<?xml version="1.0" encoding="ascii"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<title>netCDF4</title>
<link rel="stylesheet" href="epydoc.css" type="text/css" />
<script type="text/javascript" src="epydoc.js"></script>
</head>
<body bgcolor="white" text="black" link="blue" vlink="#204080"
alink="#204080">
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Home link -->
<th bgcolor="#70b0f0" class="navbar-select"
> Home </th>
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<th class="navbar" width="100%"></th>
</tr>
</table>
<table width="100%" cellpadding="0" cellspacing="0">
<tr valign="top">
<td width="100%">
<span class="breadcrumbs">
Module netCDF4
</span>
</td>
<td>
<table cellpadding="0" cellspacing="0">
<!-- hide/show private -->
</table>
</td>
</tr>
</table>
<!-- ==================== MODULE DESCRIPTION ==================== -->
<h1 class="epydoc">Module netCDF4</h1><p class="nomargin-top"></p>
<h1 class="heading">Introduction</h1>
<p>Python interface to the netCDF version 4 library. <a
href="http://www.unidata.ucar.edu/software/netcdf/netcdf-4"
target="_top">netCDF version 4</a> has many features not found in
earlier versions of the library and is implemented on top of <a
href="http://www.hdfgroup.org/HDF5" target="_top">HDF5</a>. This module
can read and write files in both the new netCDF 4 and the old netCDF 3
format, and can create files that are readable by HDF5 clients. The API
modelled after <a
href="http://dirac.cnrs-orleans.fr/plone/software/scientificpython/"
target="_top">Scientific.IO.NetCDF</a>, and should be familiar to users
of that module.</p>
<p>Most new features of netCDF 4 are implemented, such as multiple
unlimited dimensions, groups and zlib data compression. All the new
numeric data types (such as 64 bit and unsigned integer types) are
implemented. Compound and variable length (vlen) data types are
supported, but the enum and opaque data types are not. Mixtures of
compound and vlen data types (compound types containing vlens, and
vlens containing compound types) are not supported.</p>
<h1 class="heading">Download</h1>
<ul>
<li>
Latest bleeding-edge code from the <a
href="http://github.com/Unidata/netcdf4-python"
target="_top">github repository</a>.
</li>
<li>
Latest <a href="https://pypi.python.org/pypi/netCDF4"
target="_top">releases</a> (source code and windows installers).
</li>
</ul>
<h1 class="heading">Requires</h1>
<ul>
<li>
Python 2.5 or later (python 3 works too).
</li>
<li>
numpy array module <a href="http://numpy.scipy.org"
target="_top">http://numpy.scipy.org</a>, version 1.3.0 or later
(1.5.1 or higher recommended, required if using python 3).
</li>
<li>
<a href="http://cython.org" target="_top">Cython</a>, version 0.19
or later, is optional - if it is installed setup.py will use it to
recompile the Cython source code into C, using conditional
compilation to enable features in the netCDF API that have been
added since version 4.1.1. If Cython is not installed, these
features (such as the ability to rename Group objects) will be
disabled to preserve backward compatibility with older versions of
the netCDF library.
</li>
<li>
For python < 2.7, the ordereddict module <a
href="http://python.org/pypi/ordereddict"
target="_top">http://python.org/pypi/ordereddict</a>.
</li>
<li>
The HDF5 C library version 1.8.4-patch1 or higher (1.8.8 or higher
recommended) from <a href="ftp://ftp.hdfgroup.org/HDF5/current/src"
target="_top">ftp://ftp.hdfgroup.org/HDF5/current/src</a>. Be sure
to build with '<code>--enable-hl --enable-shared</code>'.
</li>
<li>
<a href="http://curl.haxx.se/libcurl/" target="_top">Libcurl</a>,
if you want <a href="http://opendap.org/" target="_top">OPeNDAP</a>
support.
</li>
<li>
<a href="http://www.hdfgroup.org/products/hdf4/"
target="_top">HDF4</a>, if you want to be able to read HDF4
"Scientific Dataset" (SD) files.
</li>
<li>
The netCDF-4 C library from <a
href="ftp://ftp.unidata.ucar.edu/pub/netcdf"
target="_top">ftp://ftp.unidata.ucar.edu/pub/netcdf</a>. Version
4.1.1 or higher is required (4.2 or higher recommended). Be sure to
build with '<code>--enable-netcdf-4 --enable-shared</code>', and
set <code>CPPFLAGS="-I $HDF5_DIR/include"</code> and
<code>LDFLAGS="-L $HDF5_DIR/lib"</code>, where
<code>$HDF5_DIR</code> is the directory where HDF5 was installed.
If you want <a href="http://opendap.org/" target="_top">OPeNDAP</a>
support, add '<code>--enable-dap</code>'. If you want HDF4 SD
support, add '<code>--enable-hdf4</code>' and add the location of
the HDF4 headers and library to <code>CPPFLAGS</code> and
<code>LDFLAGS</code>.
</li>
</ul>
<h1 class="heading">Install</h1>
<ul>
<li>
install the requisite python modules and C libraries (see above).
It's easiest if all the C libs are built as shared libraries.
</li>
<li>
By default, the utility <code>nc-config</code>, installed with
netcdf 4.1.2 or higher, will be run used to determine where all the
dependencies live.
</li>
<li>
If <code>nc-config</code> is not in your default <code>PATH</code>,
rename the file <code>setup.cfg.template</code> to
<code>setup.cfg</code>, then edit in a text editor (follow the
instructions in the comments). In addition to specifying the path
to <code>nc-config</code>, you can manually set the paths to all
the libraries and their include files (in case
<code>nc-config</code> does not do the right thing).
</li>
<li>
run <code>python setup.py build</code>, then <code>python setup.py
install</code> (as root if necessary).
</li>
<li>
run the tests in the 'test' directory by running <code>python
run_all.py</code>.
</li>
</ul>
<h1 class="heading">Tutorial</h1>
<h2 class="heading">1) Creating/Opening/Closing a netCDF file</h2>
<p>To create a netCDF file from python, you simply call the <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>
constructor. This is also the method used to open an existing netCDF
file. If the file is open for write access (<code>w, r+</code> or
<code>a</code>), you may write any type of data including new
dimensions, groups, variables and attributes. netCDF files come in
several flavors (<code>NETCDF3_CLASSIC, NETCDF3_64BIT,
NETCDF4_CLASSIC</code>, and <code>NETCDF4</code>). The first two
flavors are supported by version 3 of the netCDF library.
<code>NETCDF4_CLASSIC</code> files use the version 4 disk format
(HDF5), but do not use any features not found in the version 3 API.
They can be read by netCDF 3 clients only if they have been relinked
against the netCDF 4 library. They can also be read by HDF5 clients.
<code>NETCDF4</code> files use the version 4 disk format (HDF5) and
use the new features of the version 4 API. The <code>netCDF4</code>
module can read and write files in any of these formats. When
creating a new file, the format may be specified using the
<code>format</code> keyword in the <code>Dataset</code> constructor.
The default format is <code>NETCDF4</code>. To see how a given file
is formatted, you can examine the <code>data_model</code> <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> attribute.
Closing the netCDF file is accomplished via the <a
href="netCDF4.Dataset-class.html#close" class="link">close</a> method
of the <a href="netCDF4.Dataset-class.html" class="link">Dataset</a>
instance.</p>
<p>Here's an example:</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">from</span> netCDF4 <span class="py-keyword">import</span> Dataset
<span class="py-prompt">>>> </span>rootgrp = Dataset(<span class="py-string">'test.nc'</span>, <span class="py-string">'w'</span>, format=<span class="py-string">'NETCDF4'</span>)
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp.data_model
<span class="py-output">NETCDF4</span>
<span class="py-output"></span><span class="py-prompt">>>></span>
<span class="py-prompt">>>> </span>rootgrp.close()</pre>
<p>Remote <a href="http://opendap.org"
target="_top">OPeNDAP</a>-hosted datasets can be accessed for reading
over http if a URL is provided to the <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>
constructor instead of a filename. However, this requires that the
netCDF library be built with OPenDAP support, via the
<code>--enable-dap</code> configure option (added in version
4.0.1).</p>
<h2 class="heading">2) Groups in a netCDF file</h2>
<p>netCDF version 4 added support for organizing data in hierarchical
groups, which are analagous to directories in a filesystem. Groups
serve as containers for variables, dimensions and attributes, as well
as other groups. A <code>netCDF4.Dataset</code> defines creates a
special group, called the 'root group', which is similar to the root
directory in a unix filesystem. To create <a
href="netCDF4.Group-class.html" class="link">Group</a> instances, use
the <a href="netCDF4.Dataset-class.html#createGroup"
class="link">createGroup</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> instance. <a
href="netCDF4.Dataset-class.html#createGroup"
class="link">createGroup</a> takes a single argument, a python string
containing the name of the new group. The new <a
href="netCDF4.Group-class.html" class="link">Group</a> instances
contained within the root group can be accessed by name using the
<code>groups</code> dictionary attribute of the <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> instance.
