The local CRS is defined by a projection (in this case transverse mercator) plus vertical and horizontal adjustment. Horizontal adjustment is a 2D similarity and thus can be described with a 2D Helmert transform. Vertical adjustement is a geoid model in the form of an inclined plane thus it could be described using an affine transform.
I have written a proj pipeline able to realize this transformation. It works great for vector file transformations (thanks to the new GDAL 3 release) but it doesn't work for DSM transformation since the affine transform step used for the vertical adjustment seems to be not supported by gdalwarp (the pixels values stay unchanged).
I guess that gdalwarp only support vertical grid shift to handle pixel value modifications, but anybody can confirm me this ?
If no, which method can accomplish this DSM transformation ? If yes, do you plan to support other 3D transformation with gdalwarp at short or mid terms ?
Aurélien Vila Computer Vision Engineer 676, Rue Max Planck – 31670 Labège - FRANCE Tel: +33 9 71 16 99 01