Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Julien Minet]

Hi Joost,

Yes, I'm interested in your land-use analysis that you did. Did you share
or present the code somewhere? Maybe we can plan smthg for the next
FOSS4G.be event?

A discussion about a tag "forest_management_style=*"? Wow, it can be
dangerous, especially if we have hunting supporters vs timber producers vs
naturalists in the osm tagging list;-). I doubt that we can find global tag
that apply to the many different way forests can be managed. Otherwise,
certification of the forests (eg FSC, PEFC) can be less equivocal and I've
seen this tag was already proposed
.

Julien

On Thu, Apr 27, 2017 at 10:03 AM, joost schouppe 
wrote:

> Hi Julien,
>
> * How would you feel about building a proposal about
> forest_management_style=* ? To my great surprise, I noticed I kind of like
> mingling in the endless discussions at the tagging mailing list. So I would
> be willing to help out.
>
> * Quantitative analysis of landuse mapping in Belgium: I did that. The
> idea was to generate a useful dataset of landuse at the level of the
> statistical sectors. So I made a classification of several layers
> (residential, nature, water, transport infrastructure). Then I did some GIS
> processing: to create polygons out of roads and POIs. Then I needed to
> choose in which layer to count the landuse, for example sometimes there's a
> huge residential area with everything just mapped on top of that. So then
> you need to decide if a park in a residential area is a park or a
> residential area.
> I can share method and result. We could set up a voice meeting on Riot (
> https://riot.im/app/#/room/#osmbe:matrix.org) so others can join in case
> they're interested.
>
> * The end result should be a landuse convention, yes.
>
> 2017-04-27 9:51 GMT+02:00 Julien Minet :
>
>> Thanks for your reactions!
>>
>> * About the tag natural=wood, I also think it is over-represented in
>> Belgium. Belgian forests are indeed not only managed for timber production
>> but also hunting, tourism and nature conservation, but often in an
>> integrated manner under the same areas (at least in theory!). Note that the
>> Natura 2000 program does not preclude at all that timber wood is produced!
>> See on this link how much forests in south of Wallonia are covered by
>> Natura 2000: 
>> http://geoportail.wallonie.be/walonmap#SHARE=4E2203C158780AB5E053D0AFA49D7D23.
>>
>> 
>>
>> * I did not talk about the landcover tag in my article but it'd be worth
>> talking more about it!
>>
>> * It would be interesting to quantitavely analyse the ways landuse is
>> mapped in Belgium. I'm thinking about it. Any ideas on how, what kind of
>> analysis are welcome...
>>
>> * Summarizing this discussion in osm.be could be nice. But why not also
>> create a "landuse convention" page on the OSM wiki
>> ?
>>
>>
>> Julien
>>
>> On Wed, Apr 26, 2017 at 10:26 AM, joost schouppe <
>> joost.schou...@gmail.com> wrote:
>>
>>> *  About forests, I tend to agree with the natural=wood not really
>>> existing in Belgium. The only exception I know of is a bit of the
>>> Zoniënwoud (Kersselaerplein) that has had "zero management" for 34 years
>>> now.
>>> But most natural=wood I've seen is wrong.
>>>
>>> Just recently, I changed the Bois de La Houssiere (
>>> http://www.openstreetmap.org/#map=14/50.6189/4.1948) from wood to
>>> forest. It's a bit of a special case: it's a Natura 2000 protected area,
>>> but it is also actively used as a forestry area.
>>>
>>> I agree with the comments above that landuse=forest for any kind of
>>> group of trees is annoying too.
>>>
>>> - In cases where you have residential areas in a forest, or wooded areas
>>> in gardens, maybe we should really encourage the use of the landcover tag?
>>> - In cases where forests are managed, but as some kind of nature reserve
>>> or natural area, maybe we could use a subtag to indicate the management
>>> style? That would allow to differentiate between real forestry and forests
>>> with nature-friendly management. You could use one of the many nature
>>> reserve tags of course, but I'm not sure all naturally managed forests are
>>> protected and the Bois de la Houssiere shows the opposite also exists.
>>>
>>>
>>> * On a more detailed note: I had never heard of the taxon tag, I've only
>>> used species before. I'm completely confused now :)
>>> And are the values REALLY comma separated, not " ; " seperated?
>>>
>>> * About OSM.be: we're still thinking about what exactly we want to use
>>> the Projects for - the fact we don't really know was shown quite clearly by
>>> Marc's latest article.
>>>
>>> I think we could have an OSM.be project on "Harmonizing tagging in
>>> OpenStreetMap". It would first explain really short how tagging works, and
>>> why it can be 

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

I have no problem to use landuse=forest, and not natural=wood for any
patch that is "big enough". I would not use it in gardens, especially
not private garden. I would try to avoid using it when
natural=tree_row is the better choice.