Only <code>NETCDF4</code> formatted files support Groups, if you try
to create a Group in a netCDF 3 file you will get an error
message.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>rootgrp = Dataset(<span class="py-string">'test.nc'</span>, <span class="py-string">'a'</span>)
<span class="py-prompt">>>> </span>fcstgrp = rootgrp.createGroup(<span class="py-string">'forecasts'</span>)
<span class="py-prompt">>>> </span>analgrp = rootgrp.createGroup(<span class="py-string">'analyses'</span>)
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp.groups
<span class="py-output">OrderedDict([('forecasts', <netCDF4.Group object at 0x1b4b7b0>),</span>
<span class="py-output"> ('analyses', <netCDF4.Group object at 0x1b4b970>)])</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Groups can exist within groups in a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>, just as
directories exist within directories in a unix filesystem. Each <a
href="netCDF4.Group-class.html" class="link">Group</a> instance has a
<code>'groups'</code> attribute dictionary containing all of the
group instances contained within that group. Each <a
href="netCDF4.Group-class.html" class="link">Group</a> instance also
has a <code>'path'</code> attribute that contains a simulated unix
directory path to that group.</p>
<p>Here's an example that shows how to navigate all the groups in a
<a href="netCDF4.Dataset-class.html" class="link">Dataset</a>. The
function <code>walktree</code> is a Python generator that is used to
walk the directory tree. Note that printing the <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> object yields
summary information about it's contents.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>fcstgrp1 = fcstgrp.createGroup(<span class="py-string">'model1'</span>)
<span class="py-prompt">>>> </span>fcstgrp2 = fcstgrp.createGroup(<span class="py-string">'model2'</span>)
<span class="py-prompt">>>> </span><span class="py-keyword">def</span> <span class="py-defname">walktree</span>(top):
<span class="py-prompt">>>> </span> <span class="py-builtin">values</span> = top.groups.values()
<span class="py-prompt">>>> </span> yield <span class="py-builtin">values</span>
<span class="py-prompt">>>> </span> <span class="py-keyword">for</span> value <span class="py-keyword">in</span> top.groups.values():
<span class="py-prompt">>>> </span> <span class="py-keyword">for</span> children <span class="py-keyword">in</span> walktree(value):
<span class="py-prompt">>>> </span> yield children
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp
<span class="py-prompt">>>> </span><span class="py-keyword">for</span> children <span class="py-keyword">in</span> walktree(rootgrp):
<span class="py-prompt">>>> </span> <span class="py-keyword">for</span> child <span class="py-keyword">in</span> children:
<span class="py-prompt">>>> </span> <span class="py-keyword">print</span> child
<span class="py-output"><type 'netCDF4.Dataset'></span>
<span class="py-output">root group (NETCDF4 file format):</span>
<span class="py-output"> dimensions: </span>
<span class="py-output"> variables: </span>
<span class="py-output"> groups: forecasts, analyses</span>
<span class="py-output"><type 'netCDF4.Group'></span>
<span class="py-output">group /forecasts:</span>
<span class="py-output"> dimensions:</span>
<span class="py-output"> variables:</span>
<span class="py-output"> groups: model1, model2</span>
<span class="py-output"><type 'netCDF4.Group'></span>
<span class="py-output">group /analyses:</span>
<span class="py-output"> dimensions:</span>
<span class="py-output"> variables:</span>
<span class="py-output"> groups:</span>
<span class="py-output"><type 'netCDF4.Group'></span>
<span class="py-output">group /forecasts/model1:</span>
<span class="py-output"> dimensions:</span>
<span class="py-output"> variables:</span>
<span class="py-output"> groups:</span>
<span class="py-output"><type 'netCDF4.Group'></span>
<span class="py-output">group /forecasts/model2:</span>
<span class="py-output"> dimensions:</span>
<span class="py-output"> variables:</span>
<span class="py-output"> groups:</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<h2 class="heading">3) Dimensions in a netCDF file</h2>
<p>netCDF defines the sizes of all variables in terms of dimensions,
so before any variables can be created the dimensions they use must
be created first. A special case, not often used in practice, is that
of a scalar variable, which has no dimensions. A dimension is created
using the <a href="netCDF4.Dataset-class.html#createDimension"
class="link">createDimension</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> instance. A
Python string is used to set the name of the dimension, and an
integer value is used to set the size. To create an unlimited
dimension (a dimension that can be appended to), the size value is
set to <code>None</code> or 0. In this example, there both the
<code>time</code> and <code>level</code> dimensions are unlimited.
Having more than one unlimited dimension is a new netCDF 4 feature,
in netCDF 3 files there may be only one, and it must be the first
(leftmost) dimension of the variable.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>level = rootgrp.createDimension(<span class="py-string">'level'</span>, None)
<span class="py-prompt">>>> </span>time = rootgrp.createDimension(<span class="py-string">'time'</span>, None)
<span class="py-prompt">>>> </span>lat = rootgrp.createDimension(<span class="py-string">'lat'</span>, 73)
<span class="py-prompt">>>> </span>lon = rootgrp.createDimension(<span class="py-string">'lon'</span>, 144)</pre>
<p>All of the <a href="netCDF4.Dimension-class.html"
class="link">Dimension</a> instances are stored in a python
dictionary.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp.dimensions
<span class="py-output">OrderedDict([('level', <netCDF4.Dimension object at 0x1b48030>),</span>
<span class="py-output"> ('time', <netCDF4.Dimension object at 0x1b481c0>),</span>
<span class="py-output"> ('lat', <netCDF4.Dimension object at 0x1b480f8>),</span>
<span class="py-output"> ('lon', <netCDF4.Dimension object at 0x1b48a08>)])</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Calling the python <code>len</code> function with a <a
href="netCDF4.Dimension-class.html" class="link">Dimension</a>
instance returns the current size of that dimension. The <a
href="netCDF4.Dimension-class.html#isunlimited"
class="link">isunlimited</a> method of a <a
href="netCDF4.Dimension-class.html" class="link">Dimension</a>
instance can be used to determine if the dimensions is unlimited, or
appendable.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> len(lon)
<span class="py-output">144</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> len.is_unlimited()
<span class="py-output">False</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> time.is_unlimited()
<span class="py-output">True</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Printing the <a href="netCDF4.Dimension-class.html"
class="link">Dimension</a> object provides useful summary info,
including the name and length of the dimension, and whether it is
unlimited.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">for</span> dimobj <span class="py-keyword">in</span> rootgrp.dimensions.values():
<span class="py-prompt">>>> </span> <span class="py-keyword">print</span> dimobj
<span class="py-output"><type 'netCDF4.Dimension'> (unlimited): name = 'level', size = 0</span>
<span class="py-output"><type 'netCDF4.Dimension'> (unlimited): name = 'time', size = 0</span>
<span class="py-output"><type 'netCDF4.Dimension'>: name = 'lat', size = 73</span>
<span class="py-output"><type 'netCDF4.Dimension'>: name = 'lon', size = 144</span>
<span class="py-output"><type 'netCDF4.Dimension'> (unlimited): name = 'time', size = 0</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p><a href="netCDF4.Dimension-class.html" class="link">Dimension</a>
names can be changed using the <a
href="netCDF4.Dataset-class.html#renameDimension"
class="link">renameDimension</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> instance.</p>
<h2 class="heading">4) Variables in a netCDF file</h2>
<p>netCDF variables behave much like python multidimensional array
objects supplied by the <a href="http://numpy.scipy.org"
target="_top">numpy module</a>. However, unlike numpy arrays, netCDF4
variables can be appended to along one or more 'unlimited'
dimensions. To create a netCDF variable, use the <a
href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> instance. The
<a href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a> method has two mandatory arguments,
the variable name (a Python string), and the variable datatype. The
variable's dimensions are given by a tuple containing the dimension
names (defined previously with <a
href="netCDF4.Dataset-class.html#createDimension"
class="link">createDimension</a>). To create a scalar variable,
simply leave out the dimensions keyword. The variable primitive
datatypes correspond to the dtype attribute of a numpy array. You can
specify the datatype as a numpy dtype object, or anything that can be
converted to a numpy dtype object. Valid datatype specifiers
include: <code>'f4'</code> (32-bit floating point), <code>'f8'</code>
(64-bit floating point), <code>'i4'</code> (32-bit signed integer),
<code>'i2'</code> (16-bit signed integer), <code>'i8'</code> (64-bit
singed integer), <code>'i1'</code> (8-bit signed integer),
<code>'u1'</code> (8-bit unsigned integer), <code>'u2'</code> (16-bit
unsigned integer), <code>'u4'</code> (32-bit unsigned integer),
<code>'u8'</code> (64-bit unsigned integer), or <code>'S1'</code>
(single-character string). The old Numeric single-character
typecodes (<code>'f'</code>,<code>'d'</code>,<code>'h'</code>,
<code>'s'</code>,<code>'b'</code>,<code>'B'</code>,<code>'c'</code>,<code>'i'</code>,<code>'l'</code>),
corresponding to
(<code>'f4'</code>,<code>'f8'</code>,<code>'i2'</code>,<code>'i2'</code>,<code>'i1'</code>,<code>'i1'</code>,<code>'S1'</code>,<code>'i4'</code>,<code>'i4'</code>),
will also work. The unsigned integer types and the 64-bit integer
type can only be used if the file format is <code>NETCDF4</code>.</p>
<p>The dimensions themselves are usually also defined as variables,
called coordinate variables. The <a
href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a> method returns an instance of the <a
href="netCDF4.Variable-class.html" class="link">Variable</a> class
whose methods can be used later to access and set variable data and
attributes.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>times = rootgrp.createVariable(<span class="py-string">'time'</span>,<span class="py-string">'f8'</span>,(<span class="py-string">'time'</span>,))
<span class="py-prompt">>>> </span>levels = rootgrp.createVariable(<span class="py-string">'level'</span>,<span class="py-string">'i4'</span>,(<span class="py-string">'level'</span>,))
<span class="py-prompt">>>> </span>latitudes = rootgrp.createVariable(<span class="py-string">'latitude'</span>,<span class="py-string">'f4'</span>,(<span class="py-string">'lat'</span>,))
<span class="py-prompt">>>> </span>longitudes = rootgrp.createVariable(<span class="py-string">'longitude'</span>,<span class="py-string">'f4'</span>,(<span class="py-string">'lon'</span>,))
<span class="py-prompt">>>> </span><span class="py-comment"># two dimensions unlimited.</span>
<span class="py-prompt">>>> </span>temp = rootgrp.createVariable(<span class="py-string">'temp'</span>,<span class="py-string">'f4'</span>,(<span class="py-string">'time'</span>,<span class="py-string">'level'</span>,<span class="py-string">'lat'</span>,<span class="py-string">'lon'</span>,))</pre>
<p>All of the variables in the <a href="netCDF4.Dataset-class.html"
class="link">Dataset</a> or <a href="netCDF4.Group-class.html"
class="link">Group</a> are stored in a Python dictionary, in the same
way as the dimensions:</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp.variables
<span class="py-output">OrderedDict([('time', <netCDF4.Variable object at 0x1b4ba70>),</span>
<span class="py-output"> ('level', <netCDF4.Variable object at 0x1b4bab0>), </span>
<span class="py-output"> ('latitude', <netCDF4.Variable object at 0x1b4baf0>),</span>
<span class="py-output"> ('longitude', <netCDF4.Variable object at 0x1b4bb30>),</span>
<span class="py-output"> ('temp', <netCDF4.Variable object at 0x1b4bb70>)])</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>To get summary info on a <a href="netCDF4.Variable-class.html"
class="link">Variable</a> instance in an interactive session, just
print it.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp.variables[<span class="py-string">'temp'</span>]
<span class="py-output"><type 'netCDF4.Variable'></span>
<span class="py-output">float32 temp(time, level, lat, lon)</span>
<span class="py-output"> least_significant_digit: 3</span>
<span class="py-output"> units: K</span>
<span class="py-output">unlimited dimensions: time, level</span>
<span class="py-output">current shape = (0, 0, 73, 144)</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p><a href="netCDF4.Variable-class.html" class="link">Variable</a>
names can be changed using the <a
href="netCDF4.Dataset-class.html#renameVariable"
class="link">renameVariable</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>
instance.</p>
<h2 class="heading">5) Attributes in a netCDF file</h2>
<p>There are two types of attributes in a netCDF file, global and
variable. Global attributes provide information about a group, or the
entire dataset, as a whole. <a href="netCDF4.Variable-class.html"
class="link">Variable</a> attributes provide information about one of
the variables in a group. Global attributes are set by assigning
values to <a href="netCDF4.Dataset-class.html"
class="link">Dataset</a> or <a href="netCDF4.Group-class.html"
class="link">Group</a> instance variables. <a
href="netCDF4.Variable-class.html" class="link">Variable</a>
attributes are set by assigning values to <a
href="netCDF4.Variable-class.html" class="link">Variable</a>
instances variables. Attributes can be strings, numbers or sequences.
Returning to our example,</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">import</span> time
<span class="py-prompt">>>> </span>rootgrp.description = <span class="py-string">'bogus example script'</span>
<span class="py-prompt">>>> </span>rootgrp.history = <span class="py-string">'Created '</span> + time.ctime(time.time())
<span class="py-prompt">>>> </span>rootgrp.source = <span class="py-string">'netCDF4 python module tutorial'</span>
<span class="py-prompt">>>> </span>latitudes.units = <span class="py-string">'degrees north'</span>
<span class="py-prompt">>>> </span>longitudes.units = <span class="py-string">'degrees east'</span>
<span class="py-prompt">>>> </span>levels.units = <span class="py-string">'hPa'</span>
<span class="py-prompt">>>> </span>temp.units = <span class="py-string">'K'</span>
<span class="py-prompt">>>> </span>times.units = <span class="py-string">'hours since 0001-01-01 00:00:00.0'</span>
<span class="py-prompt">>>> </span>times.calendar = <span class="py-string">'gregorian'</span></pre>
<p>The <a href="netCDF4.Dataset-class.html#ncattrs"
class="link">ncattrs</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>, <a
href="netCDF4.Group-class.html" class="link">Group</a> or <a
href="netCDF4.Variable-class.html" class="link">Variable</a> instance
can be used to retrieve the names of all the netCDF attributes. This
method is provided as a convenience, since using the built-in
<code>dir</code> Python function will return a bunch of private
methods and attributes that cannot (or should not) be modified by the
user.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">for</span> name <span class="py-keyword">in</span> rootgrp.ncattrs():
<span class="py-prompt">>>> </span> <span class="py-keyword">print</span> <span class="py-string">'Global attr'</span>, name, <span class="py-string">'='</span>, getattr(rootgrp,name)
<span class="py-output">Global attr description = bogus example script</span>
<span class="py-output">Global attr history = Created Mon Nov 7 10.30:56 2005</span>
<span class="py-output">Global attr source = netCDF4 python module tutorial</span></pre>
<p>The <code>__dict__</code> attribute of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>, <a
href="netCDF4.Group-class.html" class="link">Group</a> or <a
href="netCDF4.Variable-class.html" class="link">Variable</a> instance
provides all the netCDF attribute name/value pairs in a python
dictionary:</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> rootgrp.__dict__
<span class="py-output">OrderedDict([(u'description', u'bogus example script'),</span>
<span class="py-output"> (u'history', u'Created Thu Mar 3 19:30:33 2011'), </span>
<span class="py-output"> (u'source', u'netCDF4 python module tutorial')])</span></pre>
<p>Attributes can be deleted from a netCDF <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a>, <a
href="netCDF4.Group-class.html" class="link">Group</a> or <a
href="netCDF4.Variable-class.html" class="link">Variable</a> using
the python <code>del</code> statement (i.e. <code>del grp.foo</code>
removes the attribute <code>foo</code> the the group
<code>grp</code>).</p>
<h2 class="heading">6) Writing data to and retrieving data from a netCDF variable</h2>
<p>Now that you have a netCDF <a href="netCDF4.Variable-class.html"
class="link">Variable</a> instance, how do you put data into it? You
can just treat it like an array and assign data to a slice.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">import</span> numpy
<span class="py-prompt">>>> </span>lats = numpy.arange(-90,91,2.5)
<span class="py-prompt">>>> </span>lons = numpy.arange(-180,180,2.5)
<span class="py-prompt">>>> </span>latitudes[:] = lats
<span class="py-prompt">>>> </span>longitudes[:] = lons
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'latitudes =\n'</span>,latitudes[:]
<span class="py-output">latitudes =</span>
<span class="py-output">[-90. -87.5 -85. -82.5 -80. -77.5 -75. -72.5 -70. -67.5 -65. -62.5</span>
