[postgis-users] [postgis-devel] Sphere or Spheroid

[pcreso at pcreso.com]

Some questions:

Which spheroid is used by default?  (I commonly use 3 different ones)

Does the geography datatype require a SRID (or spheroid?) to be specified, and reject a null one? (I'm not familiar with the geography implementation details)

If there is no default spheroid available does ST_Distance() then fail unless it is explicitly told to use spherical?

As I see it, the issue is less about providing the right (or best) answer by default, & more about protecting users from their ignorance.

I'm a bit cautious about making assumptions about which spheroid should be applied by default for users who do not fully understand the issues.

As long as the issue is well covered in the documentation, and both spherical & spheroidal versions are available, I don't have a strong preference either way.


Cheers,

 Brent


--- On Sat, 10/31/09, Paul Ramsey <pramsey at cleverelephant.ca> wrote:

> From: Paul Ramsey <pramsey at cleverelephant.ca>
> Subject: Re: [postgis-users] [postgis-devel] Sphere or Spheroid
> To: "PostGIS Users Discussion" <postgis-users at postgis.refractions.net>
> Date: Saturday, October 31, 2009, 1:28 PM
> My feeling as well, particularly when
> comparing the spheroid distance
> to the sphere distance to the distance on a properly
> projected plane.
> The spheroid and projected plane distances are very very
> close, the
> sphere is as much as 0.5% off them, depending on how long a
> distance
> you measure.
> 
> P.
> 
> On Fri, Oct 30, 2009 at 5:17 PM, Richard Greenwood
> <richard.greenwood at gmail.com>
> wrote:
> > On Fri, Oct 30, 2009 at 5:48 PM, Paragon Corporation
> <lr at pcorp.us>
> wrote:
> >>
> >> The choice of default sphere vs. spheroid is
> probably something we should
> >> ask the general PostGIS users group so cc'ing them
> since more people
> >> frequent that list that would have strong opinions
> on this subject.
> >>
> >> So question folks -- is which do you prefer as
> default behavior for new
> >> geography data type
> >>
> >> faster sphere calculation (with 0.05% - 1% margin
> error) (I vaguely recall
> >> that's what my studies suggested around the globe
> compared to UTM)
> >>
> >> or
> >> slower more accurate spheroid calculation
> >
> > I think that the default should be to "do it right"
> and that "doing it
> > fast" should be optional.
> >
> > --
> > Richard Greenwood
> > richard.greenwood at gmail.com
> > www.greenwoodmap.com
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at postgis.refractions.net
> > http://postgis.refractions.net/mailman/listinfo/postgis-users
> >
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users
>