different extent than is visible in aerial imagery. - sometimes a lesser extent, sometimes
a greater extent.  And in some of those cases where the OS layer is larger than visible
in aerial imagery, the aerial imagery shows a fence matching up with the OS layer AND
what appears to be tree stumps or scrub or young trees or whatever where the two views
disagree.  If I map the visible extent of the trees, years from now somebody will have to
change the outline to match new growth.  If I include tree stumps then somebody might
change the outline the next day to match what is visible.  Having landuse=forestry that
really does mean forestry (as opposed to landuse=forest that was intended to mean forestry
but rarely does) would deal with some of the issues.  It would be up to the mapper to decide
whether it's worth the hassle of using landcover=trees to show the current extent of trees.

[...]
That renderer also processes landuse=forest the same way - see
https://www.openstreetmap.org/way/44018882 and
https://map.atownsend.org.uk/maps/map/map.html#zoom=15&lat=53.21319&lon=-1.18217
for an example of that.

And there's the rub.  The standard carto ignores landuse=forestry.  Which means that people
end up tagging for the renderer by using landuse=forest or natural=wood.  Because woodland
is tedious to map and there's no point going to all that effort if it's not going to render.  It's
unrealistic to expect most mappers to use landuse=forestry unless it renders.

Around and around we go.  This list cannot agree on approving landuse=forestry because it
doesn't get rendered.  The carto people refuse to render landuse=forestry because nobody
uses it.  Sometimes the semi-anarchic nature of OSM tagging can be very frustrating.  There
are days when I yearn for joined-up thinking.

How about...  I expect it will get shouted down for many reasons, but...

What if we suggest in the wiki that where trees are used for actual forestry people are
encouraged to dual-tag with landuse=forestry + natural=wood on the basis that with
enough usage the carto group will render landuse=forestry AND that when they do there
will be an effort to remove natural=wood when it appears in combination with
landuse=forestry.  What was I thinking?  That might actually get us somewhere, and we
wouldn't want to do that.

--
Paul


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Silent Spike
On Mon, Jan 21, 2019 at 12:45 PM Paul Allen <[hidden email]> wrote:
Around and around we go.  This list cannot agree on approving landuse=forestry because it
doesn't get rendered.  The carto people refuse to render landuse=forestry because nobody
uses it.  Sometimes the semi-anarchic nature of OSM tagging can be very frustrating.  There
are days when I yearn for joined-up thinking.

How about...  I expect it will get shouted down for many reasons, but...

What if we suggest in the wiki that where trees are used for actual forestry people are
encouraged to dual-tag with landuse=forestry + natural=wood on the basis that with
enough usage the carto group will render landuse=forestry AND that when they do there
will be an effort to remove natural=wood when it appears in combination with
landuse=forestry.  What was I thinking?  That might actually get us somewhere, and we
wouldn't want to do that.

I tend to share the same sentiment as yourself regarding this seemingly endless back and forth between rendering and tagging. It clearly indicates where change would be beneficial and a join system could be established by both to move things forward (similar to how there's a clear system for tag proposals, it might not be perfect, but it promotes progress).

As for your suggestion, this is already my approach for "landuse=grass" and "landcover=grass". Tag both, hope that one day the old confusing tag has less isolated usage than the combination and someone realises we can remove the old one from all combined cases. 

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Kevin Kenny-3
In reply to this post by Paul Allen
On Mon, Jan 21, 2019 at 7:45 AM Paul Allen <[hidden email]> wrote:
> What if we suggest in the wiki that where trees are used for actual forestry people are
> encouraged to dual-tag with landuse=forestry + natural=wood on the basis that with
> enough usage the carto group will render landuse=forestry AND that when they do there
> will be an effort to remove natural=wood when it appears in combination with
> landuse=forestry.  What was I thinking?  That might actually get us somewhere, and we
> wouldn't want to do that.

Your post was dead on target, right up to here -  and the dual-tagging
suggestion is a good one. I do that a lot - precise tagging that
doesn't render combined with imprecise tagging 'for the renderer'. By
this I do not mean tagging falsehoods because they render nicely,
which is unacceptable, but rather tagging features like
'leisure=nature_reserve', which covers anything from the vacant lot
that the city has set aside for birds to the vast tracts of a national
forest. I can try to be accurate with protect_class and the like while
not sacrificing the ability to have the tagged feature show up on
maps.

Nevertheless, even if all the intended tags render correctly, there's
nothing wrong with tagging both land USE and land COVER - which are
two different things. One is human and social - "to what use are
people putting this land?" "They're growing trees on it."  The other
is physical, "if I look down on this land from above, what will I
see?" "It's a beaver pond in the middle of a forest."

A landuse=forest[ry] will surely be largely natural=wood and/or
landcover=trees. But forestry is a long game. Near me there are state
forests that are unquestionably managed for production (with public
recreation a secondary goal), that are mostly 'trees', but some
'grass' or 'scrub' (recently acquired parcels, or recently logged
ones), and even some water and wetland (thank you, beavers!).  The
fact that an area is a beaver pond today doesn't mean that it won't
progress through marsh, scrub, laurel meadow, alder thicket, and so on
to forest over the years - and the land is managed for the long haul,
in the expectation of such a succession. Forestry is the land USE -
the land COVER varies. Both are important.

