HDF4, problem transforming the world corners (-180,90) to equirectangular.

[jluis]

This file reports 0s for all X's ​http://oceancolor.gsfc.nasa.gov/cgi/getfile/S20061822006212.L3m_MO_L555_9.bz2

....
Corner Coordinates:
Upper Left  (       0.000,10018754.171) (  0d 0'0.01"E, 90d 0'0.00"N)
Lower Left  (       0.000,-10018754.171) (  0d 0'0.01"E, 90d 0'0.00"S)
Upper Right (       0.000,10018754.171) (  0d 0'0.01"E, 90d 0'0.00"N)
Lower Right (       0.000,-10018754.171) (  0d 0'0.01"E, 90d 0'0.00"S)
Center      (   0.0000000,   0.0000000) (  0d 0'0.01"E,  0d 0'0.01"N)

[warmerdam]

The HDF4 driver tries to reproject the lat/long boundary to equirectangular coordinates using OGRCoordinateTransform(). Unfortunately, deep down in PROJ.4 the equirectangular projection is treated as spherical instead of ellipsoidal and so the one spatial reference gets changed to a spherical ellipse. The confuses the pj_datum_transform() code and it tries to convert through geocentric coordinates resulting in non-trivial adjustments and as it turns out some sort of failure at the bounds of the world.

I am working on a PROJ.4 fix for this issue that will be part of PROJ 4.6.0.

I am not aware of a workaround at this time.

[warmerdam]

I have established a proj bug to track this issue, and have committed a preliminary fix in PROJ CVS head.

​http://bugzilla.remotesensing.org/show_bug.cgi?id=1602

This fix should come out in a PROJ 4.6.0 release at some point.