I think landuse=residential is wrong under amenity=school,
place_of_worship (except for living area of abbeys). It would be
better if we had landuse=education / worship (or something similar).
We could recommend to avoid such overlapping.
I also saw some discussion recently to let e.g. amenity=pub enclose
the complete area with parking, playground etc. of the pub. Some
people even split landuse=farmyard, to exclude the living area and use
landuse=residential there. I think there is a large gap between the
current situation and the very fine detailed mapping of landuse some
other mappers are applying. While we can recommend the fine grained
mapping, I wonder whether what is the most important thing we have to
do with landuse right now.

I am more annoyed by connected landuse over paths/tracks. If there is
a forest on one side and farmland on the other side of the track, the
map should reflect this. So the landuse has to split and both have the
end on different sides of the track. But I assume everybody has their
own "battles" and priorities.

m.



On Sat, Apr 29, 2017 at 3:42 PM, Lionel Giard  wrote:
> That's definitely an interesting answer. It seems that dividing the large
> landuse=residential is something that we should do (as it seems logical,
> even if can be tedious sometimes).
>
> I did some digging into the wiki for trees tagging and came to these
> conclusions. When we think about the key definition of landuse, landcover
> (including its proposed page) and natural, i came to the conclusion that the
> only landuse tag for tree is "landuse=forest". Because the key "natural" is
> described as a landcover representation. It seems that natural=wood is one
> of the only special case where a natural tag does represent landuse in
> common usage (and it seems wrong relatively to the definition).
>
> If we follow strict definition, the only landuse tag for trees/forest is
> "landuse=forest". The others are for landcover (like the proposed
> landcover=trees). Should we then be conservative and use only landuse=forest
> in Belgium (especially because the definition for natural=wood is very rare
> for us) ? And use landcover tag on top of others landuses if needed (like
> for tree in parks).
>
> Following all these definition note that landuse include the keys landuse=*,
> amenity=*, leisure=* and tourism=* all as landuse representation, it implies
> that we should also remove the landuse=residential (or any other) where we
> have something like amenity=school (because it is already a landuse that
> probably better fit than the landuse=residential).  What do you think about
> that ?
>
> 2017-04-28 22:16 GMT+02:00 Marc Gemis :
>>
>> Here is one answer I got, Martin was so kind to put it into a diary
>> entry: http://www.openstreetmap.org/user/dieterdreist/diary/40993
>>
>> On Fri, Apr 28, 2017 at 3:19 PM, Marc Gemis  wrote:
>> > On Fri, Apr 28, 2017 at 2:23 PM, Lionel Giard 
>> > wrote:
>> >> But for the roads, ideally, it should ideally be an area (like on the
>> >> GRB of
>> >> Vlaandereen or the PICC of Wallonia) with also the existing line to
>> >> allow
>> >> routing. I don't know, if we must change existing residential area when
>> >> adding area for the road, because it will probably look good on the
>> >> map, but
>> >> maybe it would be a problem for people using the data ?! At least it
>> >> shouldn't be a problem for the big highways, because they often don't
>> >> have
>> >> landuse at the moment (look at http://osmlanduse.org/ ).
>> >
>> > as long as you keep the current way for navigation, and just add
>> > area:highway there is no problem.
>> > Just follow the area:highway instructions on the wiki and the
>> > navigation will not get broken. I experimented with in on a small area
>> > and navigation still works.
>> >
>> > I contacted 2 mappers that map landuse in great detail (one in
>> > Germany/Italy, one in Japan) and asked them for some samples.
>> > I doubt that multipolygons are the way forward, too complex to
>> > maintain I fear. We should look at detailed areas in e.g. Germany and
>> > see how they do it.
>> >
>> >
>> > m
>>
>> ___
>> Talk-be mailing list
>> Talk-be@openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/talk-be
>
>
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Lionel Giard]

That's definitely an interesting answer. It seems that dividing the large
landuse=residential is something that we should do (as it seems logical,
even if can be tedious sometimes).

I did some digging into the wiki for trees tagging and came to these
conclusions. When we think about the key definition of landuse
, landcover
 (including its proposed page
) and
natural , i came to the
conclusion that the only landuse tag for tree is "landuse=forest". Because
the key "natural" is described as a landcover representation. It seems that
natural=wood is one of the only special case where a natural tag does
represent landuse in common usage (and it seems wrong relatively to the
definition).

If we follow strict definition, the only landuse tag for trees/forest is
"landuse=forest". The others are for landcover (like the proposed
landcover=trees). Should we then be conservative and use only
landuse=forest in Belgium (especially because the definition for
natural=wood is very rare for us) ? And use landcover tag on top of others
landuses if needed (like for tree in parks).

Following all these definition note that landuse include the keys
landuse=*,  amenity=*, leisure=* and tourism=* all as landuse
representation, it implies that we should also remove the
landuse=residential (or any other) where we have something like
amenity=school (because it is already a landuse that probably better fit
than the landuse=residential).  What do you think about that ?

