Re: [Geopriv] [geopriv] #14: uncertainty specification

> I think that uncertainty should be speicified in meters, independent
> from the CRS, and the members of the WG that i have spoken to seem to
> agree.

I agree too.

With uncertainty expressed in degrees one has to understand whether that means:

1.  (lat+/-u', lon+/-u')

    i.e. the region bounded by (lat-u -> lat+u, lon-u -> lon+u).  The shape of
    and area of this region depends on where you are on the globe, and you also
    get large problems with wrap-around at the poles.

  or

2.  (lat, lon) +/- u'

    i.e. anywhere within the 3d-chord specified with center (lat, lon) subtending
    u degrees.  This is a circular region that has almost the same shape and area
    everywhere (WGS84 ellipsoid notwithstanding).

It's slightly tricky - personally if someone said " (51'N, 1'W) +/- 1' " I'd probably assume definition 1 above, and take that to mean "50 - 52 N, 2 - 0 W".

However IMHO definition 2 is more intuitive.  It's also much more consistent with specifying the uncertainty in meters.

Ray