Python netCDF：从地理区域创建离散坐标变量

Question

I'm looking to create a netCDF4 file, 'area_nc', with dimensions of time and a discrete, alphanumeric area code, eg 'A0', 'A1', 'A2', 'B0' etc, that doesn't fit to a regular grid (instead of latitude/longitude).

The NetCDF Climate and Forecast (CF) Metadata Conventions pdf suggests that this can be done in Section 4.5, but I can't find any examples.

Is there a way to do this?

So far I've tried:

import netCDF4 as nc

area_nc = nc.Dataset('area.nc', 'w')

area_nc.createDimension('time', None)
area_nc.createDimension('nhood', 6)

time = area_nc.createVariable('time', int, ('time'))
nhood = area_nc.createVariable('nhood', str, ('nhood'))

in_temp = area_nc.createVariable('in_temp', int, ('time', 'nhood'))

nhood[:] = ['A0', 'A1', 'A2', 'B0', 'B1', 'B2']

But I get the error:

IndexError: data can only be assigned to VLEN variables using integer indices

Achieving this, I would then like to add data to the file using this code as a reference, instead of an integer slice, eg:

area_nc.variables['in_temp'][0, 'A0'] = 23

Thanks!

Answer 1

The first part of your question, the error: The error indicates that python considers your dimension to be of variable length (VLEN). The solution lies in better defining the dimension. A solution is changing two of your lines:

nhood = area_nc.createVariable('nhood', str, ('nhood')) ->
nhood = area_nc.createVariable('nhood', '<U13', ('nhood'))

nhood[:] = ['A0', 'A1', 'A2', 'B0', 'B1', 'B2'] ->
nhood[:] = np.array(['A0', 'A1', 'A2', 'B0', 'B1', 'B2'])

Unfortunately, your last line won't work yet still. I could think of the following workaround:

def int_area(area):
    import numpy as np
    return np.where(np.array(['A0', 'A1', 'A2', 'B0', 'B1', 'B2'])==area)[0][0]
area_nc.variables['in_temp'][0, int_area('A0')] = 23