Re: [OSM-talk-ie] Ireland EEZ boundary

[moltonel 3x Combo]

On 22/09/2015, Stephen Roulston  wrote:
> Yes, sorry about that :(
>
> I automatically thought that such a short list is bound to be wrong. I was
> very embarrassed to see the posting later.

No worries, it was quick to redo and it showcased the problems that
the old EEZ caused :)

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

Re: [OSM-talk-ie] Ireland EEZ boundary

[moltonel 3x Combo]

On 20/09/2015, moltonel 3x Combo  wrote:
> Done:
> https://www.openstreetmap.org/changeset/34148691

Reuploaded in http://www.openstreetmap.org/changeset/34157957, as
Stephen mistakenly undid the changes when he encountered an upload
conflic while working on townlands. Precisely the kind of problems
that should occur less often with the simplified EEZ :p

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

Re: [OSM-talk-ie] Ireland EEZ boundary

[Stephen Roulston]

Yes, sorry about that :( 

I automatically thought that such a short list is bound to be wrong. I was very 
embarrassed to see the posting later. 

Stephen

> On 21 Sep 2015, at 11:26, moltonel 3x Combo  wrote:
> 
>> On 20/09/2015, moltonel 3x Combo  wrote:
>> Done:
>> https://www.openstreetmap.org/changeset/34148691
> 
> Reuploaded in http://www.openstreetmap.org/changeset/34157957, as
> Stephen mistakenly undid the changes when he encountered an upload
> conflic while working on townlands. Precisely the kind of problems
> that should occur less often with the simplified EEZ :p
> 
> ___
> Talk-ie mailing list
> Talk-ie@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ie

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

Re: [OSM-talk-ie] Ireland EEZ boundary

[moltonel 3x Combo]

On 18/09/2015, Colm Moore  wrote:
> You will probably find that there is a statutory instrument that defines
> much of the EEZ boundaries.

Yes, the corresponding document and points can be found at
http://www.irishstatutebook.ie/eli/2014/si/86 and its title is tagged
on the osm objects. I have no intention of touching these coordinates.
But they only describe the boundary at sea. Nothing is said about the
coastline/inland boundary, which is the one I'm addressing in my
email.

> I think islands should be included within the EEZ - its all the one
> territorial claim - as islands change in size.

That would fit with suggestion b) or c).

> What is the OSM standard for this?

http://overpass-turbo.eu/?w=%22border_type%22%3D%22eez%22+global is
interesting. All kinds of layouts exist:
* border-only (no area) boundaries (eg between Japan and Korea)
* area encompassing the land (eg Philipines)
* area following the offshore country border rather than the exact
coastline (eg South Africa)
* area crossing the land very roughly (eg Italy)
* the Irish EEZ seems unique in OSM in following the coastline precisely

Following these observations, and the fact that the irish statute
document's points actually start at sea very near the territorial
water boundary, I suggest a new option : make the EEZ follow its
current offshore boundary, and the territorial boundary (which starts
at http://www.openstreetmap.org/way/144387894 and ends at
http://www.openstreetmap.org/way/92987647). This follows the South
Africa model, and actually results in a very lightweight and
standard-conforming osm object.


Thanks for pushing me to do these extra checks Colm, this new solution
looks much more satisfying to me. Unless there's a contrary opinion
I'm going to enact it this sunday (schedule permiting), but a couple
of "+1" replies wouldn't hurt either.

Cheers.

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

[OSM-talk-ie] Ireland EEZ boundary

[moltonel 3x Combo]

Hi,

the EEZ boundary is a maritime exclusive economic zone. It is mapped
as relation 4121287 in OSM:
http://localhost:8111/import?url=http://www.openstreetmap.org/api/0.6/relation/4121287/full
(using a josm/mercator remote-control link instead of a website link
because the website is likely to timeout).

There are a few problems with it:
* At ~4000 members, it is enormous. JOSM validator takes over a minute
to check it on my machine. It's the biggest relation we have, even the
four regions are about half the size.
* It only has a fraction of the islands it ought to have, and most of
those are mistakenly tagged as outer instead of inner.
* It keeps breaking (non-closed ways), and the more members a MP has
the more laborious it is to fix it.

What should we do with it ? Either...

a) Add all the missing islands, fix their roles
  pros: that's the best theoretical way to map a maritime area
  cons: will exacerbate all the size-related problems

b) Keep as-is but drop all the islands
  pros: reduces the size strain a bit
  cons: not a true maritime area anymore. Boat navigation will surely
not get fooled by this, but other tasks like for example area
calculation gets more complicated.

c) Same as b) but follow the ROI-NI border instead of the coastline
  Same pros and cons as b), only stronger

d) Only keep the maritime ways
  pros: simplest version
  cons: not much can be done with it appart from rendering (for
example a bot can't determine wether it is navigating within the EEZ
anymore)

e) Use multilinestring relations to simplify the geometries
  pros: great reduction in member count. Same technique could be used
to simplify the other county and region MPs
  cons: not all tools support multilinestring relations (what about
nominatim ?), they are used in France for example but are not yet a
standard osm feature. Islands still need to be supported, as in a).



I'm leaning towards solution c), but could be convinced of either.
What do ye think ?

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