Generating NetCDF files with

Python
Thanks to:

Stephen Pascoe.
 We are using netCDF4-python!
There are many options for working with NetCDF files
in Python. In this example we have chosen to highlight
the use of the netCDF4-python module.

The netCDF4-python module is useful because:

• It implements the basic “classic” model as well as
more advanced features.
• It provides a simple interface to the NetCDF
structure.
• It has been used as the underlying NetCDF I/O layer
for many more advanced packages.
 Creating/Opening a netCDF file

To create a netCDF file from python, you simply call the

Dataset() constructor. This is also the method used
to open an existing netCDF file.

If the file is open for write access (w, r+ or a), you

may write any type of data including new dimensions,
groups, variables and attributes.
 Working with “classic” NetCDF
The netCDF4 module can read and write files in any
netCDF format. When creating a new file, the format
may be specified using the format keyword in the
Dataset constructor. The default format is
NETCDF4.

This tutorial will focus exclusively on the NetCDF-

”classic” data model using: NETCDF4_CLASSIC
 Creating a NetCDF file
Open a new NetCDF file in “write” ('w') mode:

>>> from netCDF4 import Dataset

>>> dataset = Dataset('data/test.nc', 'w',
format='NETCDF4_CLASSIC')

>>> print dataset.file_format

NETCDF4_CLASSIC
 Create dimensions
Define a set of dimensions used for your variables:

level = dataset.createDimension('level', 10)

lat = dataset.createDimension('lat', 73)
lon = dataset.createDimension('lon', 144)
time = dataset.createDimension('time', None)

And you can query dimensions with:

>>> print len(lon)
144
>>> print time.isunlimited() # is special!
True
 Dimensions
All of the Dimension instances are stored in a python
dictionary. This allows you to access each dimension by
its name using dictionary key access:
print 'Lon dimension:', dataset.dimensions['lon']
for dimname in dataset.dimensions.keys():
dim = dataset.dimensions[dimname]
print dimname, len(dim), dim.isunlimited()

Lon dimension: <type 'netCDF4.Dimension'>: name = 'lon', size = 144

level 10 False
time 0 True
lat 73 False
lon 144 False
 Variables
NetCDF variables behave much like python multi-
dimensional arrays in numpy. However, unlike numpy
arrays, netCDF4 variables can be appended to along the
'unlimited' dimension (a.k.a. the "record dimension").
To create a netCDF variable, use:
Dataset.createVariable(<var_id>, <type>,
<dimensions>)

This method has two mandatory arguments: the variable

ID (a Python string) and the variable datatype.
Additionally a tuple of dimensions can be provided.
 Creating a variable
import numpy as np

# Create coordinate variables for 4-dimensions

times = dataset.createVariable('time', np.float64, ('time',))

levels = dataset.createVariable('level', np.int32, ('level',))
latitudes = dataset.createVariable('latitude', np.float32,
('lat',))
longitudes = dataset.createVariable('longitude', np.float32,
('lon',))

# Create the actual 4-d variable

temp = dataset.createVariable('temp', np.float32,
('time','level','lat','lon'))
 Accessing variables
All of the variables in the Dataset are stored in a
Python dictionary, in the same way as the dimensions:
>>> print 'temp variable:', dataset.variables['temp']

>>> for varname in dataset.variables.keys():

... var = dataset.variables[varname]
... print varname, var.dtype, var.dimensions, var.shape

temp variable: <type 'netCDF4.Variable'> float32 temp(time, level,

lat, lon) unlimited dimensions: time
current shape = (0, 10, 73, 144)
time float64 (u'time',) (0,)
level int32 (u'level',) (10,)
latitude float32 (u'lat',) (73,)
longitude float32 (u'lon',) (144,)
temp float32 (u'time', u'level', u'lat', u'lon') (0, 10, 73, 144)
 Attributes (global)
Global attributes are set by assigning values to
Dataset instance variables. Attributes can be strings,
numbers or sequences. Returning to our example:

import time

# Global Attributes
dataset.description = 'bogus example script'
dataset.history = 'Created ' + time.ctime(time.time())
dataset.source = 'netCDF4 python module tutorial'
 Attributes (variable)
Variable attributes are set by assigning values to
Variable instance variables:

