python/netcdf4中的字符串处理

我是netCDF4新手，但我有以下几点：

import netCDF4
nc = netCDF4.Dataset('test.nc','w',format='NETCDF4_CLASSIC')
lat = nc.createVariable('lat','d')
lon = nc.createVariable('lon','d')
place = nc.createVariable('place','c')
lat[:]=17.002
lon[:]=-81.501
#place[:]='test'
nc.close()

我想为变量“place”指定一个大于1个字符的值。如何在不使用属性（例如place.name='test'）的情况下执行此操作？

关键是使用函数将字符串数组转换为字符数组

import netCDF4
import numpy as np

str_out = netCDF4.stringtochar(np.array(['test'], 'S4'))

nc = netCDF4.Dataset('./test.nc', 'w', format='NETCDF4_CLASSIC')
nc.createDimension('lat', 1)
nc.createDimension('lon', 1)
nc.createDimension('nchar', 4)

lat = nc.createVariable('lat', 'f4', ('lat',))
lon = nc.createVariable('lon', 'f4', ('lon',))
place = nc.createVariable('place', 'S1', ('nchar'))

lat[:] = 17.002
lon[:] = -81.501
place[:] = str_out

nc.close()

您可以使用

>>> ncks test.nc 
...
lat[0]=17.002 
lon[0]=-81.501 
nchar[0] place[0--3]='test'

请注意，通过删除“NETCDF4_CLASSIC”的格式，您可以通过另一种方式完成此操作：

str_out = np.array(['test'], dtype='object')

nc = netCDF4.Dataset('./test.nc', 'w')
nc.createDimension('lat', 1)
nc.createDimension('lon', 1)
nc.createDimension('str_dim', 1)

lat = nc.createVariable('lat', 'f4', ('lat',))
lon = nc.createVariable('lon', 'f4', ('lon',))
# Now we can use the data type of 'str' 
place = nc.createVariable('place', str, ('str_dim',))

lat[:] = 17.002
lon[:] = -81.501
place[:] = str_out

nc.close()

>>> ncks test.nc 
...
lat[0]=17.002 
lon[0]=-81.501 
str_dim[0] place[0]=test 

---

对于较长的文本，根据N1B4的有用答案，您可以执行以下操作：

from __future__ import print_function                                                                                                                        
import netCDF4 as nc
import numpy as np

message = """This message is spread over two lines.
I won't even bother counting the characters."""

fname = './test.nc'

# Write the message
with nc.Dataset(fname, 'w') as ds:

    ds.createDimension('nchar', len(message))
    data = ds.createVariable('message', 'S1', ('nchar'))
    data[:] = nc.stringtochar(np.array([message], 'S'))

# Read the message
with nc.Dataset(fname, 'r') as ds:

    reconstructed = ds.variables['message'][...].tostring().decode()

print(reconstructed)