[postgis-devel] closed surfaces, volum or just closed surface

[Olivier Courtin]

On Aug 17, 2010, at 1:35 PM, Nicklas Avén wrote:

Nicklas,

> in #568 you write:
> "As a consequence closed surface (so volume) have a 0 perimeter  
> value return"
>
> Is it so. Isn't it nessecary to be able to describe a closed survace  
> without handling it as a volum or polyhedron?
> I mean like a closed linestring. Sometimes you just want a closed  
> linestring not a polygon.
>
> I thought this was to opff topic in #568 so I put it in a separate  
> post.
>
> Does the spec say anything about this?

Well in fact in ISO 19107 we have all the kind of geometries we could
dream in GIS field fully describded.
And so we've got points, curves, surfaces, solids, conic and so on...

Right now this full model is really ambitous.
(This model is also used in GML 3, and so in CityGML)

Current spatial databases standard only take a part of it.
(Even if we have strong clues that the next one will take in count
the full ISO 19107 model.)

And in fact PolyhedralSurface or TIN are initialy surfaces.
So by analogy they behave more like your linestring (who could be  
closed).

And still by analogy your polygon equivalent in 3D is a solid
in ISO 19107, but there's still no database standard to describe it.

So the best we have right now (in spatial database standards) is 3D  
closed surface...


>  The difference shows for example when we measure distance from a  
> closed survace/volum
> to a point inside it. If it is a volum the distance is 0 if it is  
> just a closed surface there is a distance
> to calculate to the closest facet.

I'm agree with you that's a really good idea to have
a flag to indicate on the geometry if it's closed or not (volume or  
surface)

I've just began to think on a topology serialization,
need to work on it far more...


--
Olivier