CODEWITHSUNDEEP
python-xarray
Jack Sillin
Jack Sillin

Reputation: 85

Extract Single Value From Xarray Dataset

I have an Xarray dataset with just two pieces of information, a time referenced by 'time' and a time referenced by 'reftime':

<xarray.Dataset>
Dimensions:  ()
Coordinates:
    reftime  datetime64[ns] 2020-03-31T06:00:00
    time     datetime64[ns] 2020-03-31T12:00:00
    crs      object Projection: latitude_longitude
Data variables:
    *empty*
Attributes:
    Originating_or_generating_Center:                                        ...
    Originating_or_generating_Subcenter:                                     ...
    GRIB_table_version:                                                      ...
    Type_of_generating_process:                                              ...
    Analysis_or_forecast_generating_process_identifier_defined_by_originating...
    Conventions:                                                             ...
    history:                                                                 ...
    featureType:                                                             ...
    History:                                                                 ...
    geospatial_lat_min:                                                      ...
    geospatial_lat_max:                                                      ...
    geospatial_lon_min:                                                      ...
    geospatial_lon_max:                                                      ...

everything else is empty. My goal is to get whatever date is referenced by 'reftime' into a string format. Normally, I understand that this can be done by calling dataset['reftime'], but the catch is that this code is intended to run in the background and sometimes it will find xarray datasets where the time I want is referenced by 'reftimeX' where X is some number. How can I extract whatever information is stored in the first coordinate (be it reftime, reftime1, or reftimeX') such that it could be stored as a string?

I've tried turning it into a DataArray in the hopes that I could then turn it into a numpy array and extract the string from there, but when I try to turn it into a DataArray:

filtered_dataarray = filtered_ds.to_array()

I get an error:

ValueError: at least one array or dtype is required

The Xarray docs suggest that this function needs some self parameter:

Dataset.to_array(self, dim='variable', name=None)

But thus far I have been unable to figure out to what this is referring.

Upvotes: 1

Views: 3719

Answers (1)

astoeriko
astoeriko

Reputation: 890

You can get a list of all coordinates in the dataset like this:

coord_names = list(ds.coords)

If you are sure that the coordinate you want is always the first one, you could access it via

ds[coord_names[0]]

However I would rather go through the coords and check what is the exact name of the coordinate you want. Given that you know that it must contain "reftime" you could do:

reftime_name = [var for var in ds.coords if "reftime" in var][0]
ds[reftime_name]

The to_array method does not do what you expect it to. It would take all data variables of the dataset and concatenate them along a new dimension. However, your data set does not contain any data variables (only coords) so it throws an error. The self argument indicates that to_array is an object method. self is a reference to the current instance of the class. Usually, you would call the method on an object (e.g. ds.to_array()) and then you do not need to explicitly pass the self parameter (see also here).

Upvotes: 2

Related Questions