Ticket #4285 (new defect)
OGR does not support PROJ.4 +proj=ob_tran , rotated_pole is not supported by the netcdf driver
|Reported by:||etourigny||Owned by:||warmerdam|
|Severity:||normal||Keywords:||ogr, proj, netcdf,|
|Cc:||pds, peifer, rouault, warmerdam|
It seems OGR and geotiff (maybe proj.4 too?) need to be modified to support the ob_tran projection (General Oblique Transformation).
The problem was initially reported in bug #4251:
rotated pole grids ARE supported by PROJ.4 (through ob_tran transformation, in a somewhat counter-intuitive way, see my recent post at: http://osgeo-org.1803224.n2.nabble.com/forward-inverse-projection-logic-for-ob-tran-td6874351.html). So the CRS of the data is, in PROJ.4 notation: +proj=ob_tran +o_proj=latlon +o_lon_p=-162 +o_lat_p=39.25 +lon_0=180 However, GDAL/OGR's limited projection support via gdal/ogr/ogr_srs_proj4.cpp doesn't include ob_tran transformation, which is indeed a pity.
As rotated_pole is quite common in netcdf-cf files created by RCMs (mostly from European centers), it is important to support it within gdal. In fact, ob_tran may be the solution to support any other projections we might not support yet in the netcdf driver?
Perhaps Pat and 'peifer' would like to work on this? Adding Frank and Even in cc as they might have something to say.
It seems from this message that in fact ob_tran is not fully supported by proj.4? http://osgeo-org.1803224.n2.nabble.com/forward-inverse-projection-logic-for-ob-tran-td6874351.html
From this post it seems we need to use the inverse projection from rotated pole to lat/lon. http://osgeo-org.1803224.n2.nabble.com/Re-rotated-pole-ob-tran-help-needed-tt5149504.html#none
Perhaps this workaround suggested by Frank, using a WKT EXTENSION node works?: http://osgeo-org.1803224.n2.nabble.com/Does-GDAL-handle-rotated-spheres-td2030419.html