Putting landcover=trees or natural=wood on the pond is wrong. There's
no wood there.  Putting landuse=forest[ry] on the pond may be right.
When the beavers move away, that land will be productive again . In
fact, it will likely be more productive than before they arrived.
There's pretty solid evidence that beavers improve the forest over
time.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Warin
On 22/01/19 04:29, Kevin Kenny wrote:

> On Mon, Jan 21, 2019 at 7:45 AM Paul Allen <[hidden email]> wrote:
>> What if we suggest in the wiki that where trees are used for actual forestry people are
>> encouraged to dual-tag with landuse=forestry + natural=wood on the basis that with
>> enough usage the carto group will render landuse=forestry AND that when they do there
>> will be an effort to remove natural=wood when it appears in combination with
>> landuse=forestry.  What was I thinking?  That might actually get us somewhere, and we
>> wouldn't want to do that.
> Your post was dead on target, right up to here -  and the dual-tagging
> suggestion is a good one. I do that a lot - precise tagging that
> doesn't render combined with imprecise tagging 'for the renderer'. By
> this I do not mean tagging falsehoods because they render nicely,
> which is unacceptable, but rather tagging features like
> 'leisure=nature_reserve', which covers anything from the vacant lot
> that the city has set aside for birds to the vast tracts of a national
> forest. I can try to be accurate with protect_class and the like while
> not sacrificing the ability to have the tagged feature show up on
> maps.
>
> Nevertheless, even if all the intended tags render correctly, there's
> nothing wrong with tagging both land USE and land COVER - which are
> two different things. One is human and social - "to what use are
> people putting this land?" "They're growing trees on it."

It is more than just growing.

The growing thing has to produce something for human benefit, usually some form of harvesting to provide a produce.

Grass can be grown to produce turf. Trees can be grown to produce timber.

If there is no human benefit from the plant growth then the landuse cannot be the plant.
The area may not have any 'land use' (i.e. a wasteland) or it may be used for conservation, but it is not related to the plant.

> The other
> is physical, "if I look down on this land from above, what will I
> see?" "It's a beaver pond in the middle of a forest."
>
> A landuse=forest[ry] will surely be largely natural=wood and/or
> landcover=trees. But forestry is a long game. Near me there are state
> forests that are unquestionably managed for production (with public
> recreation a secondary goal), that are mostly 'trees', but some
> 'grass' or 'scrub' (recently acquired parcels, or recently logged
> ones), and even some water and wetland (thank you, beavers!).  The
> fact that an area is a beaver pond today doesn't mean that it won't
> progress through marsh, scrub, laurel meadow, alder thicket, and so on
> to forest over the years - and the land is managed for the long haul,
> in the expectation of such a succession. Forestry is the land USE -
> the land COVER varies. Both are important.
>
> Putting landcover=trees or natural=wood on the pond is wrong. There's
> no wood there.  Putting landuse=forest[ry] on the pond may be right.
> When the beavers move away, that land will be productive again . In
> fact, it will likely be more productive than before they arrived.
> There's pretty solid evidence that beavers improve the forest over
> time.

Some forestry operations used to use rivers to transport the timber down stream to a lake.

-------------------

My problem with going to landuse=forestry with natural=wood...

what happens to the remaining landuse=forest?
Will that finally be recognised as the same as natural=wood and be migrated to natural=wood???


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Paul Allen
On Mon, 21 Jan 2019 at 20:21, Warin <[hidden email]> wrote:

My problem with going to landuse=forestry with natural=wood...

what happens to the remaining landuse=forest?
Will that finally be recognised as the same as natural=wood and be migrated to natural=wood???

Ideally, if we get landuse=forestry and it eventually renders, landuse=forest would be
deprecated and slowly replaced when a mapper encounters it.  It's a misbegotten tag
that has been used inconsistently.  It was intended to mean what the suggested
landuse=forestry means, but has largely been used to mean what natural=wood means.

landuse=forest is wrong two ways.  A forest is not landuse.  You might be able to justify
landcover=forest but that's already dealt with by landcover=trees.  You might be able to
make an argument for natural=forest (a big wood) in the same way we draw a
distinction between rivers and streams.  The only way it can be considered landuse is
if the land is used for forestry, but then we have the mismatch with natural English which
is part of the reason it was misused and part of the reason people keep proposing
landuse=forestry.

Any migration would have to be on a case-by-case basis.  If land used for forestry is
tagged as landuse=forest it should (eventually) be migrated to landuse=forestry.
If not used for forestry then landcover=trees or natural=wood.

But all that requires that this list and the carto people manage to get all our shit in the
same sock.

