Python

There are two useful libraries for reading OpenDAP. The first is the netCDF4 library. To access opendap urls the linked netcdf.dll/.so/.dylib should have with dap support. The second one is the pydap library.

Library

reads netcdf files

creates netcdf file

netcdf4 support

reads opendap

no need for netcdf.dll

pydap.client

(error)

(error)

(error)

(tick)

(tick)

pupynere

(tick)

(tick)

(error)

(error)

(tick)

Scientific.IO.NetCDF

(tick)

(tick)

(error)

(tick)

(error)

PyNIO

(tick)

(tick)

(tick)

(tick)

(error)

netcdf4-python

(tick)

(tick)

(tick)

(tick) (recent)

(error)

Other, not actively developed, libraries are: pycdf, ncmodule, pynetcdf,

Suggestion: use netcdf4-python or pupynere + pydap

The pydap client and pupynere client both do not require any dll to be present, which can be convenient. The netcdf4-python has the best support for netcdf4 and is installed by python(x,y) under windows, which increases compatibility.

Install

To install the netCDF4 library: if you're on windows the best way to install these is to use python(x,y). On other platforms emerge/apt-get/port install netcdf and pip install netCDF4 or easy_install netCDF4.
To install the pydap library do easy_install pydap or pip install pydap.

Example

All netcdf libraries have approximately the same interface. Below you can find some examples.

Reading opendap with netCDF4 library

#!/usr/bin/env python
# Read data from an opendap server
import netCDF4
# specify an url, the JARKUS dataset in this case
url = 'http://dtvirt5.deltares.nl:8080/thredds/dodsC/opendap/rijkswaterstaat/jarkus/profiles/transect.nc'
# for local windows files, note that '\t' defaults to the tab character in python, so use prefix r to indicate that it is a raw string.
url = r'f:\opendap\rijkswaterstaat\jarkus\profiles\transect.nc' 
# create a dataset object
dataset = netCDF4.Dataset(url)

# lookup a variable
variable = dataset.variables['id']
# print the first 10 values
print variable[0:10]

Reading opendap with pydap.client

#!/usr/bin/env python
# use the pydap client this time
import pydap.client
# specify an url, the JARKUS dataset in this case
url = 'http://dtvirt5.deltares.nl:8080/thredds/dodsC/opendap/rijkswaterstaat/jarkus/profiles/transect.nc'
# create a dataset object
dataset = pydap.client.open_url(url)

# lookup a variable (directly on the dataset)
variable = dataset['id']
# print the first 10 values
print variable[0:10]

Creating a netcdf file with pupynere

#!/usr/bin/env python
import pupynere
# create a new netcdf file
f = pupynere.NetCDFFile('test.nc', 'w')
# create a dimension with length of 10
f.createDimension('npoints', 10)
# create a new variable of type double of dimension npoints
variable = f.createVariable('x', 'd', ('npoints',))
# store an attribute
variable.unit = 'm'
# store some values
variable[:] = range(10)
f.close()

# Now try and open it again
f = pupynere.NetCDFFile('test.nc', 'r')
# read all the data
print f.variables['x'][:]
f.close()

Creating a netcdf file with netCDF4

#!/usr/bin/env python
import netCDF4
# create a new netcdf file
f = netCDF4.Dataset('test.nc', 'w', format='NETCDF3_64BIT')
# create a dimension with length of 10
f.createDimension('npoints', 10)
# create a new variable of type double of dimension npoints
variable = f.createVariable('x', 'd', ('npoints',))
# store an attribute
variable.unit = 'm'
# store some values
variable[:] = range(10)
f.close()

# Now try and open it again
f = netCDF4.Dataset('test.nc', 'r')
# read all the data
print f.variables['x'][:]
f.close()

Disclaimer