# Variable Attributes
latitudes.units = 'degree_north'
longitudes.units = 'degree_east'
levels.units = 'hPa'
temp.units = 'K'

times.units = 'hours since 0001-01-01 00:00:00'

times.calendar = 'gregorian'
 Accessing attributes
Attributes are accessed as attributes of their
relevant instances:

>>> print dataset.description

bogus example script

>>> print dataset.history

Created Mon Mar 17 01:12:31 2014
 Writing data
Now that you have a netCDF Variable instance, how
do you put data into it? You can just treat it like an array
and assign data to a slice.
>>> lats = np.arange(-90,91,2.5)
>>> lons = np.arange(-180,180,2.5)
>>> latitudes[:] = lats
>>> longitudes[:] = lons
>>> print 'latitudes =\n', latitudes[:]
latitudes =
[-90. -87.5 -85. -82.5 -80. -77.5 -75. -72.5 -70.
...
-60. -57.5 -55. -52.5 -50. -47.5 -45. -42.5 -40
...
 Growing data along unlimited
dimension (1)
Unlike NumPy's array objects, netCDF Variable
objects that have an unlimited dimension will grow
along that dimension if you assign data outside the
currently defined range of indices.

>>> print 'temp shape before adding data = ', temp.shape

temp shape before adding data = (0, 10, 73, 144)

 Growing data along unlimited
dimension (2)
>>> from numpy.random import uniform
>>> nlats = len(dataset.dimensions['lat'])
>>> nlons = len(dataset.dimensions['lon'])
>>> temp[0:5,:,:,:] = uniform(size=(5,10,nlats,nlons))
>>> print 'temp shape after adding data = ', temp.shape

temp shape after adding data = (5, 10, 73, 144)

temp shape before adding data = (0, 10, 73, 144)

NOTE: numpy.random.uniform(size = X) returns values

from a uniform distribution in a numpy array with
dimensions expressed in a tuple X.
 Defining date/times correctly (1)
Time coordinate values pose a special challenge to
netCDF users. Most metadata standards (such as CF and
COARDS) specify that time should be measure relative
to a fixed date using a certain calendar, with units
specified like "hours since YY:MM:DD hh-mm-ss".

These units can be awkward to deal with, without a

utility to convert the values to and from calendar dates.
The functions num2date() and date2num() are
provided with this package to do just that. Here's an
example of how they can be used…
 Defining date/times correctly (2)
>>> # Fill in times.
>>> from datetime import datetime, timedelta
>>> from netCDF4 import num2date, date2num
>>> dates = []
>>> for n in range(temp.shape[0]):
... dates.append(datetime(2001, 3, 1) + n *
timedelta(hours=12))
>>> times[:] = date2num(dates, units = times.units,
calendar = times.calendar)
>>> print 'time values (in units %s): ' % times.units +
'\n', times[:]
time values (in units hours since 0001-01-01
00:00:00.0): [ 17533104. 17533116. 17533128. 17533140.
17533152.]
 Defining date/times correctly (3)
>>> dates = num2date(times[:], units=times.units,
calendar=times.calendar)
>>> print 'dates corresponding to time values:\n', dates

dates corresponding to time values:

[datetime.datetime(2001, 3, 1, 0, 0)
datetime.datetime(2001, 3, 1, 12, 0)
datetime.datetime(2001, 3, 2, 0, 0)
datetime.datetime(2001, 3, 2, 12, 0)
datetime.datetime(2001, 3, 3, 0, 0)]

num2date() converts numeric values of time in the specified

units and calendar to datetime objects, and
date2num() does the reverse.
 Finally, let’s write the file
Simply…

>>> dataset.close()
# and the file is written!
 I wonder what it looks like
netcdf test {
dimensions:
level = 10 ;
time = UNLIMITED ; // (5 currently)
lat = 73 ;
lon = 144 ;
variables:
double time(time) ;
time:units = "hours since 0001-01-01 00:00:00.0" ;
time:calendar = "gregorian" ;
int level(level) ;
level:units = "hPa" ;
float latitude(lat) ;
latitude:units = "degrees north" ;
float longitude(lon) ;
longitude:units = "degrees east" ;
float temp(time, level, lat, lon) ;
temp:units = "K" ;
// global attributes:
:description = "bogus example script" ;
:history = "Created Mon Mar 17 01:12:31 2014" ;
:source = "netCDF4 python module tutorial" ;
}
 Further reading
Python-netCDF4:
http://netcdf4-python.googlecode.com/svn/trunk/docs/netCDF4-module.html