CODEWITHSUNDEEP
python-3.xnetcdfpython-xarray
Light_B
Light_B

Reputation: 1800

How to stop xarray from automatically changing time attributes when writing out a netcdf file?

I noticed that while concatenating multiple yearly NetCDF files into one file or while splitting a time-series file into yearly groups, xarray's .to_netcdf() automatically updates the time units. Example of what I mean

# time attribute of the file
ncdump -h file_1970_2017.nc
>>double time(time) ;
    time:_FillValue = NaN ;
    time:units = "Hours since 1900-01-01T00:00:00+00:00" ;
    time:calendar = "proleptic_gregorian" ;

#  after splitting the files into yearly files using group-by method the time attribute is automatically modified

# example
ncdump -h file_splitted_2005.nc
>>double time(time) ;
    time:_FillValue = NaN ;
    time:units = "Hours since 2005-01-01T00:00:00+00:00" ;
    time:calendar = "proleptic_gregorian" ;

The same problem is encountered when I do the vice-versa, that is when I concatenate individual yearly files into a common file. Is there some way in which I can force it to not change the time attribute? From the documentation, it seems 'encoding' argument might help but I couldn't figure out how?

Upvotes: 1

Views: 912

Answers (1)

Light_B
Light_B

Reputation: 1800

Figured it out. Using encoding argument as a nested dictionary this could be achieved

# when writing out the dataset ds encoding can be used as
ds.to_netcdf('file_splitted_2005.nc', encoding={'my_variable':{'_FillValue': -999.0},'time':{'units': "seconds since 1900-01-01 00:00:00"}})

If I understood correctly, while writing out the data, xarray will automatically re-calculate the time values based on what units attribute we specify if our time array is a datetime object. In that case it uses the intelligent datetime features under the hood. That means I could also specify

'time':{'units': "seconds since 2000-01-01 00:00:00"}

and it will automatically re-calculate the values it stores in the time array making life much easier for us.

Upvotes: 4

Related Questions