Hello all
I should finish in the next few days the implementation of datum shifts
backed by interpolation grids (NADCON, NTv2, RGF93, etc.). I'm about to
finish the RGF93 one (used in France), which is more complicated than
the others because they basically interpolate the Bursa-Wolf parameters
instead than interpolating the changes in geographic coordinates. The
javadoc at [1] tries to summarize the algorithm. In comparison, NADCON
and NTv2 are much easier and should be implemented fast.
Note: last time I tried RGF93 datum shift with GDAL/Proj.4, I got a
translation in the opposite direction compared to the reference
implementation provided by the French mapping agency. I'm not completely
sure that I have not done something wrong - I would need to test again.
So there is a possibility that SIS and GDAL/Proj.4 give opposite answers
regarding RGF93 datum shifts (a difference of a few hundred metres). But
according my early tests, SIS seems to match the expected answers
reasonably well (by 1 cm).
We will need to take a decision about in which directory (on the user
machine) the users will need to put their grid files (I don't think that
we will be allowed to distribute them with SIS for licensing reasons,
except maybe NADCON). The Proj.4 library put them in
"/usr/local/share/proj" on Unix machines. I will start a new thread on
this subject next week.
Martin
[1]
https://builds.apache.org/job/sis-dev/javadoc/org/apache/sis/referencing/operation/transform/InterpolatedGeocentricTransform.html