2017-04-28 22:16 GMT+02:00 Marc Gemis :

> Here is one answer I got, Martin was so kind to put it into a diary
> entry: http://www.openstreetmap.org/user/dieterdreist/diary/40993
>
> On Fri, Apr 28, 2017 at 3:19 PM, Marc Gemis  wrote:
> > On Fri, Apr 28, 2017 at 2:23 PM, Lionel Giard 
> wrote:
> >> But for the roads, ideally, it should ideally be an area (like on the
> GRB of
> >> Vlaandereen or the PICC of Wallonia) with also the existing line to
> allow
> >> routing. I don't know, if we must change existing residential area when
> >> adding area for the road, because it will probably look good on the
> map, but
> >> maybe it would be a problem for people using the data ?! At least it
> >> shouldn't be a problem for the big highways, because they often don't
> have
> >> landuse at the moment (look at http://osmlanduse.org/ ).
> >
> > as long as you keep the current way for navigation, and just add
> > area:highway there is no problem.
> > Just follow the area:highway instructions on the wiki and the
> > navigation will not get broken. I experimented with in on a small area
> > and navigation still works.
> >
> > I contacted 2 mappers that map landuse in great detail (one in
> > Germany/Italy, one in Japan) and asked them for some samples.
> > I doubt that multipolygons are the way forward, too complex to
> > maintain I fear. We should look at detailed areas in e.g. Germany and
> > see how they do it.
> >
> >
> > m
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

Here is one answer I got, Martin was so kind to put it into a diary
entry: http://www.openstreetmap.org/user/dieterdreist/diary/40993

On Fri, Apr 28, 2017 at 3:19 PM, Marc Gemis  wrote:
> On Fri, Apr 28, 2017 at 2:23 PM, Lionel Giard  wrote:
>> But for the roads, ideally, it should ideally be an area (like on the GRB of
>> Vlaandereen or the PICC of Wallonia) with also the existing line to allow
>> routing. I don't know, if we must change existing residential area when
>> adding area for the road, because it will probably look good on the map, but
>> maybe it would be a problem for people using the data ?! At least it
>> shouldn't be a problem for the big highways, because they often don't have
>> landuse at the moment (look at http://osmlanduse.org/ ).
>
> as long as you keep the current way for navigation, and just add
> area:highway there is no problem.
> Just follow the area:highway instructions on the wiki and the
> navigation will not get broken. I experimented with in on a small area
> and navigation still works.
>
> I contacted 2 mappers that map landuse in great detail (one in
> Germany/Italy, one in Japan) and asked them for some samples.
> I doubt that multipolygons are the way forward, too complex to
> maintain I fear. We should look at detailed areas in e.g. Germany and
> see how they do it.
>
>
> m

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

On Fri, Apr 28, 2017 at 2:23 PM, Lionel Giard  wrote:
> But for the roads, ideally, it should ideally be an area (like on the GRB of
> Vlaandereen or the PICC of Wallonia) with also the existing line to allow
> routing. I don't know, if we must change existing residential area when
> adding area for the road, because it will probably look good on the map, but
> maybe it would be a problem for people using the data ?! At least it
> shouldn't be a problem for the big highways, because they often don't have
> landuse at the moment (look at http://osmlanduse.org/ ).

as long as you keep the current way for navigation, and just add
area:highway there is no problem.
Just follow the area:highway instructions on the wiki and the
navigation will not get broken. I experimented with in on a small area
and navigation still works.

I contacted 2 mappers that map landuse in great detail (one in
Germany/Italy, one in Japan) and asked them for some samples.
I doubt that multipolygons are the way forward, too complex to
maintain I fear. We should look at detailed areas in e.g. Germany and
see how they do it.


m

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Lionel Giard]

I understand your point, and that's why we should say if something must be
done or not : describing the best practices.
For landuse inside an existing residential area, it is always possible to
just change the residential zone into a multipolygon relation and make the
new landuse (like the park) an inner polygon. So it will still display the
same on the map, but will not count as having two landuse at the same
place. It avoids to tediously redraw a large residential area.

But for the roads, ideally, it should ideally be an area (like on the GRB
of Vlaandereen or the PICC of Wallonia) with also the existing line to
allow routing. I don't know, if we must change existing residential area
when adding area for the road, because it will probably look good on the
map, but maybe it would be a problem for people using the data ?! At least
it shouldn't be a problem for the big highways, because they often don't
have landuse at the moment (look at http://osmlanduse.org/ ).

About the area you linked, I really don't have a definitive answer, is the
wooded area part of the garden of the houses ? Maybe do a mix of
landuse=residential and landuse=forest/lancover=trees (to have the house
and garden not included in the forest) ?
___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

the landuse under roads should be highway, this has been discussed in
the past. Never got popular. Alternative is area:highway, which is
more popular I believe.

But you won't split the current residential areas on each street, would you ?
I think overlaying landuse=residential with leisure=park is just an
easy way to map, otherwise you would have to start splitting all large
residential landuses each time, you map amenity=place_of_worship,
parkings, parks, gardens, etc. etc.

