>> Any particular reason not to allow for an uncertainty information to
>> be specified in the URI?
> I am definitely not the expert who had addressed that for previous
> protocols, but i remember there was always a heated discussion (3825
> comes to my mind).
ACK.
> First, there's a difference between "uncertainty" and "object
> dimension" (uncertainty is how "unsure" the measurement is, while
> what you describe is more "what is the approx. Size of the location
> to be represented).
To my mind, these are not as distinct as it may seem.
For example, I could post a geo: URI to describe my own position
for, say, today. I'm certainly to go visit a few places during
the day (at least home and work), so there's uncertainty
associated with my position, and it comes not from the
instrument's inability to perform the precise measurement, but
also from my own ``inability'' to stay precisely in one place.
In other words, the uncertainty always gives the interpreting
party the area (for 2D) or volume (for 3D) where the object is
most likely (at 95% confidence) to be found. The respective
extents depend, and could be estimated from: the precision of
the instrument, the characteristic size of the object which
location is being measured, its characteristic movement
distance, the intended use of the information being published,
or, I believe, most commonly, taking into account all of these.
The above being said, I don't think that a geo: URI may need to
allow for a detailed specification of all the sources of the
uncertainty, or why they were assumed, or how they were
estimated or measured. However, the simply ``uncertainty as
delta'' specification is, I guess, will be handy.
One more example: in different contexts, either of the following
may be used to refer to the city of Barnaul:
* geo:53,83;
* geo:53.4,83.8;
* geo:53.36,83.75.
While these aren't ``equal'' per the geo: scheme specification,
the addition of the uncertainty information could allow for
these to be related to each other.
Also, it should be noted, that for a physicist, the following
URIs look quite /distinct/:
* geo:53.4,83.8;
* geo:53.40,83.80.
While these are ``equal'' per the specification. Don't know if
it's of any significance, though. It may, however, be useful to
note in the specification that the interpretation of the geo:
URIs is somewhat contrary to the expectations a physicist may
have.
> Then, what is the "dimension" of those values? Is is a circle, or a
> sphere? In "natural" space, or in the "projected" space (lat/lon
> circle, which would definitely not give a circle in "real world"?)
> What is the dimension (meters, inches, degrees?).
In my opinion, it should always refer to the ``projected''
space, and use the dimensions used for that projected space.
For EPSG 4979, for instance, it would be degrees for Y and X,
and meters for the altitude.
> If you want, we can set up a quick chat here in Stockholm about it
? Could you please be more specific?
It'd certainly be hard for me to visit Stockholm in the
foreseeable future.
> - and i would love to have someone who was involved in other
> "uncertainty" discussions joining...
--
FSF associate member #7257
_______________________________________________
Geopriv mailing list
Geopriv@ietf.org
https://www.ietf.org/mailman/listinfo/geopriv
