added band_as_variable option to open_rasterio #592
Conversation
|
Thanks for taking a stab at this @clausmichele 👍 The design in #296 would be to load the band data in separate variables so you can do this: xds = rioxarray.open_rasterio('./results/S2_L2A_data_mod.tiff',band_as_variable=True)
xds["B02"]This enables storing the description, nodata values, etc separately in the attributes of each variable. An example implementation is here: https://github.com/bopen/xarray-gdal/blob/main/xarray_gdal/xarray_plugin.py |
|
Ok, second iteration, thanks for the link to
|
| # If we have a Dataset with variables and each variable has already 'grid_mapping: spatial_ref' | ||
| # in the variable attrs, doing the next steps would add another 'spatial_ref' as a | ||
| # coordinate and will break the functionality of xarray.to_nectdf() | ||
| # TODO: Probably we would need to check if the transform is already available in each band? |
There was a problem hiding this comment.
The transform should be the same for each band.
| rio_crs = riods.crs or result.rio.crs | ||
| if rio_crs: | ||
| result.rio.write_crs(rio_crs, inplace=True) | ||
| # If we have a Dataset with variables and each variable has already 'grid_mapping: spatial_ref' |
There was a problem hiding this comment.
I recommend not adding the CRS to each variable as you did previously and instead add it to the dataset here.
| for attr, value in result.attrs.items() | ||
| if not attr.startswith(f"{result.name}#") | ||
| } | ||
| # remove duplicate tags for Dataset, otherwise there are issues opening the netCDF file in QGIS |
There was a problem hiding this comment.
This is interesting. Mind adding a separate function called _clean_attrs and moving the duplicate cleanup there?
| output_is_dataset = True | ||
| data_vars = {} | ||
| for i in riods.indexes: | ||
| data_var_attrs = { |
There was a problem hiding this comment.
I am wondering if band specific tags should be added here?
|
@clausmichele thanks for working on this. I added some comments and will try to do a more thorough review later. |
|
Okay, this got me down a rabbit hole to a different implementation to preserve the lazy loading (see #600). Want to give it a try and see what you think? |
First draft of a possible implementation of
read_variable_as_bands.@snowman2 what do you think? Is it too naive as an approach?
docs/history.rstfor all changes anddocs/rioxarray.rstfor new API