Ncdata :

• is likewise focussed on a neutral representation of netcdf data
• has less rich data manipulation
• includes explicit support for translation to and from iris and xarray

A Python object representation of NetCDF4 data, allowing more flexible programmatic handling of NetCDF files.
Enables quick+easy copying of data from one netCDF4 file to another, with arbitrary changes.
Intended scope similar to NCO commands.

For example::

import ncobj.nc_dataset as ncds
with netCDF4.Dataset(file_in_path) as ds_in:
    in_group = ncds.read(ds_in)
    out_group = ncobj.Group()
    var = in_group.variables['my_temp']
    var.data = var.data[:100, 1::4]
    out_group.variables.add(var)
    ncds.write(file_out_path, out_group)

Latest web docs : http://pp-mo.github.io/build/html/index.html

Important changes in the offing (branch py3):

• drop Python2 + transition to Python3
• adopt ordered containers throughout : accepting that ordering is a "real" property of datasets

VERSION "0.4" : 2018-02-13

• added "ncobj.nc_fake" module, to make ncobj data mimic a readable fake netCDF4.Dataset

VERSION "0.3" : 2014-09-12

• Core classes written and full unit tests.
• File i/o via netCDF4 (ncobj.nc_dataset)
• CDL generation facility for ncobj elements.
• Documentation with Sphinx
• Working examples:
  • generate netcdf file from scratch
  • subset all data by dimensions
  • alphabetic-ordered CDL dumps
  • 'semantic containers' manipulations
• Non-working usecase examples demonstrate intended coding forms + api.