<span class="py-output"> -60. -57.5 -55. -52.5 -50. -47.5 -45. -42.5 -40. -37.5 -35. -32.5</span>
<span class="py-output"> -30. -27.5 -25. -22.5 -20. -17.5 -15. -12.5 -10. -7.5 -5. -2.5</span>
<span class="py-output"> 0. 2.5 5. 7.5 10. 12.5 15. 17.5 20. 22.5 25. 27.5</span>
<span class="py-output"> 30. 32.5 35. 37.5 40. 42.5 45. 47.5 50. 52.5 55. 57.5</span>
<span class="py-output"> 60. 62.5 65. 67.5 70. 72.5 75. 77.5 80. 82.5 85. 87.5</span>
<span class="py-output"> 90. ]</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Unlike NumPy's array objects, netCDF <a
href="netCDF4.Variable-class.html" class="link">Variable</a> objects
with unlimited dimensions will grow along those dimensions if you
assign data outside the currently defined range of indices.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-comment"># append along two unlimited dimensions by assigning to slice.</span>
<span class="py-prompt">>>> </span>nlats = len(rootgrp.dimensions[<span class="py-string">'lat'</span>])
<span class="py-prompt">>>> </span>nlons = len(rootgrp.dimensions[<span class="py-string">'lon'</span>])
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'temp shape before adding data = '</span>,temp.shape
<span class="py-output">temp shape before adding data = (0, 0, 73, 144)</span>
<span class="py-output"></span><span class="py-prompt">>>></span>
<span class="py-prompt">>>> </span><span class="py-keyword">from</span> numpy.random <span class="py-keyword">import</span> uniform
<span class="py-prompt">>>> </span>temp[0:5,0:10,:,:] = uniform(size=(5,10,nlats,nlons))
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'temp shape after adding data = '</span>,temp.shape
<span class="py-output">temp shape after adding data = (6, 10, 73, 144)</span>
<span class="py-output"></span><span class="py-prompt">>>></span>
<span class="py-prompt">>>> </span><span class="py-comment"># levels have grown, but no values yet assigned.</span>
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'levels shape after adding pressure data = '</span>,levels.shape
<span class="py-output">levels shape after adding pressure data = (10,)</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Note that the size of the levels variable grows when data is
appended along the <code>level</code> dimension of the variable
<code>temp</code>, even though no data has yet been assigned to
levels.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-comment"># now, assign data to levels dimension variable.</span>
<span class="py-prompt">>>> </span>levels[:] = [1000.,850.,700.,500.,300.,250.,200.,150.,100.,50.]</pre>
<p>However, that there are some differences between NumPy and netCDF
variable slicing rules. Slices behave as usual, being specified as a
<code>start:stop:step</code> triplet. Using a scalar integer index
<code>i</code> takes the ith element and reduces the rank of the
output array by one. Boolean array and integer sequence indexing
behaves differently for netCDF variables than for numpy arrays. Only
1-d boolean arrays and integer sequences are allowed, and these
indices work independently along each dimension (similar to the way
vector subscripts work in fortran). This means that</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>temp[0, 0, [0,1,2,3], [0,1,2,3]]</pre>
<p>returns an array of shape (4,4) when slicing a netCDF variable,
but for a numpy array it returns an array of shape (4,). Similarly, a
netCDF variable of shape <code>(2,3,4,5)</code> indexed with
<code>[0, array([True, False, True]), array([False, True, True,
True]), :]</code> would return a <code>(2, 3, 5)</code> array. In
NumPy, this would raise an error since it would be equivalent to
<code>[0, [0,1], [1,2,3], :]</code>. When slicing with integer
sequences, the indices must be sorted in increasing order and contain
no duplicates. While this behaviour may cause some confusion for
those used to NumPy's 'fancy indexing' rules, it provides a very
powerful way to extract data from multidimensional netCDF variables
by using logical operations on the dimension arrays to create
slices.</p>
<p>For example,</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>tempdat = temp[::2, [1,3,6], lats>0, lons>0]</pre>
<p>will extract time indices 0,2 and 4, pressure levels 850, 500 and
200 hPa, all Northern Hemisphere latitudes and Eastern Hemisphere
longitudes, resulting in a numpy array of shape (3, 3, 36, 71).</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'shape of fancy temp slice = '</span>,tempdat.shape
<span class="py-output">shape of fancy temp slice = (3, 3, 36, 71)</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>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 <code>hours since YY:MM:DD hh-mm-ss</code>.
These units can be awkward to deal with, without a utility to convert
the values to and from calendar dates. The functione called <a
href="netCDF4-module.html#num2date" class="link">num2date</a> and <a
href="netCDF4-module.html#date2num" class="link">date2num</a> are
provided with this package to do just that. Here's an example of how
they can be used:</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-comment"># fill in times.</span>
<span class="py-prompt">>>> </span><span class="py-keyword">from</span> datetime <span class="py-keyword">import</span> datetime, timedelta
<span class="py-prompt">>>> </span><span class="py-keyword">from</span> netCDF4 <span class="py-keyword">import</span> num2date, date2num
<span class="py-prompt">>>> </span>dates = [datetime(2001,3,1)+n*timedelta(hours=12) <span class="py-keyword">for</span> n <span class="py-keyword">in</span> range(temp.shape[0])]
<span class="py-prompt">>>> </span>times[:] = date2num(dates,units=times.units,calendar=times.calendar)
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'time values (in units %s): '</span> % times.units+<span class="py-string">'\n'</span>,times[:]
<span class="py-output">time values (in units hours since January 1, 0001): </span>
<span class="py-output">[ 17533056. 17533068. 17533080. 17533092. 17533104.]</span>
<span class="py-output"></span><span class="py-prompt">>>></span>
<span class="py-prompt">>>> </span>dates = num2date(times[:],units=times.units,calendar=times.calendar)
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'dates corresponding to time values:\n'</span>,dates
<span class="py-output">dates corresponding to time values:</span>
<span class="py-output">[2001-03-01 00:00:00 2001-03-01 12:00:00 2001-03-02 00:00:00</span>
<span class="py-output"> 2001-03-02 12:00:00 2001-03-03 00:00:00]</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p><a href="netCDF4-module.html#num2date" class="link">num2date</a>
converts numeric values of time in the specified <code>units</code>
and <code>calendar</code> to datetime objects, and <a
href="netCDF4-module.html#date2num" class="link">date2num</a> does
the reverse. All the calendars currently defined in the <a
href="http://cf-pcmdi.llnl.gov/documents/cf-conventions/"
target="_top">CF metadata convention</a> are supported. A function
called <a href="netCDF4-module.html#date2index"
class="link">date2index</a> is also provided which returns the
indices of a netCDF time variable corresponding to a sequence of
datetime instances.</p>
<h2 class="heading">7) Reading data from a multi-file netCDF dataset.</h2>
<p>If you want to read data from a variable that spans multiple
netCDF files, you can use the <a href="netCDF4.MFDataset-class.html"
class="link">MFDataset</a> class to read the data as if it were
contained in a single file. Instead of using a single filename to
create a <a href="netCDF4.Dataset-class.html"
class="link">Dataset</a> instance, create a <a
href="netCDF4.MFDataset-class.html" class="link">MFDataset</a>
instance with either a list of filenames, or a string with a wildcard
(which is then converted to a sorted list of files using the python
glob module). Variables in the list of files that share the same
unlimited dimension are aggregated together, and can be sliced across
multiple files. To illustrate this, let's first create a bunch of
netCDF files with the same variable (with the same unlimited
dimension). The files must in be in <code>NETCDF3_64BIT</code>,
<code>NETCDF3_CLASSIC</code> or <code>NETCDF4_CLASSIC format</code>
(<code>NETCDF4</code> formatted multi-file datasets are not
supported).</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">for</span> nfile <span class="py-keyword">in</span> range(10):
<span class="py-prompt">>>> </span> f = Dataset(<span class="py-string">'mftest'</span>+repr(nfile)+<span class="py-string">'.nc'</span>,<span class="py-string">'w'</span>,format=<span class="py-string">'NETCDF4_CLASSIC'</span>)
<span class="py-prompt">>>> </span> f.createDimension(<span class="py-string">'x'</span>,None)
<span class="py-prompt">>>> </span> x = f.createVariable(<span class="py-string">'x'</span>,<span class="py-string">'i'</span>,(<span class="py-string">'x'</span>,))
<span class="py-prompt">>>> </span> x[0:10] = numpy.arange(nfile*10,10*(nfile+1))
<span class="py-prompt">>>> </span> f.close()</pre>
<p>Now read all the files back in at once with <a
href="netCDF4.MFDataset-class.html" class="link">MFDataset</a></p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">from</span> netCDF4 <span class="py-keyword">import</span> MFDataset
<span class="py-prompt">>>> </span>f = MFDataset(<span class="py-string">'mftest*nc'</span>)
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.variables[<span class="py-string">'x'</span>][:]
<span class="py-output">[ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24</span>
<span class="py-output"> 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49</span>
<span class="py-output"> 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74</span>
<span class="py-output"> 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99]</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Note that MFDataset can only be used to read, not write,
multi-file datasets.</p>
<h2 class="heading">8) Efficient compression of netCDF variables</h2>
<p>Data stored in netCDF 4 <a href="netCDF4.Variable-class.html"
class="link">Variable</a> objects can be compressed and decompressed
on the fly. The parameters for the compression are determined by the
<code>zlib</code>, <code>complevel</code> and <code>shuffle</code>
keyword arguments to the <a
href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a> method. To turn on compression, set
<code>zlib=True</code>. The <code>complevel</code> keyword regulates
the speed and efficiency of the compression (1 being fastest, but
lowest compression ratio, 9 being slowest but best compression
ratio). The default value of <code>complevel</code> is 4. Setting
<code>shuffle=False</code> will turn off the HDF5 shuffle filter,
which de-interlaces a block of data before compression by reordering
the bytes. The shuffle filter can significantly improve compression
ratios, and is on by default. Setting <code>fletcher32</code>
keyword argument to <a
href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a> to <code>True</code> (it's
<code>False</code> by default) enables the Fletcher32 checksum
algorithm for error detection. It's also possible to set the HDF5
chunking parameters and endian-ness of the binary data stored in the
HDF5 file with the <code>chunksizes</code> and <code>endian</code>
keyword arguments to <a
href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a>. These keyword arguments only are
relevant for <code>NETCDF4</code> and <code>NETCDF4_CLASSIC</code>
files (where the underlying file format is HDF5) and are silently
ignored if the file format is <code>NETCDF3_CLASSIC</code> or
<code>NETCDF3_64BIT</code>,</p>
<p>If your data only has a certain number of digits of precision (say
for example, it is temperature data that was measured with a
precision of 0.1 degrees), you can dramatically improve zlib
compression by quantizing (or truncating) the data using the
<code>least_significant_digit</code> keyword argument to <a
href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a>. The least significant digit is the
power of ten of the smallest decimal place in the data that is a
reliable value. For example if the data has a precision of 0.1, then
setting <code>least_significant_digit=1</code> will cause data the
data to be quantized using
<code>numpy.around(scale*data)/scale</code>, where scale = 2**bits,
and bits is determined so that a precision of 0.1 is retained (in
this case bits=4). Effectively, this makes the compression 'lossy'
instead of 'lossless', that is some precision in the data is
sacrificed for the sake of disk space.</p>
<p>In our example, try replacing the line</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>temp = rootgrp.createVariable(<span class="py-string">'temp'</span>,<span class="py-string">'f4'</span>,(<span class="py-string">'time'</span>,<span class="py-string">'level'</span>,<span class="py-string">'lat'</span>,<span class="py-string">'lon'</span>,))</pre>
<p>with</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>temp = dataset.createVariable(<span class="py-string">'temp'</span>,<span class="py-string">'f4'</span>,(<span class="py-string">'time'</span>,<span class="py-string">'level'</span>,<span class="py-string">'lat'</span>,<span class="py-string">'lon'</span>,),zlib=True)</pre>
<p>and then</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>temp = dataset.createVariable(<span class="py-string">'temp'</span>,<span class="py-string">'f4'</span>,(<span class="py-string">'time'</span>,<span class="py-string">'level'</span>,<span class="py-string">'lat'</span>,<span class="py-string">'lon'</span>,),zlib=True,least_significant_digit=3)</pre>
<p>and see how much smaller the resulting files are.</p>
<h2 class="heading">9) Beyond homogenous arrays of a fixed type - compound data types</h2>
<p>Compound data types map directly to numpy structured (a.k.a
'record' arrays). Structured arrays are akin to C structs, or
derived types in Fortran. They allow for the construction of
table-like structures composed of combinations of other data types,
including other compound types. Compound types might be useful for
representing multiple parameter values at each point on a grid, or at
each time and space location for scattered (point) data. You can then
access all the information for a point by reading one variable,
instead of reading different parameters from different variables.
Compound data types are created from the corresponding numpy data
type using the <a
href="netCDF4.Dataset-class.html#createCompoundType"
class="link">createCompoundType</a> method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> instance.
Since there is no native complex data type in netcdf, compound types
are handy for storing numpy complex arrays. Here's an example:</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>f = Dataset(<span class="py-string">'complex.nc'</span>,<span class="py-string">'w'</span>)
<span class="py-prompt">>>> </span>size = 3 <span class="py-comment"># length of 1-d complex array</span>
<span class="py-prompt">>>> </span><span class="py-comment"># create sample complex data.</span>
<span class="py-prompt">>>> </span>datac = numpy.exp(1j*(1.+numpy.linspace(0, numpy.pi, size)))
<span class="py-prompt">>>> </span><span class="py-comment"># create complex128 compound data type.</span>
<span class="py-prompt">>>> </span>complex128 = numpy.dtype([(<span class="py-string">'real'</span>,numpy.float64),(<span class="py-string">'imag'</span>,numpy.float64)])
<span class="py-prompt">>>> </span>complex128_t = f.createCompoundType(complex128,<span class="py-string">'complex128'</span>)
<span class="py-prompt">>>> </span><span class="py-comment"># create a variable with this data type, write some data to it.</span>
<span class="py-prompt">>>> </span>f.createDimension(<span class="py-string">'x_dim'</span>,None)
<span class="py-prompt">>>> </span>v = f.createVariable(<span class="py-string">'cmplx_var'</span>,complex128_t,<span class="py-string">'x_dim'</span>)
<span class="py-prompt">>>> </span>data = numpy.empty(size,complex128) <span class="py-comment"># numpy structured array</span>
<span class="py-prompt">>>> </span>data[<span class="py-string">'real'</span>] = datac.real; data[<span class="py-string">'imag'</span>] = datac.imag
<span class="py-prompt">>>> </span>v[:] = data <span class="py-comment"># write numpy structured array to netcdf compound var</span>
<span class="py-prompt">>>> </span><span class="py-comment"># close and reopen the file, check the contents.</span>
<span class="py-prompt">>>> </span>f.close(); f = Dataset(<span class="py-string">'complex.nc'</span>)
<span class="py-prompt">>>> </span>v = f.variables[<span class="py-string">'cmplx_var'</span>]
<span class="py-prompt">>>> </span>datain = v[:] <span class="py-comment"># read in all the data into a numpy structured array</span>
<span class="py-prompt">>>> </span><span class="py-comment"># create an empty numpy complex array</span>
<span class="py-prompt">>>> </span>datac2 = numpy.empty(datain.shape,numpy.complex128)
<span class="py-prompt">>>> </span><span class="py-comment"># .. fill it with contents of structured array.</span>
<span class="py-prompt">>>> </span>datac2.real = datain[<span class="py-string">'real'</span>]; datac2.imag = datain[<span class="py-string">'imag'</span>]
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> datac.dtype,datac <span class="py-comment"># original data</span>
<span class="py-output">complex128 [ 0.54030231+0.84147098j -0.84147098+0.54030231j -0.54030231-0.84147098j]</span>
<span class="py-output"></span><span class="py-prompt">>>></span>
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> datac2.dtype,datac2 <span class="py-comment"># data from file</span>
<span class="py-output">complex128 [ 0.54030231+0.84147098j -0.84147098+0.54030231j -0.54030231-0.84147098j]</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Compound types can be nested, but you must create the 'inner' ones
first. All of the compound types defined for a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> are stored in
a Python dictionary, just like variables and dimensions. As always,
printing objects gives useful summary information in an interactive
session:</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> f
<span class="py-output"><type 'netCDF4.Dataset'></span>
<span class="py-output">root group (NETCDF4 file format):</span>
<span class="py-output"> dimensions: x_dim</span>
<span class="py-output"> variables: cmplx_var</span>
<span class="py-output"> groups:</span>
<span class="py-output"><type 'netCDF4.Variable'></span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.variables[<span class="py-string">'cmplx_var'</span>]
<span class="py-output">compound cmplx_var(x_dim)</span>
<span class="py-output">compound data type: [('real', '<f8'), ('imag', '<f8')]</span>
<span class="py-output">unlimited dimensions: x_dim</span>
<span class="py-output">current shape = (3,)</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.cmptypes
<span class="py-output">OrderedDict([('complex128', <netCDF4.CompoundType object at 0x1029eb7e8>)])</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.cmptypes[<span class="py-string">'complex128'</span>]
<span class="py-output"><type 'netCDF4.CompoundType'>: name = 'complex128', numpy dtype = [(u'real','<f8'), (u'imag', '<f8')]</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<h2 class="heading">10) Variable-length (vlen) data types.</h2>
<p>NetCDF 4 has support for variable-length or "ragged"
arrays. These are arrays of variable length sequences having the
same type. To create a variable-length data type, use the <a
href="netCDF4.Dataset-class.html#createVLType"
class="link">createVLType</a> method method of a <a
href="netCDF4.Dataset-class.html" class="link">Dataset</a> or <a
href="netCDF4.Group-class.html" class="link">Group</a> instance.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>f = Dataset(<span class="py-string">'tst_vlen.nc'</span>,<span class="py-string">'w'</span>)
<span class="py-prompt">>>> </span>vlen_t = f.createVLType(numpy.int32, <span class="py-string">'phony_vlen'</span>)</pre>
<p>The numpy datatype of the variable-length sequences and the name
of the new datatype must be specified. Any of the primitive datatypes
can be used (signed and unsigned integers, 32 and 64 bit floats, and
characters), but compound data types cannot. A new variable can then
be created using this datatype.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>x = f.createDimension(<span class="py-string">'x'</span>,3)
<span class="py-prompt">>>> </span>y = f.createDimension(<span class="py-string">'y'</span>,4)
<span class="py-prompt">>>> </span>vlvar = f.createVariable(<span class="py-string">'phony_vlen_var'</span>, vlen_t, (<span class="py-string">'y'</span>,<span class="py-string">'x'</span>))</pre>
<p>Since there is no native vlen datatype in numpy, vlen arrays are
represented in python as object arrays (arrays of dtype
<code>object</code>). These are arrays whose elements are Python
object pointers, and can contain any type of python object. For this
application, they must contain 1-D numpy arrays all of the same type
but of varying length. In this case, they contain 1-D numpy
<code>int32</code> arrays of random length betwee 1 and 10.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span><span class="py-keyword">import</span> random
<span class="py-prompt">>>> </span>data = numpy.empty(len(y)*len(x),object)
<span class="py-prompt">>>> </span><span class="py-keyword">for</span> n <span class="py-keyword">in</span> range(len(y)*len(x)):
<span class="py-prompt">>>> </span> data[n] = numpy.arange(random.randint(1,10),dtype=<span class="py-string">'int32'</span>)+1
<span class="py-prompt">>>> </span>data = numpy.reshape(data,(len(y),len(x)))
<span class="py-prompt">>>> </span>vlvar[:] = data
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'vlen variable =\n'</span>,vlvar[:]
<span class="py-output">vlen variable =</span>
<span class="py-output">[[[ 1 2 3 4 5 6 7 8 9 10] [1 2 3 4 5] [1 2 3 4 5 6 7 8]]</span>
<span class="py-output"> [[1 2 3 4 5 6 7] [1 2 3 4 5 6] [1 2 3 4 5]]</span>
<span class="py-output"> [[1 2 3 4 5] [1 2 3 4] [1]]</span>
<span class="py-output"> [[ 1 2 3 4 5 6 7 8 9 10] [ 1 2 3 4 5 6 7 8 9 10]</span>
<span class="py-output"> [1 2 3 4 5 6 7 8]]]</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f
<span class="py-output"><type 'netCDF4.Dataset'></span>
<span class="py-output">root group (NETCDF4 file format):</span>
<span class="py-output"> dimensions: x, y</span>
<span class="py-output"> variables: phony_vlen_var</span>
<span class="py-output"> groups:</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.variables[<span class="py-string">'phony_vlen_var'</span>]
<span class="py-output"><type 'netCDF4.Variable'></span>
<span class="py-output">vlen phony_vlen_var(y, x)</span>
<span class="py-output">vlen data type: int32</span>
<span class="py-output">unlimited dimensions:</span>
<span class="py-output">current shape = (4, 3)</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.VLtypes[<span class="py-string">'phony_vlen'</span>]
<span class="py-output"><type 'netCDF4.VLType'>: name = 'phony_vlen', numpy dtype = int32</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>Numpy object arrays containing python strings can also be written
as vlen variables, For vlen strings, you don't need to create a vlen
data type. Instead, simply use the python <code>str</code> builtin
(or a numpy string datatype with fixed length greater than 1) when
calling the <a href="netCDF4.Dataset-class.html#createVariable"
class="link">createVariable</a> method.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>z = f.createDimension(<span class="py-string">'z'</span>,10)
<span class="py-prompt">>>> </span>strvar = rootgrp.createVariable(<span class="py-string">'strvar'</span>, str, <span class="py-string">'z'</span>)</pre>
<p>In this example, an object array is filled with random python
strings with random lengths between 2 and 12 characters, and the data
in the object array is assigned to the vlen string variable.</p>
<pre class="py-doctest">
<span class="py-prompt">>>> </span>chars = <span class="py-string">'1234567890aabcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'</span>
<span class="py-prompt">>>> </span>data = numpy.empty(10,<span class="py-string">'O'</span>)
<span class="py-prompt">>>> </span><span class="py-keyword">for</span> n <span class="py-keyword">in</span> range(10):
<span class="py-prompt">>>> </span> stringlen = random.randint(2,12)
<span class="py-prompt">>>> </span> data[n] = <span class="py-string">''</span>.join([random.choice(chars) <span class="py-keyword">for</span> i <span class="py-keyword">in</span> range(stringlen)])
<span class="py-prompt">>>> </span>strvar[:] = data
<span class="py-prompt">>>> </span><span class="py-keyword">print</span> <span class="py-string">'variable-length string variable:\n'</span>,strvar[:]
<span class="py-output">variable-length string variable:</span>
<span class="py-output">[aDy29jPt jd7aplD b8t4RM jHh8hq KtaPWF9cQj Q1hHN5WoXSiT MMxsVeq td LUzvVTzj</span>
<span class="py-output"> 5DS9X8S]</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f
<span class="py-output"><type 'netCDF4.Dataset'></span>
<span class="py-output">root group (NETCDF4 file format):</span>
<span class="py-output"> dimensions: x, y, z</span>
<span class="py-output"> variables: phony_vlen_var, strvar</span>
<span class="py-output"> groups:</span>
<span class="py-output"></span><span class="py-prompt">>>> </span><span class="py-keyword">print</span> f.variables[<span class="py-string">'strvar'</span>]
<span class="py-output"><type 'netCDF4.Variable'></span>
<span class="py-output">vlen strvar(z)</span>
<span class="py-output">vlen data type: <type 'str'></span>
<span class="py-output">unlimited dimensions:</span>
<span class="py-output">current size = (10,)</span>
<span class="py-output"></span><span class="py-prompt">>>></span></pre>
<p>It is also possible to set contents of vlen string variables with
numpy arrays of any string or unicode data type. Note, however, that
accessing the contents of such variables will always return numpy
arrays with dtype <code>object</code>.</p>
<p>All of the code in this tutorial is available in
<code>examples/tutorial.py</code>, Unit tests are in the
<code>test</code> directory.</p>
<hr />
<div class="fields"> <p><strong>Contact:</strong>
Jeffrey Whitaker <jeffrey.s.whitaker@noaa.gov>
</p>
<p><strong>Copyright:</strong>
2008 by Jeffrey Whitaker.
</p>
<p><strong>License:</strong>
Permission to use, copy, modify, and distribute this software and
its documentation for any purpose and without fee is hereby
granted, provided that the above copyright notice appear in all
copies and that both the copyright notice and this permission
notice appear in supporting documentation. THE AUTHOR DISCLAIMS ALL
WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
AUTHOR BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES
OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
PERFORMANCE OF THIS SOFTWARE.
</p>
<p><strong>Version:</strong>
1.1.7
</p>
</div><!-- ==================== CLASSES ==================== -->
<a name="section-Classes"></a>
<table class="summary" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr bgcolor="#70b0f0" class="table-header">
<td align="left" colspan="2" class="table-header">
<span class="table-header">Classes</span></td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.CompoundType-class.html" class="summary-name">CompoundType</a><br />
A <a href="netCDF4.CompoundType-class.html"
class="link">CompoundType</a> instance is used to describe a
compound data type.
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.Dataset-class.html" class="summary-name">Dataset</a><br />
Dataset(self, filename, mode="r", clobber=True,
diskless=False, persist=False, keepweakref=False, format='NETCDF4')
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.Dimension-class.html" class="summary-name">Dimension</a><br />
Dimension(self, group, name, size=None)
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.Group-class.html" class="summary-name">Group</a><br />
Group(self, parent, name)
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.MFDataset-class.html" class="summary-name">MFDataset</a><br />
MFDataset(self, files, check=False, aggdim=None, exclude=[])
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.MFTime-class.html" class="summary-name">MFTime</a><br />
MFTime(self, time, units=None)
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.VLType-class.html" class="summary-name">VLType</a><br />
A <a href="netCDF4.VLType-class.html" class="link">VLType</a>
instance is used to describe a variable length (VLEN) data type.
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4.Variable-class.html" class="summary-name">Variable</a><br />
Variable(self, group, name, datatype, dimensions=(), zlib=False,
complevel=4, shuffle=True, fletcher32=False, contiguous=False,
chunksizes=None, endian='native',
least_significant_digit=None,fill_value=None)
</td>
</tr>
</table>
<!-- ==================== FUNCTIONS ==================== -->
<a name="section-Functions"></a>
<table class="summary" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr bgcolor="#70b0f0" class="table-header">
<td align="left" colspan="2" class="table-header">
<span class="table-header">Functions</span></td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a href="netCDF4-module.html#chartostring" class="summary-sig-name">chartostring</a>(<span class="summary-sig-arg">b</span>)</span><br />
convert a character array to a string array with one less dimension.</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a href="netCDF4-module.html#date2index" class="summary-sig-name">date2index</a>(<span class="summary-sig-arg">dates</span>,
<span class="summary-sig-arg">nctime</span>,
<span class="summary-sig-arg">calendar</span>=<span class="summary-sig-default">None</span>,
<span class="summary-sig-arg">select</span>=<span class="summary-sig-default">'exact'</span>)</span><br />
Return indices of a netCDF time variable corresponding to the given
dates.</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a href="netCDF4-module.html#date2num" class="summary-sig-name">date2num</a>(<span class="summary-sig-arg">dates</span>,
<span class="summary-sig-arg">units</span>,
<span class="summary-sig-arg">calendar</span>=<span class="summary-sig-default">'standard'</span>)</span><br />
Return numeric time values given datetime objects.</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a name="getlibversion"></a><span class="summary-sig-name">getlibversion</span>()</span><br />
returns a string describing the version of the netcdf library used to
build the module, and when it was built.</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a href="netCDF4-module.html#num2date" class="summary-sig-name">num2date</a>(<span class="summary-sig-arg">times</span>,
<span class="summary-sig-arg">units</span>,
<span class="summary-sig-arg">calendar</span>=<span class="summary-sig-default">'standard'</span>)</span><br />
Return datetime objects given numeric time values.</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a href="netCDF4-module.html#stringtoarr" class="summary-sig-name">stringtoarr</a>(<span class="summary-sig-arg">a</span>,
<span class="summary-sig-arg">NUMCHARS</span>,
<span class="summary-sig-arg">dtype</span>=<span class="summary-sig-default">'S'</span>)</span><br />
convert a string to a character array of length NUMCHARS</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr>
<td><span class="summary-sig"><a href="netCDF4-module.html#stringtochar" class="summary-sig-name">stringtochar</a>(<span class="summary-sig-arg">a</span>)</span><br />
convert a string array to a character array with one extra dimension</td>
<td align="right" valign="top">
</td>
</tr>
</table>
</td>
</tr>
</table>
<!-- ==================== VARIABLES ==================== -->
<a name="section-Variables"></a>
<table class="summary" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr bgcolor="#70b0f0" class="table-header">
<td align="left" colspan="2" class="table-header">
<span class="table-header">Variables</span></td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="MINYEAR"></a><span class="summary-name">MINYEAR</span> = <code title="1">1</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="NC_DISKLESS"></a><span class="summary-name">NC_DISKLESS</span> = <code title="8">8</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="__has_nc_inq_format_extended__"></a><span class="summary-name">__has_nc_inq_format_extended__</span> = <code title="1">1</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="__has_nc_inq_path__"></a><span class="summary-name">__has_nc_inq_path__</span> = <code title="1">1</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="__has_rename_grp__"></a><span class="summary-name">__has_rename_grp__</span> = <code title="1">1</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="__hdf5libversion__"></a><span class="summary-name">__hdf5libversion__</span> = <code title="'1.8.14'"><code class="variable-quote">'</code><code class="variable-string">1.8.14</code><code class="variable-quote">'</code></code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="__netcdf4libversion__"></a><span class="summary-name">__netcdf4libversion__</span> = <code title="u'4.3.3-rc2'"><code class="variable-quote">u'</code><code class="variable-string">4.3.3-rc2</code><code class="variable-quote">'</code></code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="__package__"></a><span class="summary-name">__package__</span> = <code title="None">None</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="default_encoding"></a><span class="summary-name">default_encoding</span> = <code title="'utf-8'"><code class="variable-quote">'</code><code class="variable-string">utf-8</code><code class="variable-quote">'</code></code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a href="netCDF4-module.html#default_fillvals" class="summary-name">default_fillvals</a> = <code title="{'S1': '\x00',
'U1': '\x00',
'f4': 9.96920996839e+36,
'f8': 9.96920996839e+36,
'i1': -127,
'i2': -32767,
'i4': -2147483647,
'i8': -9223372036854775806,
..."><code class="variable-group">{</code><code class="variable-quote">'</code><code class="variable-string">S1</code><code class="variable-quote">'</code><code class="variable-op">: </code><code class="variable-quote">'</code><code class="variable-string">\x00</code><code class="variable-quote">'</code><code class="variable-op">, </code><code class="variable-quote">'</code><code class="variable-string">U1</code><code class="variable-quote">'</code><code class="variable-op">: </code><code class="variable-quote">'</code><code class="variable-string">\x00</code><code class="variable-quote">'</code><code class="variable-op">, </code><code class="variable-quote">'</code><code class="variable-string">f4</code><code class="variable-quote">'</code><code class="variable-op">: </code>9.969209<code class="variable-ellipsis">...</code></code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="gregorian"></a><span class="summary-name">gregorian</span> = <code title="datetime.datetime(1582, 10, 15, 0, 0)">datetime.datetime(1582, 10, 15, 0, 0)</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="is_native_big"></a><span class="summary-name">is_native_big</span> = <code title="False">False</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="is_native_little"></a><span class="summary-name">is_native_little</span> = <code title="True">True</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="python3"></a><span class="summary-name">python3</span> = <code title="False">False</code>
</td>
</tr>
<tr>
<td width="15%" align="right" valign="top" class="summary">
<span class="summary-type"> </span>
</td><td class="summary">
<a name="unicode_error"></a><span class="summary-name">unicode_error</span> = <code title="'replace'"><code class="variable-quote">'</code><code class="variable-string">replace</code><code class="variable-quote">'</code></code>
</td>
</tr>
</table>
<!-- ==================== FUNCTION DETAILS ==================== -->
<a name="section-FunctionDetails"></a>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr bgcolor="#70b0f0" class="table-header">
