There have been many issues related to reprojection of the native spatialite (WGS84) to the model projection. Basically, 1D structures end up at a different place than the user expects.
Projections themselves are well-defined by EPSG, but the transition between different projects are not. There are many details that you should consider when reprojecting:
Reprojection methods get updated from time to time. RD is now for instance in version 4.
Some reprojection methods may be refined by using a "correction grid"
When transitioning from WGS84 to projected system, you should consider tectonic drift ("date dependent datum shift")
User just use some default setting of their GDAL/PostGIS/QGis version to reproject their data from their native projection to WGS84. We then use our default setting to convert it back. Big chance there is an error of a few centimeters to a few meters.
Related to this: how strict are we in processing rasters? Do we inspect the EPSG code advertised by the raster, or do we go into the precise version of the CRS definition and compare that?
caspervdw
commented
2 years ago
There have been many issues related to reprojection of the native spatialite (WGS84) to the model projection. Basically, 1D structures end up at a different place than the user expects.
Projections themselves are well-defined by EPSG, but the transition between different projects are not. There are many details that you should consider when reprojecting:
User just use some default setting of their GDAL/PostGIS/QGis version to reproject their data from their native projection to WGS84. We then use our default setting to convert it back. Big chance there is an error of a few centimeters to a few meters.
Related to this: how strict are we in processing rasters? Do we inspect the EPSG code advertised by the raster, or do we go into the precise version of the CRS definition and compare that?
Possible implementation in the global settings: