Instead of renaming the files to create a multi-file dataset, we use a .vrt file to create a single-file dataset. This has several advantages:
First and foremost it fixes the problems reported in issue #24. Multi-file datasets fail to properly handle the boundary regions, where the tiles meet each other.
The setup is simplified. Renaming files to match the SRTM conventions is not necessary any more.
Also the config.yaml gets simpler. The parameters filename_epsg and filename_tile_size are not needed.
One small downside is that gdal is required for building the .vrt, but I don't see this as a big hurdle.
Probably also the instructions for the other datasets will need to be fixed in this respect, but I since I found and verified this problem only with EU-DEM so far, I'm starting with this one.
Instead of renaming the files to create a multi-file dataset, we use a .vrt file to create a single-file dataset. This has several advantages:
config.yaml
gets simpler. The parametersfilename_epsg
andfilename_tile_size
are not needed.One small downside is that gdal is required for building the .vrt, but I don't see this as a big hurdle.
Probably also the instructions for the other datasets will need to be fixed in this respect, but I since I found and verified this problem only with EU-DEM so far, I'm starting with this one.