[Geojson] circles & units

[Tim Schaub]

I don't think we disagree.

The proposal I linked to says this:

    A properties value of "axis_units" is optional and may be used to
    specify units. A default of kilometers is otherwise assumed.

https://github.com/GeoJSONWG/geojson-spec/wiki/Proposal---Circles-and-Ellipses-Geoms

There has been discussion about it being silly to have center
specified in degrees and radius also in degrees.  Without this silly
default (same units), clients can't really do much with circles
(unless they know the meters per unit of all CRS they might
encounter).  Which is silly.

I think you agree with this.

Tim

On Fri, Nov 4, 2011 at 6:04 PM, Matt Priour <mpriour at kestrelcomputer.com> wrote:
> I disagree.
> In the absence of a stated CRS, GeoJSON spec states:
> "If no crs member can be so acquired, the default CRS shall apply to the
> GeoJSON object.
>
> The default CRS is a geographic coordinate reference system, using the WGS84
> datum, and with longitude and latitude units of decimal degrees."
>
> Thus, in absence of any additional information, AND IF the circles spec
> stated that radius MUST be stated in CRS coordinate units (as has been
> suggested), then the following information could be determined:
> All units are in decimal degrees of latitude & longitude in the WGS 84
> datum.
> circle:
>  center: 0.0 deg longitude, 0.0 deg latitude
>  radius: 1.0 degrees
>
> point:
>   x: 0.5 deg longitude
>   y: 0.0 deg latitude
>
> Most geometric operators would probably convert to a reasonable geometric
> (rather than geographic) coordinate system before doing Point in Polygon or
> Area calculations, but that is beyond the concern of GeoJSON.
>
> As long as the requirement is "radius MUST be stated in CRS coordinate
> units" then all the required information is known.
> However, if the requirement is "radius MUST be assumed to be meters unless a
> specific measurement unit is indicated", then you are right. All kinds of
> information is unknown and requires outside information to know even the
> basic positional information.
>
> Matt Priour
>
> --------------------------------------------------------------------------------------------
>
> From: Tim Schaub
> Sent: Friday, November 04, 2011 4:25 PM
> To: geojson at lists.geojson.org
> Subject: [Geojson] circles & units
>
> I'll try not to combine or confuse concerns here.  This thread is
> strictly about my concern with the circles proposal with regard to
> units.
>
> https://github.com/GeoJSONWG/geojson-spec/wiki/Proposal---Circles-and-Ellipses-Geoms
>
> An otherwise lightweight client cannot do a whole lot with circles.
> Take, for example, these two objects:
>
>   [{
>       "type": "Circle",
>       "coordinates": [0, 0],
>       "radius": 1
>   }, {
>       "type": "Point",
>       "coordinates": [0.5, 0]
>   }]
>
> Without knowing the number of meters per unit for the geometries' CRS,
> a client cannot answer the very simple question: is the point in the
> circle?  Lacking this same information, a client cannot even render
> the two objects.
>
> This makes circles unique among GeoJSON objects.  Without the EPSG
> database, a service that returns CRS details*, or some other
> dependency, circles cannot be used together with other
> features/geometries any interesting way.
>
> Tim
>
> * No handy service that I know of returns the number of meters per
> linear unit for a given CRS.  Though this is critical for so many
> things, specs like WMS & WMTS assume everybody knows this, and it
> drives me nuts.
>
> --
> Tim Schaub
> OpenGeo http://opengeo.org/
> Expert service straight from the developers.
> _______________________________________________
> Geojson mailing list
> Geojson at lists.geojson.org
> http://lists.geojson.org/listinfo.cgi/geojson-geojson.org
>
>



-- 
Tim Schaub
OpenGeo http://opengeo.org/
Expert service straight from the developers.