Maybe it's worth a formal proposal for landuse=forestry suggesting dual-tagging as
an interim workaround for it not being rendered, with a later clean-up.  Because we're
going to keep coming back to this one until we finally do something about it.

--
Paul


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

David Marchal
Paul,

Your landuse=forestry proposal seems good to me: it is clear enough, and the transition process you describe here seems consistent with what I know about such transitions which already happened. If I understand you, the main problem for landuse=forestry is to include it in the standard style to not discourage its use, but style devs rejected adding its rendering before its use spread a bit. Some sort of vicious circle, in fact?

Awaiting your answer,

Regards.

> Le 21 janv. 2019 à 21:44, Paul Allen <[hidden email]> a écrit :
>
> Ideally, if we get landuse=forestry and it eventually renders, landuse=forest would be
> deprecated and slowly replaced when a mapper encounters it.  It's a misbegotten tag
> that has been used inconsistently.  It was intended to mean what the suggested
> landuse=forestry means, but has largely been used to mean what natural=wood means.
>
> landuse=forest is wrong two ways.  A forest is not landuse.  You might be able to justify
> landcover=forest but that's already dealt with by landcover=trees.  You might be able to
> make an argument for natural=forest (a big wood) in the same way we draw a
> distinction between rivers and streams.  The only way it can be considered landuse is
> if the land is used for forestry, but then we have the mismatch with natural English which
> is part of the reason it was misused and part of the reason people keep proposing
> landuse=forestry.
>
> Any migration would have to be on a case-by-case basis.  If land used for forestry is
> tagged as landuse=forest it should (eventually) be migrated to landuse=forestry.
> If not used for forestry then landcover=trees or natural=wood.
>
> But all that requires that this list and the carto people manage to get all our shit in the
> same sock.
>
> Maybe it's worth a formal proposal for landuse=forestry suggesting dual-tagging as
> an interim workaround for it not being rendered, with a later clean-up.  Because we're
> going to keep coming back to this one until we finally do something about it.
>
> --
> Paul
>
> _______________________________________________
> Tagging mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/tagging

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Paul Allen
On Tue, 22 Jan 2019 at 18:14, David Marchal <[hidden email]> wrote:

Your landuse=forestry proposal seems good to me: it is clear enough, and the transition process you describe here seems consistent with what I know about such transitions which already happened. If I understand you, the main problem for landuse=forestry is to include it in the standard style to not discourage its use,

Yup.  If it rendered, people who read this list would use it.  If enough people used it, editors
would offer it as a preset (for iD somebody would have to raise the issue on github since
Bryan Housel recently announced he was no longer following this list).  A couple of vicious
circles there.

but style devs rejected adding its rendering before its use spread a bit.

I don't know if they have rejected this specific idea, or even if they were asked.  It's just
that they often require that a tag has been used sufficiently in the wild before they consider
adding it.

Some sort of vicious circle, in fact?

As I said, two of them.  It won't be widely used until editors offer it as a preset and it
renders.  So we're at an impasse.  A proposal to introduce it that suggests dual-tagging
until it takes off enough for editors and carto to support it seems the only way forward -
not guaranteed to succeed but it might.

I might even write the proposal myself.  But only after I get a feel for the mood here.  So
far nobody has heaped scorn on the idea, which is a good sign, but I'd like to see a little
more support first because if people here don't see it as sensible then neither will
most ordinary mappers.

--
Paul


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Mateusz Konieczny-3
Jan 22, 2019, 8:38 PM by [hidden email]:
On Tue, 22 Jan 2019 at 18:14, David Marchal <[hidden email]> wrote:

Your landuse=forestry proposal seems good to me: it is clear enough, and the transition process you describe here seems consistent with what I know about such transitions which already happened. If I understand you, the main problem for landuse=forestry is to include it in the standard style to not discourage its use,

Yup.  If it rendered, people who read this list would use it

It is completely normal for tags to not be rendered at beginning, before getting popular.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Mateusz Konieczny-3
In reply to this post by Paul Allen



Jan 21, 2019, 9:44 PM by [hidden email]:
On Mon, 21 Jan 2019 at 20:21, Warin <[hidden email]> wrote:

My problem with going to landuse=forestry with natural=wood...

what happens to the remaining landuse=forest?
Will that finally be recognised as the same as natural=wood and be migrated to natural=wood???

Ideally, if we get landuse=forestry and it eventually renders
Note that it is not obvious to me that a typical map would want to display landuse=forestry.


_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
Reply | Threaded
Open this post in threaded view
|

Re: Forest parcel with other landcover (scrub, scree…): how to map?

Kevin Kenny-3
In reply to this post by Paul Allen
On Tue, Jan 22, 2019 at 2:39 PM Paul Allen <[hidden email]> wrote:
> I don't know if they have rejected this specific idea, or even if they were asked.  It's just
> that they often require that a tag has been used sufficiently in the wild before they consider
> adding it.


A commoner objection is that it's redundant with landuse=forest. That
would even be true if anyone tagged landuse=forest to mean what it
says on the box.

_______________________________________________
Tagging mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/tagging
1234