When installing DAM, there might be issues with the package gdal and its gdal_array functionality. This happens if pip trys to cache a "bad" version of gdal instead of creating new wheels with the version required for this package (see here). This issue really only comes into play when using the warp.match_grid function with nodata_threshold != None at this point, but in the future, it might become a bigger issue.
To solve it. Either install gdal separately before installing DAM, ensure its wheels are build and not taken from cache.
# Create a new virtual environment venv_dir=".venv" python3 -m venv $venv_dir # Activate the virtual environment source $venv_dir/bin/activate # Install the version of GDAL's python bindings matching the one on your machine (must be >=3.4.3) gdalv=$(gdal-config --version) pip install --no-cache-dir --force-reinstall "GDAL[numpy]==$gdalv" # Install DAM pip install https://github.com/c-hydro/dam/archive/dev.zip
Otherwise, pass the same options directly to the DAM installation. This will, however apply the settings to the installation of every required package and be therefore slightly slower.
# Create a new virtual environment venv_dir=".venv" python3 -m venv $venv_dir # Activate the virtual environment source $venv_dir/bin/activate # Install DAM pip install --no-cache-dir --force-reinstall https://github.com/c-hydro/dam/archive/dev.zip