Using python-xarray to extract a single variable from a multidimensional netcdf file while keeping attributes/metadata

Question

I have a large netcdf file with many dimensions and attributes. I want to extract a single variable from this file and save this as a new netcdf file, while keeping all of the original metadata. I am using xarray.

I open the dataset using:

dr=xr.open_dataset("path_to_file")

And it looks like this when I print it (some dimensions and metadata removed for simplicity:

<xarray.Dataset>
Dimensions:                (Time: 464, bottom_top: 39, bottom_top_stag: 40, snow_layers_stag: 3, snso_layers_stag: 7, soil_layers_stag: 4, south_north: 186, south_north_stag: 187, west_east: 246, west_east_stag: 247)
Coordinates:
    XLAT                   (Time, south_north, west_east) float32 ...
    XLONG                  (Time, south_north, west_east) float32 ...
    XTIME                  (Time) datetime64[ns] ...
    XLAT_U                 (Time, south_north, west_east_stag) float32 ...
    XLONG_U                (Time, south_north, west_east_stag) float32 ...
    XLAT_V                 (Time, south_north_stag, west_east) float32 ...
    XLONG_V                (Time, south_north_stag, west_east) float32 ...
Dimensions without coordinates: Time, bottom_top, bottom_top_stag, snow_layers_stag, snso_layers_stag, soil_layers_stag, south_north, south_north_stag, west_east, west_east_stag
Data variables:
    Times                  (Time) |S19 ...
    UST                    (Time, south_north, west_east) float32 ...
    ZNU                    (Time, bottom_top) float32 ...
    ZNW                    (Time, bottom_top_stag) float32 ...
    ZS                     (Time, soil_layers_stag) float32 ...
    DZS                    (Time, soil_layers_stag) float32 ...

Attributes:
    TITLE:                            OUTPUT FROM WRF V3.9 MODEL
    START_DATE:                      2017-10-31_00:00:00
    SIMULATION_START_DATE:           2017-10-01_00:00:00
    WEST-EAST_GRID_DIMENSION:        247
    SOUTH-NORTH_GRID_DIMENSION:      187
    BOTTOM-TOP_GRID_DIMENSION:       40

    HYBRID_OPT:                      -1
    ETAC:                            0.0

I want to extract just UST, so I try:

dr_u = dr['UST']

But when I print the resulting dr_u, the metadata is gone:

<xarray.Dataset>
Dimensions:  (Time: 464, south_north: 186, west_east: 246)
Coordinates:
    XLAT     (Time, south_north, west_east) float32 ...
    XLONG    (Time, south_north, west_east) float32 ...
    XTIME    (Time) datetime64[ns] ...
Dimensions without coordinates: Time, south_north, west_east
Data variables:
    UST      (Time, south_north, west_east) float32 ...

I want to be able to keep all of the information under the Attributes heading in the original file. I'm aware there is a flag called keep_attrs in the xarray package that seems like it would be useful for this, but I can't work out how to use it on this operation.

bwc · Accepted Answer · 2019-11-06 17:51:34Z

1

You can retrieve the attribute dictionary from an xarray object using ds.attrs

You can just assign the attributes manually:

dr_u.attrs = dr.attrs

answered Nov 6, 2019 at 17:51

bwc

1,1158 silver badges19 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

agorapotatoes Over a year ago

I thought this had solved the problem but unfortunately I find that if I then try to save the xarray as a netcdf file and read it back in, the attributes have disappeared. I don't have this problem if I read in a netcdf file and then immediately save it again, only if I go through the process of extracting a single variable and then adding the attributes. Do you know how to solve this issue? I have looked at the possible arguments for to_netcdf and found nothing that seems useful.

kmuehlbauer Over a year ago

You would need to create a Dataset with your DataArray inside. Then add the attrs to the Dataset.

Collectives™ on Stack Overflow

Using python-xarray to extract a single variable from a multidimensional netcdf file while keeping attributes/metadata

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related