Hi Jon,
Thanks for the link. Definitely clear as mud.
I'm still not seeing any text in ISO 19111 that spells out that you
can't combine a geographic CRS and an ellipsoid based vertical CRS. Just
that quote that ellipsoidal heights can't be captured in a vertical CRS
because vertical CRS are based on the direction of gravity and
ellipsoids are not. Which seems OK except when I see that some vertical
datums are based on ellipsoids. Don't those two things contradict each
other?
That you have to use a full 3D CRS when doing certain coordinate
transformations also makes sense to me. But isn't that the case for most
3D CRS since the vertical may not (is not in most cases) orthogonal to
the horizontal?
Is the restriction on 3D geographic CRS, that they can not be compound
CRS, simply to make sure the vertical and horizontal use the same base
ellipsoid? Which would seem to mean that they are the only 3D CRS that
are defined with orthogonal horizontal and vertical.
OK. Enough mud scraping and stirring for now.
Thanks,
Ethan
Jon Blower wrote:
> Hi Ethan,
>
> I'm afraid I've never actually read ISO 19111 (I don't read many
> standards, I let other people tell me about them then I pretend I know
> what I'm talking about ;-)) so I can't point you to the information.
> I can however suggest that you look at a recent conversation I had on
> this subject with the (very helpful) GeoTools guys on the GeoTools
> mailing list where this was discussed in more detail.
>
> (ISO 19111 distinguishes between "compound CRSs", which are 2D
> horizontal + 1D vertical CRSs, and "true" 3D CRSs, which are combined.
> If your horizontal CRS is geographic and your vertical CRS is "height
> above ellipsoid" then ISO19111 doesn't allow you to separate them,
> although GeoTools actually relaxes this for practical purposes. I
> can't say I understand this fully yet.)
>
> The conversation is here
> http://www.nabble.com/Vertical-and-3D-coordinate-reference-systems-td18859312.html,
> but I think the main issue is that you can't separate the horizontal
> and vertical components if you're doing a coordinate transform that
> involves a datum shift (e.g. WGS84 lat-lon to British National Grid).
>
> Clear as mud?
> Jon
>
>
> On Thu, Aug 7, 2008 at 9:58 PM, Ethan Davis <edavis at unidata.ucar.edu> wrote:
>
>> Hi Jon,
>>
>> Jon Blower wrote:
>>
>>> Hi Philip,
>>>
>>>
>>>> This situation is especially pertinent to the complex realm of coordinate
>>>> reference systems, where it is difficult to focus on one particular facet
>>>> -
>>>> vertical coordinate systems, say - without also having to consider other
>>>> facets - horizontal coordinate systems, geodetic datums, ellipsoids,
>>>> projections, and so on. It's an all-or-nothing kind of problem domain.
>>>>
>>>>
>>> This is a very good point - in fact ISO 19111 says that you can't
>>> separate horizontal and vertical CRSs in some cases. I'm wary of
>>> making such proposals myself because I frankly don't understand all
>>> the issues properly (we need some geodesists or GIS experts or
>>> something). So it's a tricky one.
>>>
>>>
>> Can you point out where in ISO 19111 it discusses the relation between
>> horizontal and vertical CRSs?
>>
>> I see the various subtypes of CRS: geodetic, vertical, engineering, and
>> image (which all are based on some kind of datum); as well as some more
>> complex CRS: derived, projected, and compound (which are all based on one or
>> more of the simpler types). To make a 3D CRS you combine a horizontal CRS
>> and a vertical CRS with a compound CRS. My impression has always been that
>> vertical CRSs are very separate from horizontal CRSs.
>>
>> I did find in Annex B 1.2.1 a discussion of the relation between vertical
>> datum and ellipsoidal height:
>>
>> "Vertical CRSs make use of the direction of gravity to define the
>> concept of height or depth. By implication therefore, ellipsoidal
>> heights (h) cannot be captured in a vertical CRS: ellipsoidal
>> heights cannot exist independently, but only as an inseparable part
>> of the 3D coordinate tuple defined in a geodetic 3D CRS."
>>
>> I'm not sure I understand since I thought a vertical datum could be based on
>> an ellipsoidal earth. (Though I only see two based on an ellipsoid out of
>> over 100 vertical datums defined by EPSG. Maybe they aren't that useful.)
>>
>> I'm with you, I really don't have a good grasp of all the issues.
>>
>> Is there a geodesist in the house?!
>>
>> Ethan
>>
>> --
>> Ethan R. Davis Telephone: (303) 497-8155
>> Software Engineer Fax: (303) 497-8690
>> UCAR Unidata Program Center E-mail: edavis at ucar.edu
>> P.O. Box 3000
>> Boulder, CO 80307-3000 http://www.unidata.ucar.edu/
>> ---------------------------------------------------------------------------
>>
>>
>>
>
>
>
>
--
Ethan R. Davis Telephone: (303) 497-8155
Software Engineer Fax: (303) 497-8690
UCAR Unidata Program Center E-mail: edavis at ucar.edu
P.O. Box 3000
Boulder, CO 80307-3000 http://www.unidata.ucar.edu/
---------------------------------------------------------------------------
Received on Fri Aug 08 2008 - 14:24:40 BST
