[Geojson] [OSGeo-Standards] EPSG + Coordinate Ordering (Was Re:GeoJSON '1.0'?)

Christopher Schmidt crschmidt at metacarta.com
Fri Mar 14 04:37:56 PDT 2008

On Thu, Mar 13, 2008 at 05:50:27PM -0600, Carl Reed wrote:
> Not to be a bother, but Please wait until I send you the axis order 
> guidance document the OGC has been working hard on. You may find the 
> document useful and we would encourage your feedback.

After waiting more than 4 months for the current discussion paper on WMS
Tiling to be made available on the OGC website, I'm not particularly
convinced that waiting on the OGC to provide guidance is in the
best-interest of a rapid-development specification design process.  

> Axis order is a non-trivial element of any geo application and the 
> implications of having different approaches to dealing with CRS and axis 
> order in different communities will be deadly in terms of consistency, data 
> sharing, and interoperability in general.

That's fine if you expect to be able to use GeoJSON without translation to 
some data format that is designed to encode that information. GeoJSON is
not designed for that. The primary use case of GeoJSON is for the
*simple* cases -- and no matter what the simple case looks like, it will
be trivial (so far as I can tell) for a user to map that into data that
makes sense in their client if they have a knowledge of CRS, because the
definition of coordinate reference systems includes the information
needed to reverse it.

> And just to make things interesting, remember that a CRS does not have to 
> be related to the earth :-)

How is that relevant? We don't describe things as relating to the earth?
Only to X,Y,Z in a projected coordinate system. How they map to the
earth is the job of the CRS, not of GeoJSON.

> Also, read  http://www.ogcnetwork.net/node/338 which provides excellent 
> grounding as to what happens to the same "point" on the surface of the 
> earth when different datums are used. One more reason using full CRS 
> metadata such as defined in the EPSG database is so important.

Absolutely. Though I'm not convinced that that drawing is realistic: I'd
be interested in seeing the source data (vector + imagery) that that
applies to. But we're not talking about ignoring the metadata defined in
the EPSG database: we're talking about how to *map it to GeoJSON* --
specifically, how you take a CRS, read GeoJSON, and get out the data in
the coordinate ordering you care about.

> Oh yeah, I forgot - (x,y,z) in South Africa is typically defined using 
> reverse Lo-Gauss, which means that they use an upside down coordinate 
> system (at least from us Northerners perspective). In other words, y "is 
> directed south and not north.

Sure. And Y can also be Westing, as in
but that doesn't mean that there *isn't* a y. I don't care if z means
"into the earth": Just that it is labeled Z. Application understanding
of a particular esoteric CRS is required in order to do fancy things
with it, but if you only care about working in projected space, then
none of this matters.  

Christopher Schmidt

More information about the GeoJSON mailing list