icsm-au/transformation_grids

NTv2 grids to transform data

transformation_grids's Issues

GeoTools/jGridShift reports gsb files as invalid

GeoTools uses the jGridShift library to read and apply NTv2 grid shift files. JGridShift is not happy with the contents of the files, and throws exceptions.

GeoTools uses an older version of of jGridShift (v1.0), but for reference, the source code is a lot like the latest one available here, where it's comparing the nodeCount with lonColumnCount times latRowCount and finds the two numbers are indeed different, hence throws an exception:
https://github.com/floscher/jGridShift/blob/master/src/main/java/au/com/objectix/jgridshift/NTv2SubGrid.java#L187

Trying to load the GDA94_GDA2020_conformal_and_distortion.gsb file the error message is:

java.lang.IllegalStateException: SubGrid WEST_DC5 has inconsistent grid dimensions

For reference, lonColumnCount=1100, latRowCount=1481, nodeCount=1630581 , so there seems to be an excess of nodes? (1481×1100=1629100)

Upon trying to load the GDA94_GDA2020_conformal.gsb file, the jGridShift library throws the following error:

java.lang.IllegalStateException: SubGrid WEST_C5 has inconsistent grid dimesions

The number of cols, rows and node counts is the same as in the other file.

Conflicting information about grid licenses

Here on GitHub the license for the grids are clearly stated as CC BY 4.0, but on http://www.icsm.gov.au/gda/grids.html the same files are said to be licensed under BSD 3-Clause. From previous discussions I know that CC BY 4.0 is the correct license so the ICSM website should be updated to reflect that.

Grid files are too large for default distribution

The grid files here are very large, and unfortunately as a result they've been blocked from inclusion in the proj-datumgrid package. This means that the grid files are not available by default on important platforms such as the osgeo4w install (accordingly they are not available in a qgis install without manual user installation).

Is there any chance the grids could be thinned to avoid this situation?

Relevant discussion is at:

OSGeo/proj-datumgrid#16 (comment)