But strictly speaking you are correct, a park is not
landuse=residential, but is a pond in a private garden ? Where do you
draw the line ?

There are mappers who loves to see more different landuses for civic
buildings, church grounds, schools  etc.

When you ask for no gaps in landuse, what is the landuse of a beach ?
of a river ? of a small ditch or path between two fields ?  What is
the use of a tree row along a road ? Of a grass patch that separates
the cycleway from the road ? Do we really want to split up landuse up
to this level ? now, or is this a long term goal ?

Right now, I am more concerned about areas like
http://osm.org/go/0Erc0BNv?m= . Which message does the current map
give you about this area around the Prinsenlaan ? Please first look at
the map and then you might look at aerial imagery or other sources ( I
hope I have some images at
https://xian.smugmug.com/OSM/OSM-2017/2017-04-17-Retie-Brug-2 soon)
Does it match ? Let me know.

@Joost
I do not mind you created your own tool, it's great that there are
multiple tools to analyse landuse. Each with their own benefits (e.g.
easy of use vs. customizability)

m.

On Thu, Apr 27, 2017 at 5:21 PM, Lionel Giard  wrote:
> Personally, i put Nature reserve on a special relation, as it is described
> on the wiki, like this one :
> http://www.openstreetmap.org/relation/7130732#map=17/50.68519/4.70461
> And the forest or other landuses are just part of this multipolygon. In this
> example, i also have a multipolygon for the forest because, i have things
> inside it.
> Nothing stop you to have multiple relation at the same place (i think of
> relation like a "special" polygon, and nothing stop us to make multiple
> polygon on each other but slightly different in shape).
>
> We should really describe/decide which tag is representing a landuse
> (because it can be landuse=*, leisure=* or natural=*) and if we need to
> avoid putting them at the same place (like allow or not leisure=park on top
> of landuse=residential ?!). Right know, we all have our own opinion and it
> create some variation of the map in Belgium.
> And we should avoid gap in the landuse. Thus it seems important to solve
> some problems like what should be the landuse tag under a road ?
>
> 2017-04-27 14:29 GMT+02:00 Marc Gemis :
>>
>> On Thu, Apr 27, 2017 at 1:50 PM, joost schouppe
>>  wrote:
>> > The examples you give are already hard work to think about. Much more
>> > basic
>> > mistakes are made too: e.g. a forest is also a nature reserve. But then
>> > someone turns the forest into a multipolygon, because there is some
>> > water or
>> > grassland inside of it. But the multipolygon is also used for the nature
>> > reserve. Which would imply the holes in the forest are unprotected, and
>> > that's usually not the case.
>>
>> I have been thinking about this as well. even the name belongs to the
>> outer way and not the relation.
>> I would say put the nature reserve tag on the outer way and the forest
>> tags on the relation. Would that work ?
>>
>> But tagging mistakes due to bad quality aerial imagery is equally
>> common I think. And those are much harder to detect I think.
>> For your case, it  is basically looking up all multipolygons on which
>> the nature reserve tags is placed and check those. While this can be a
>> long list, it is not nearly as long as checking all landuses.
>>
>>
>>
>> m
>>
>> ___
>> Talk-be mailing list
>> Talk-be@openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/talk-be
>
>
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Lionel Giard]

Personally, i put Nature reserve on a special relation, as it is described
on the wiki, like this one :
http://www.openstreetmap.org/relation/7130732#map=17/50.68519/4.70461
And the forest or other landuses are just part of this multipolygon. In
this example, i also have a multipolygon for the forest because, i have
things inside it.
Nothing stop you to have multiple relation at the same place (i think of
relation like a "special" polygon, and nothing stop us to make multiple
polygon on each other but slightly different in shape).

We should really describe/decide which tag is representing a landuse
(because it can be landuse=*, leisure=* or natural=*) and if we need to
avoid putting them at the same place (like allow or not leisure=park on top
of landuse=residential ?!). Right know, we all have our own opinion and it
create some variation of the map in Belgium.
And we should avoid gap in the landuse. Thus it seems important to solve
some problems like what should be the landuse tag under a road ?

2017-04-27 14:29 GMT+02:00 Marc Gemis :

> On Thu, Apr 27, 2017 at 1:50 PM, joost schouppe
>  wrote:
> > The examples you give are already hard work to think about. Much more
> basic
> > mistakes are made too: e.g. a forest is also a nature reserve. But then
> > someone turns the forest into a multipolygon, because there is some
> water or
> > grassland inside of it. But the multipolygon is also used for the nature
> > reserve. Which would imply the holes in the forest are unprotected, and
> > that's usually not the case.
>
> I have been thinking about this as well. even the name belongs to the
> outer way and not the relation.
> I would say put the nature reserve tag on the outer way and the forest
> tags on the relation. Would that work ?
>
> But tagging mistakes due to bad quality aerial imagery is equally
> common I think. And those are much harder to detect I think.
> For your case, it  is basically looking up all multipolygons on which
> the nature reserve tags is placed and check those. While this can be a
> long list, it is not nearly as long as checking all landuses.
>
>
>
> m
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

On Thu, Apr 27, 2017 at 1:50 PM, joost schouppe
 wrote:
> The examples you give are already hard work to think about. Much more basic
> mistakes are made too: e.g. a forest is also a nature reserve. But then
> someone turns the forest into a multipolygon, because there is some water or
> grassland inside of it. But the multipolygon is also used for the nature
> reserve. Which would imply the holes in the forest are unprotected, and
> that's usually not the case.

