expressing uncertainty. Using meters appears to be the de-facto best
practice for expressing uncertainty of geographic info - I would guess for
the same reasons that Alexander as provided: simplicity. And working with
decimal degrees as the UoM is just way too hard for many applications.
As to expressing other CRSs, Ivan does raise some good points. However,
this issue could be resolved by a very clear statement in the document
that clarifies the use cases that this geo URI is used for.
Carl
>> My point is that the geo: URIs may actually be useful in
>> metadata entries. When the data these entries describe uses
>> non-WGS 84 CRS (which is true for most of the raster data sets
>> derived from satellite observations; check Landsat 7 data on
>> http://www.landcover.org/, for instance, or you may refer to the
>> vast variety of data sets available through
>> https://wist.echo.nasa.gov/), it seems to be inappropriate to
>> use WGS 84. Especially when both the sender and receiver of the
>> geo: URI use non-WGS 84 CRS internally (consider, e. g.,
>> MapServer showing Landsat 7 data.)
>
> As you outlined below, the "professional" GIS world has multiple options
> available when conveying spatial information in a variety of CRSes. The
> "geo" URI specification aims at filling the gap that there's no simple,
> easy to use and human readable format that's well standardized.
>
> Simplicity is a key feature for me in the scheme. If you want
> complexity, there are enough other schemes to go with.
>
>> There, of course, exist other ways to express such information
>> (like GML, for instance) while preserving the CRS, which,
>> however, seem to me somewhat cumbersome, when compared to a URN
>> with one more URN embedded within it.
>>
>> Also note that when transforming from a given CRS to WGS 84, the
>> transformation of the associated uncertainty may be non-trivial.
>
> Correct. That's why i am in favour in *always* expressing uncertainty in
> meters - which would
>
> A) not require any transformations on the uncertainty value
> B) make the uncertainty parameter independent from the crs parameter
> C) make uncertainty understandable and measureable to non-GIS people too
> (my mum understands +/+100 meters, but not +/-0.03 degrees)
>
>> Having said that, I believe that the specification, if allowing
>> the use of non-WGS 84 CRSes, should warn about possible
>> interoperability problems and generally recommend /against/ such
>> CRSes, except for the special cases (like, e. g., passing the
>> locations between two agents that already use non-WGS 84 CRS
>> internally.)
>>
>> > Regarding uncertainty
>
> I would leave that to the specifications which define the additional
> CRSes.
>
>
> Alex
> _______________________________________________
> Geopriv mailing list
> Geopriv@ietf.org
> https://www.ietf.org/mailman/listinfo/geopriv
>
_______________________________________________
Geopriv mailing list
Geopriv@ietf.org
https://www.ietf.org/mailman/listinfo/geopriv
