[Geotiff] geos projection sweep parameter

[Even Rouault]

On lundi 10 juillet 2017 10:10:48 CEST David Hoese wrote:
> Even,
> 
> So to summarize, a typical geotiff reading client should look for the
> standard geotiff projection information, if not found look for "ESRI PE
> String".

Yes. But ESRI PE string is vendor specific, so a strict GeoTIFF reader will generally not understand it.

> You made it sound like ArcGIS uses its own form of WKT?

WKT v1 was not strongly standardized, so various vendors have come with their own variations,
to decide how to name projection methods, parameters, etc...
Some preview :
https://web.archive.org/web/20130728081442/http://home.gdal.org:80/projects/opengis/wktproblems.html

> Does it also use the "ESRI PE String" key?

I guess so ! I don't recall the details, but I believe that support for GeoTIFF ESRI PE string in GDAL
probably originates from merging this part of the ESRI fork of GDAL to GDAL mainline.

> 
> As you stated the two "possible" solutions are to get the "geos"
> projection (and others) added to the geotiff specification or to get
> "sweep" added to the ESRI WKT specification. Both of which would require
> client libraries to be updated to properly parse and use this
> information which could take years to reach a majority of the GIS
> community; maybe less given the right documentation and google search
> results.

For GDAL-based software, potentially no change on the reading side would be required.
If the ESRI PE string contains the EXTENSION PROJ4 node (which is rather hacky), GDAL will use it.
I've just tested it by manually crafting such a GeoTIFF. Works at least with GDAL 1.11  or later.

> 
> Perhaps I will start with contacting some of the working group and see
> what path forward they might recommend.

I actually forwarded your email to one of the members of the OGC GeoTIFF SWG

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/geotiff/attachments/20170710/a6f724b8/attachment.html>