I have been thinking about this as well. even the name belongs to the
outer way and not the relation.
I would say put the nature reserve tag on the outer way and the forest
tags on the relation. Would that work ?

But tagging mistakes due to bad quality aerial imagery is equally
common I think. And those are much harder to detect I think.
For your case, it  is basically looking up all multipolygons on which
the nature reserve tags is placed and check those. While this can be a
long list, it is not nearly as long as checking all landuses.



m

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[joost schouppe]

2017-04-27 13:14 GMT+02:00 Marc Gemis :

> On Thu, Apr 27, 2017 at 10:03 AM, joost schouppe
>  wrote:
> > So then you need to decide if a park in a residential area is a park or a
> > residential area.
>
> aren't park and residential area 2 different "layers" ?
>
> Yes, they were. But my use case was statistical analysis, not any kind of
map. I needed to pick just one landuse indication for as much land as
possible. So first I made the layers, then cut away areas from the "lower"
layers if they were also defined in the "upper" layers. But as you say,
that is not always straightforward.
And yes, I know the OSMlanduse project, though I finished mine before they
published anything.

The examples you give are already hard work to think about. Much more basic
mistakes are made too: e.g. a forest is also a nature reserve. But then
someone turns the forest into a multipolygon, because there is some water
or grassland inside of it. But the multipolygon is also used for the nature
reserve. Which would imply the holes in the forest are unprotected, and
that's usually not the case.

-- 
Joost Schouppe
OpenStreetMap  |
Twitter  | LinkedIn
 | Meetup

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

On Thu, Apr 27, 2017 at 10:03 AM, joost schouppe
 wrote:
> So then you need to decide if a park in a residential area is a park or a
> residential area.

aren't park and residential area 2 different "layers" ?

I hope that one day we can map  a park as

landuse=recreation (or leisure)
leisure=park, perhaps with nested leisure=playground, pitch, garden in it.
landcover = grass / bushes/ trees/ flowerbeds / asphalt / water / 
for the different parts of the park.

and what about the parking of a park? Does it belong to the
leisure=park area, but perhaps not to the landuse=leisure ?
what about a cafe in the park ? landuse=retail around building and
terrace ? terrace as leisure=outdoor_seating ?



BTW, do you know: http://osmlanduse.org/  ? It gives a pie chart with
the percentages of the landuse in the current view.
And although it says landuse/landcover, they ignore the landcover tag I believe.

m.

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[joost schouppe]

Hi Julien,

* How would you feel about building a proposal about
forest_management_style=* ? To my great surprise, I noticed I kind of like
mingling in the endless discussions at the tagging mailing list. So I would
be willing to help out.

* Quantitative analysis of landuse mapping in Belgium: I did that. The idea
was to generate a useful dataset of landuse at the level of the statistical
sectors. So I made a classification of several layers (residential, nature,
water, transport infrastructure). Then I did some GIS processing: to create
polygons out of roads and POIs. Then I needed to choose in which layer to
count the landuse, for example sometimes there's a huge residential area
with everything just mapped on top of that. So then you need to decide if a
park in a residential area is a park or a residential area.
I can share method and result. We could set up a voice meeting on Riot (
https://riot.im/app/#/room/#osmbe:matrix.org) so others can join in case
they're interested.

* The end result should be a landuse convention, yes.

2017-04-27 9:51 GMT+02:00 Julien Minet :

