| See discussion:
http://osgeo-org.1560.x6.nabble.com/Problem-with-General-Oblique-transform-latitude-td5257089.html Implementation:
https://github.com/geotools/geotools/blob/master/modules/library/referencing/src/main/java/org/geotools/referencing/operation/projection/GeneralOblique.java Original feature Jira issue:
GEOT-5190 General Oblique Transform for Rotated Pole Coordinates
https://osgeo-org.atlassian.net/browse/GEOT-5190 While attempting use the General Oblique transform, I have encountered a
discrepancy between the latitude of the origin and the projected latitude. In the test script:
https://github.com/geotools/geotools/blob/master/modules/library/referencing/src/test/resources/org/geotools/referencing/test-data/scripts/GeneralOblique.txt The test General Oblique coordinate system is specified with:
The following assertion is made:
This looks wrong to me because I expected the origin of the rotated
coordinate system to have the latitude specified in the parameter
latitude_of_origin, that is 37.5. If I specify my own General_Oblique CRS in GeoServer with
PARAMETER["central_meridian", -106] and {{PARAMETER["latitude_of_origin",
54]}} and use the reprojection console to reproject (0, 0) to WGS84, I get
the (lon, lat) = (-106.0, 36.0) which is incorrect. Likewise {{(-106.0,
54.0)}} projected to this same General Oblique CRS is incorrectly
projected to (0.0, 18.0) (should be (0,0)). In each case, it appears to me that the transformed latitude is 90
degrees minus the expected latitude. This is consistent with the
latitude being measured from the North Pole not the equator. As this is
a common mathematical convention, I think there is a problem with
the implementation of the General Oblique transformation. Looking at equations 3.74 on page 25 (PDF page 31) of part one of
the COSMO manual:
http://www.cosmo-model.org/content/model/documentation/core/cosmoDyncsNumcs.pdf I note that (lambda_g^N, phi_g^N) is the North Pole of the rotated grid,
so phi_g^N is not latitudeOfOrigin, which is measured from the equator. According to the javadoc of MapProjection, points are adjusted for
central_meridian but not latitude_of_origin:
https://github.com/geotools/geotools/blob/master/modules/library/referencing/src/main/java/org/geotools/referencing/operation/projection/MapProjection.java From the discussion On 18/03/16 21:01, Maciej Filocha wrote:
> You're right here. Proj's "+o_lat_p" parameter (undocumented, of course)
> used in my reference example, is latitude of rotated North Pole, not
> latitude of the origin.
> Thank you for pointing this out! |
