Support EPSG 3857

[jmalik]

Datum for EPSG 3857 is not handled properly. It uses the WGS384 ellipsoid, but should not.

See ​http://hg.orfeo-toolbox.org/OTB/rev/69e1541eddab for a fix. Might need an update in ogcwkttranslator too (I'll come back on this after syncing ossimplugins with OTB version).

Issue is discussed in more details here : ​http://hg.orfeo-toolbox.org/OTB/rev/8b064b7ae526

[jmalik]

Sorry, issue has been discussed on the OTB tracker : ​http://bugs.orfeo-toolbox.org/view.php?id=351

[jmalik]

see issue #2269 for the associated patch in ossimOgcWktTranslator

[jmalik]

The following test is run in OTB : ​http://hg.orfeo-toolbox.org/OTB/file/004c0194e3e9/Testing/Code/Projections/CMakeLists.txt#l1473

It reprojects the geo point {1.48353 43.55968} to EPSG 3857, checking the diff in the result with the theorical value : {165145.804154 5397552.58726}.

Using the WGS84 ellipsoid, it leads to an error of 10 to 20 kilometers.

Using the one in ​http://hg.orfeo-toolbox.org/OTB/rev/69e1541eddab solves the problem. But to be used, a patch is needed in ossimOgcWktTranslator.

[dburken]

Question:

Difference between (already in place) :

{"6055", "Popular Visualisation", "7059", 0, 0, 0, -180.0, 180.0, -85, 85, 0, 0, 0},

and:

{"WE-EPSG-3857", "Popular Visualisation", "PV", 0, 0, 0, -180.0, 180.0, -85, 85, 0, 0, 0},

[jmalik]

You are probably right. I did not fix this bug so have limited understanding of the mechanisms behind.

Still, it seemed like the : #define NUMBER_OF_THREE_PARAM_DATUMS 226 needed an update (so would be 228 now).

[jmalik]

Confirmed : no need for the additionnal line in ossimDatum.inc

Also the "NUMBER_OF_THREE_PARAM_DATUMS" macro is used only in disabled code, in Utilities/otbossim/src/ossim/base/ossimDatumFactory.cpp:311

Thus the patch reduces to using the 6055 datum instead of WGE, so only ossimOgcWktTranslator need an update.