> Thanks for your reactions!
>
> * About the tag natural=wood, I also think it is over-represented in
> Belgium. Belgian forests are indeed not only managed for timber production
> but also hunting, tourism and nature conservation, but often in an
> integrated manner under the same areas (at least in theory!). Note that the
> Natura 2000 program does not preclude at all that timber wood is produced!
> See on this link how much forests in south of Wallonia are covered by
> Natura 2000: http://geoportail.wallonie.be/walonmap#SHARE=
> 4E2203C158780AB5E053D0AFA49D7D23.
> 
>
> * I did not talk about the landcover tag in my article but it'd be worth
> talking more about it!
>
> * It would be interesting to quantitavely analyse the ways landuse is
> mapped in Belgium. I'm thinking about it. Any ideas on how, what kind of
> analysis are welcome...
>
> * Summarizing this discussion in osm.be could be nice. But why not also
> create a "landuse convention" page on the OSM wiki
> ?
>
>
> Julien
>
> On Wed, Apr 26, 2017 at 10:26 AM, joost schouppe  > wrote:
>
>> *  About forests, I tend to agree with the natural=wood not really
>> existing in Belgium. The only exception I know of is a bit of the
>> Zoniënwoud (Kersselaerplein) that has had "zero management" for 34 years
>> now.
>> But most natural=wood I've seen is wrong.
>>
>> Just recently, I changed the Bois de La Houssiere (
>> http://www.openstreetmap.org/#map=14/50.6189/4.1948) from wood to
>> forest. It's a bit of a special case: it's a Natura 2000 protected area,
>> but it is also actively used as a forestry area.
>>
>> I agree with the comments above that landuse=forest for any kind of group
>> of trees is annoying too.
>>
>> - In cases where you have residential areas in a forest, or wooded areas
>> in gardens, maybe we should really encourage the use of the landcover tag?
>> - In cases where forests are managed, but as some kind of nature reserve
>> or natural area, maybe we could use a subtag to indicate the management
>> style? That would allow to differentiate between real forestry and forests
>> with nature-friendly management. You could use one of the many nature
>> reserve tags of course, but I'm not sure all naturally managed forests are
>> protected and the Bois de la Houssiere shows the opposite also exists.
>>
>>
>> * On a more detailed note: I had never heard of the taxon tag, I've only
>> used species before. I'm completely confused now :)
>> And are the values REALLY comma separated, not " ; " seperated?
>>
>> * About OSM.be: we're still thinking about what exactly we want to use
>> the Projects for - the fact we don't really know was shown quite clearly by
>> Marc's latest article.
>>
>> I think we could have an OSM.be project on "Harmonizing tagging in
>> OpenStreetMap". It would first explain really short how tagging works, and
>> why it can be something problematic. Then it could define goals, one of
>> which could be "harmonizing tagging practices about landuse mapping in
>> Belgium". Next it could define a series of sub-projects, like "discussing
>> and creating consensus about best practices" (with links to this discussion
>> and Julien's article). Another one could be setting up a Maproulette task
>> to check certain suspicious cases (like the hundreds of natural=grassland
>> around the Bois de la Houssiere). Lastly, it should contain an invitation
>> and specific pointers on how to participate in the project.
>>
>> 2017-04-26 5:16 GMT+02:00 Marc Gemis :
>>
>>> Julien, and others,
>>>
>>> thanks a lot for this text. I still have to go through all the
>>> details, but here are already some remarks.
>>>
>>> - Me too, would love to see landuse=forest 

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[joost schouppe]

See, we really have no clue what we're doing :)
But maybe we should do this discussion in a separate thread


2017-04-26 12:18 GMT+02:00 Marc Gemis :

> On Wed, Apr 26, 2017 at 10:26 AM, joost schouppe
>  wrote:
> > * About OSM.be: we're still thinking about what exactly we want to use
> the
> > Projects for - the fact we don't really know was shown quite clearly by
> > Marc's latest article.
>
> which was not placed under Projects :-)
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>



-- 
Joost Schouppe
OpenStreetMap  |
Twitter  | LinkedIn
 | Meetup

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

On Wed, Apr 26, 2017 at 10:26 AM, joost schouppe
 wrote:
> * About OSM.be: we're still thinking about what exactly we want to use the
> Projects for - the fact we don't really know was shown quite clearly by
> Marc's latest article.

which was not placed under Projects :-)

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[joost schouppe]

*  About forests, I tend to agree with the natural=wood not really existing
in Belgium. The only exception I know of is a bit of the Zoniënwoud
(Kersselaerplein) that has had "zero management" for 34 years now.
But most natural=wood I've seen is wrong.

Just recently, I changed the Bois de La Houssiere (
http://www.openstreetmap.org/#map=14/50.6189/4.1948) from wood to forest.
It's a bit of a special case: it's a Natura 2000 protected area, but it is
also actively used as a forestry area.

I agree with the comments above that landuse=forest for any kind of group
of trees is annoying too.

- In cases where you have residential areas in a forest, or wooded areas in
gardens, maybe we should really encourage the use of the landcover tag?
- In cases where forests are managed, but as some kind of nature reserve or
natural area, maybe we could use a subtag to indicate the management style?
That would allow to differentiate between real forestry and forests with
nature-friendly management. You could use one of the many nature reserve
tags of course, but I'm not sure all naturally managed forests are
protected and the Bois de la Houssiere shows the opposite also exists.


* On a more detailed note: I had never heard of the taxon tag, I've only
used species before. I'm completely confused now :)
And are the values REALLY comma separated, not " ; " seperated?

* About OSM.be: we're still thinking about what exactly we want to use the
Projects for - the fact we don't really know was shown quite clearly by
Marc's latest article.

I think we could have an OSM.be project on "Harmonizing tagging in
OpenStreetMap". It would first explain really short how tagging works, and
why it can be something problematic. Then it could define goals, one of
which could be "harmonizing tagging practices about landuse mapping in
Belgium". Next it could define a series of sub-projects, like "discussing
and creating consensus about best practices" (with links to this discussion
and Julien's article). Another one could be setting up a Maproulette task
to check certain suspicious cases (like the hundreds of natural=grassland
around the Bois de la Houssiere). Lastly, it should contain an invitation
and specific pointers on how to participate in the project.

2017-04-26 5:16 GMT+02:00 Marc Gemis :

> Julien, and others,
>
> thanks a lot for this text. I still have to go through all the
> details, but here are already some remarks.
>
> - Me too, would love to see landuse=forest be used in a more strict
> way, only for areas where timber is really used for commercial
> purposes. It's even possible that at certain periods there are no
> trees in such areas. For the rest I would love that landcover=trees
> would be more accepted and rendered.
> One of the reasons is that landuse=forest clashes with e.g.
> landuse=residential in large private parks.  But I fear too any people
> stick to their "managed" definition and just want to see trees on the
> default map.
>
> - Some mapper split a landuse=farmyard and use landuse=residential
> around the farm itself. I do not do this. What do you think about this
> ?
>
> - During one of my recent walks I found some areas like
> https://xian.smugmug.com/OSM/OSM-2017/2017-04-02-Postel-AK/i-5D62hDt
> some were larger than what you see on this picture, I think I would
> use natural=grassland on those. Other suggestions ?   It was hard to
> take a better picture
>
> - Often it is better to use natural=tree_row instead of
> landuse=forest/natural=wood IMHO.
>
> regards
>
> m
>
> On Tue, Apr 25, 2017 at 9:31 AM, Julien Minet  wrote:
> > Hi list,
> >
> > Following some discussions about landuse=farmland|meadow some times ago
> in
> > this list, I've written an article here
> > (http://www.nobohan.be/2017/04/20/landuse-osm-belgium/) about land-use
> > mapping in Belgium: what could be the best practices adapted to the
> Belgian
> > landscape. Of course, there's matter for discussions about that topic ;-)
> >
> > I think this text could be used to make a page on
> > https://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/,
> since
> > it discuss what are the local conventions for land-use mapping in
> Belgium.
> >
> > Do you also want to put this text on osm.be, similarly to the Marc Gemis
> > articles? Maybe a better place for discussions...
> >
> > Cheers,
> >
> > Julien
> >
> > ___
> > Talk-be mailing list
> > Talk-be@openstreetmap.org
> > https://lists.openstreetmap.org/listinfo/talk-be
> >
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>



-- 
Joost Schouppe
OpenStreetMap  |
Twitter  | LinkedIn
 | Meetup

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Marc Gemis]

Julien, and others,

thanks a lot for this text. I still have to go through all the
details, but here are already some remarks.

- Me too, would love to see landuse=forest be used in a more strict
way, only for areas where timber is really used for commercial
purposes. It's even possible that at certain periods there are no
trees in such areas. For the rest I would love that landcover=trees
would be more accepted and rendered.
One of the reasons is that landuse=forest clashes with e.g.
landuse=residential in large private parks.  But I fear too any people
stick to their "managed" definition and just want to see trees on the
default map.

- Some mapper split a landuse=farmyard and use landuse=residential
around the farm itself. I do not do this. What do you think about this
?

- During one of my recent walks I found some areas like
https://xian.smugmug.com/OSM/OSM-2017/2017-04-02-Postel-AK/i-5D62hDt
some were larger than what you see on this picture, I think I would
use natural=grassland on those. Other suggestions ?   It was hard to
take a better picture

- Often it is better to use natural=tree_row instead of
landuse=forest/natural=wood IMHO.

regards

m

On Tue, Apr 25, 2017 at 9:31 AM, Julien Minet  wrote:
> Hi list,
>
> Following some discussions about landuse=farmland|meadow some times ago in
> this list, I've written an article here
> (http://www.nobohan.be/2017/04/20/landuse-osm-belgium/) about land-use
> mapping in Belgium: what could be the best practices adapted to the Belgian
> landscape. Of course, there's matter for discussions about that topic ;-)
>
> I think this text could be used to make a page on
> https://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/, since
> it discuss what are the local conventions for land-use mapping in Belgium.
>
> Do you also want to put this text on osm.be, similarly to the Marc Gemis
> articles? Maybe a better place for discussions...
>
> Cheers,
>
> Julien
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Lionel Giard]

Hi,

Concerning the wood-tag : I was personally using the first approach Here
 where it says that
natural=wood is for area covered by tree, and natural=forest is for area
managed for land forestry (like plantation in the Ardennes). And most of
the time, the only "wood" areas are the place around motorway intersections
for example (between the roads, you often have an area covered by tree,
without real management).

I agree with you about fenced area that correspond to pasture/meadow. But I
would also include, based on local knowledge only, some non-fenced area as
landuse=meadow when it is an area for cutting only (to feed cattle for
example) like it is often the case in some part of Wallonia. It's sometimes
area that are unsuitable for agriculture, but used just to feed cattle
without letting them graze on it (like when it is too dangerous).
And for any uncultivated area, composed of mostly grasses and herbaceous,
there always is the natural=grassland
 tag.

Every other cultivated area, should then only be orchard or farmland. The
only problem can be when a combination of different types of agriculture
occur at the same place, then i don't know how to map.

Lionel

Garanti
sans virus. www.avast.com

<#m_-9078182235892435705_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

2017-04-25 11:40 GMT+02:00 Gerard Vanderveken :

> Hi,
>
> One remark for the wood-tag:
> The tag natural=wood
>  is sometimes
> used for forests, but since natural=wood would refers to unmanaged, natural
> forests, IMHO it should not be used in Belgium since no more forests are
> completely natural: every patches of forest in Belgium has experienced
> human interventions in the recent history.
> In several forests Zoniënwoud, Meerdaalwoud, ... are some nature reserves.
> Places where nature can have its course and where only very minimal human
> intervention take place (eg clearance of paths)
> I think these reserves (Joseph Zwaenepoel, Kerselaerplein, Everzwijnbad,
> Pruikemakers) could qualify as wood.
>
> I think meadow is a good tag for the grass lands. Conversion between
> farmland and grass land takes somtetimes place but mosttimes as incidental
> crop harvest of grass.
> Making a pasture requires also fencing, which is not a temporary measure.
>
> So, when there is a grass land without fence surrounded by farmland, it is
> likely farmland.
> Other grass lands are usually fenced and are meadow.
> Local knowledge will tell if it is to be one or the other
>
> Regards,
> Gerard.
>
> Julien Minet wrote:
>
> Hi list,
>
> Following some discussions about landuse=farmland|meadow some times ago in
> this list, I've written an article here (http://www.nobohan.be/2017/
> 04/20/landuse-osm-belgium/) about land-use mapping in Belgium: what could
> be the best practices adapted to the Belgian landscape. Of course, there's
> matter for discussions about that topic ;-)
>
> I think this text could be used to make a page on
> https://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/,
> since it discuss what are the local conventions for land-use mapping in
> Belgium.
>
> Do you also want to put this text on osm.be, similarly to the Marc Gemis
> articles? Maybe a better place for discussions...
>
> Cheers,
>
> Julien
>
> --
>
> ___
> Talk-be mailing 
> listTalk-be@openstreetmap.orghttps://lists.openstreetmap.org/listinfo/talk-be
>
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-be
>
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Land-use mapping with OSM in Belgium

[Gerard Vanderveken]

Hi,

One remark for the wood-tag:
The tag natural=wood 
 is sometimes 
used for forests, but since natural=wood would refers to unmanaged, 
natural forests, IMHO it should not be used in Belgium since no more 
forests are completely natural: every patches of forest in Belgium has 
experienced human interventions in the recent history.

In several forests Zoniënwoud, Meerdaalwoud, ... are some nature reserves.
Places where nature can have its course and where only very minimal 
human intervention take place (eg clearance of paths)
I think these reserves (Joseph Zwaenepoel, Kerselaerplein, Everzwijnbad, 
Pruikemakers) could qualify as wood.


I think meadow is a good tag for the grass lands. Conversion between 
farmland and grass land takes somtetimes place but mosttimes as 
incidental crop harvest of grass.

Making a pasture requires also fencing, which is not a temporary measure.

So, when there is a grass land without fence surrounded by farmland, it 
is likely farmland.

Other grass lands are usually fenced and are meadow.
Local knowledge will tell if it is to be one or the other

Regards,
Gerard.

Julien Minet wrote:


Hi list,

Following some discussions about landuse=farmland|meadow some times 
ago in this list, I've written an article here 
(http://www.nobohan.be/2017/04/20/landuse-osm-belgium/) about land-use 
mapping in Belgium: what could be the best practices adapted to the 
Belgian landscape. Of course, there's matter for discussions about 
that topic ;-)


I think this text could be used to make a page on 
https://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/, 
since it discuss what are the local conventions for land-use mapping 
in Belgium.


Do you also want to put this text on osm.be , similarly 
to the Marc Gemis articles? Maybe a better place for discussions...


Cheers,

Julien



___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be
 

___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be

[OSM-talk-be] Land-use mapping with OSM in Belgium

[Julien Minet]

Hi list,

Following some discussions about landuse=farmland|meadow some times ago in
this list, I've written an article here (
http://www.nobohan.be/2017/04/20/landuse-osm-belgium/) about land-use
mapping in Belgium: what could be the best practices adapted to the Belgian
landscape. Of course, there's matter for discussions about that topic ;-)

I think this text could be used to make a page on
https://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Conventions/, since
it discuss what are the local conventions for land-use mapping in Belgium.

Do you also want to put this text on osm.be, similarly to the Marc Gemis
articles? Maybe a better place for discussions...

Cheers,

Julien
___
Talk-be mailing list
Talk-be@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-be