<td align="left" colspan="2" class="table-header">
<span class="table-header">Function Details</span></td>
</tr>
</table>
<a name="chartostring"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr valign="top"><td>
<h3 class="epydoc"><span class="sig"><span class="sig-name">chartostring</span>(<span class="sig-arg">b</span>)</span>
</h3>
</td><td align="right" valign="top"
>
</td>
</tr></table>
<p>convert a character array to a string array with one less
dimension.</p>
<dl class="fields">
<dt>Parameters:</dt>
<dd><ul class="nomargin-top">
<li><strong class="pname"><code>b</code></strong> - Input character array (numpy datatype 'S1' or 'U1'). Will be
converted to a array of strings, where each string has a fixed
length of b.shape[-1] characters.</li>
</ul></dd>
<dt>Returns:</dt>
<dd>A numpy string array with datatype 'SN' or 'UN' and shape
b.shape[:-1], where N=b.shape[-1].</dd>
</dl>
</td></tr></table>
</div>
<a name="date2index"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr valign="top"><td>
<h3 class="epydoc"><span class="sig"><span class="sig-name">date2index</span>(<span class="sig-arg">dates</span>,
<span class="sig-arg">nctime</span>,
<span class="sig-arg">calendar</span>=<span class="sig-default">None</span>,
<span class="sig-arg">select</span>=<span class="sig-default">'exact'</span>)</span>
</h3>
</td><td align="right" valign="top"
>
</td>
</tr></table>
<p>Return indices of a netCDF time variable corresponding to the given
dates.</p>
<dl class="fields">
<dt>Parameters:</dt>
<dd><ul class="nomargin-top">
<li><strong class="pname"><code>dates</code></strong> - A datetime object or a sequence of datetime objects. The datetime
objects should not include a time-zone offset.</li>
<li><strong class="pname"><code>nctime</code></strong> - A netCDF time variable object. The nctime object must have a
<code>units</code> attribute.</li>
<li><strong class="pname"><code>calendar</code></strong> - Describes the calendar used in the time calculation. Valid
calendars <code>'standard', 'gregorian', 'proleptic_gregorian'
'noleap', '365_day', '360_day', 'julian', 'all_leap',
'366_day'</code>. Default is <code>'standard'</code>, which is a
mixed Julian/Gregorian calendar If <code>calendar</code> is None,
its value is given by <code>nctime.calendar</code> or
<code>standard</code> if no such attribute exists.</li>
<li><strong class="pname"><code>select</code></strong> - <code>'exact', 'before', 'after', 'nearest'</code> The index
selection method. <code>exact</code> will return the indices
perfectly matching the dates given. <code>before</code> and
<code>after</code> will return the indices corresponding to the
dates just before or just after the given dates if an exact match
cannot be found. <code>nearest</code> will return the indices
that correspond to the closest dates.</li>
</ul></dd>
<dt>Returns:</dt>
<dd>an index (indices) of the netCDF time variable corresponding to
the given datetime object(s).</dd>
</dl>
</td></tr></table>
</div>
<a name="date2num"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr valign="top"><td>
<h3 class="epydoc"><span class="sig"><span class="sig-name">date2num</span>(<span class="sig-arg">dates</span>,
<span class="sig-arg">units</span>,
<span class="sig-arg">calendar</span>=<span class="sig-default">'standard'</span>)</span>
</h3>
</td><td align="right" valign="top"
>
</td>
</tr></table>
<p>Return numeric time values given datetime objects. The units of the
numeric time values are described by the <code class="link">units</code>
argument and the <code class="link">calendar</code> keyword. The datetime
objects must be in UTC with no time-zone offset. If there is a time-zone
offset in <code>units</code>, it will be applied to the returned numeric
values.</p>
<dl class="fields">
<dt>Parameters:</dt>
<dd><ul class="nomargin-top">
<li><strong class="pname"><code>dates</code></strong> - A datetime object or a sequence of datetime objects. The datetime
objects should not include a time-zone offset.</li>
<li><strong class="pname"><code>units</code></strong> - a string of the form <code>'<b>time units</b> since <b>reference
time</b></code>' describing the time units. <b><code>time
units</code></b> can be days, hours, minutes, seconds,
milliseconds or microseconds. <b><code>reference time</code></b>
is the time origin. Accuracy is somewhere between a millisecond
and a microsecond, depending on the time interval and the
calendar used.</li>
<li><strong class="pname"><code>calendar</code></strong> - describes the calendar used in the time calculations. All the
values currently defined in the <a
href="http://cf-pcmdi.llnl.gov/documents/cf-conventions/"
target="_top">CF metadata convention</a> are supported. Valid
calendars <code>'standard', 'gregorian', 'proleptic_gregorian'
'noleap', '365_day', '360_day', 'julian', 'all_leap',
'366_day'</code>. Default is <code>'standard'</code>, which is a
mixed Julian/Gregorian calendar.</li>
</ul></dd>
<dt>Returns:</dt>
<dd>a numeric time value, or an array of numeric time values.</dd>
</dl>
</td></tr></table>
</div>
<a name="num2date"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr valign="top"><td>
<h3 class="epydoc"><span class="sig"><span class="sig-name">num2date</span>(<span class="sig-arg">times</span>,
<span class="sig-arg">units</span>,
<span class="sig-arg">calendar</span>=<span class="sig-default">'standard'</span>)</span>
</h3>
</td><td align="right" valign="top"
>
</td>
</tr></table>
<p>Return datetime objects given numeric time values. The units of the
numeric time values are described by the <code>units</code> argument and
the <code>calendar</code> keyword. The returned datetime objects
represent UTC with no time-zone offset, even if the specified
<code>units</code> contain a time-zone offset.</p>
<dl class="fields">
<dt>Parameters:</dt>
<dd><ul class="nomargin-top">
<li><strong class="pname"><code>times</code></strong> - numeric time values.</li>
<li><strong class="pname"><code>units</code></strong> - a string of the form <code>'<b>time units</b> since <b>reference
time</b></code>' describing the time units. <b><code>time
units</code></b> can be days, hours, minutes, seconds,
milliseconds or microseconds. <b><code>reference time</code></b>
is the time origin. Accuracy is somewhere between a millisecond
and a microsecond, depending on the time interval and the
calendar used.</li>
<li><strong class="pname"><code>calendar</code></strong> - describes the calendar used in the time calculations. All the
values currently defined in the <a
href="http://cf-pcmdi.llnl.gov/documents/cf-conventions/"
target="_top">CF metadata convention</a> are supported. Valid
calendars <code>'standard', 'gregorian', 'proleptic_gregorian'
'noleap', '365_day', '360_day', 'julian', 'all_leap',
'366_day'</code>. Default is <code>'standard'</code>, which is a
mixed Julian/Gregorian calendar.</li>
</ul></dd>
<dt>Returns:</dt>
<dd>a datetime instance, or an array of datetime instances.
<p>The datetime instances returned are 'real' python datetime
objects if <code>calendar='proleptic_gregorian'</code>, or
<code>calendar = 'standard'</code> or <code>'gregorian'</code>
and the date is after the breakpoint between the Julian and
Gregorian calendars (1582-10-15). Otherwise, they are 'phony'
datetime objects which support some but not all the methods of
'real' python datetime objects. The datetime instances do not
contain a time-zone offset, even if the specified
<code>units</code> contains one.</p></dd>
</dl>
</td></tr></table>
</div>
<a name="stringtoarr"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr valign="top"><td>
<h3 class="epydoc"><span class="sig"><span class="sig-name">stringtoarr</span>(<span class="sig-arg">a</span>,
<span class="sig-arg">NUMCHARS</span>,
<span class="sig-arg">dtype</span>=<span class="sig-default">'S'</span>)</span>
</h3>
</td><td align="right" valign="top"
>
</td>
</tr></table>
<p>convert a string to a character array of length NUMCHARS</p>
<dl class="fields">
<dt>Parameters:</dt>
<dd><ul class="nomargin-top">
<li><strong class="pname"><code>a</code></strong> - Input python string.</li>
<li><strong class="pname"><code>NUMCHARS</code></strong> - number of characters used to represent string (if len(a) <
NUMCHARS, it will be padded on the right with blanks).</li>
<li><strong class="pname"><code>dtype</code></strong> - type of numpy array to return. Default is 'S', which means an
array of dtype 'S1' will be returned. If dtype='U', a unicode
array (dtype = 'U1') will be returned.</li>
</ul></dd>
<dt>Returns:</dt>
<dd>A rank 1 numpy character array of length NUMCHARS with datatype
'S1' (default) or 'U1' (if dtype='U')</dd>
</dl>
</td></tr></table>
</div>
<a name="stringtochar"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<table width="100%" cellpadding="0" cellspacing="0" border="0">
<tr valign="top"><td>
<h3 class="epydoc"><span class="sig"><span class="sig-name">stringtochar</span>(<span class="sig-arg">a</span>)</span>
</h3>
</td><td align="right" valign="top"
>
</td>
</tr></table>
<p>convert a string array to a character array with one extra
dimension</p>
<dl class="fields">
<dt>Parameters:</dt>
<dd><ul class="nomargin-top">
<li><strong class="pname"><code>a</code></strong> - Input numpy string array with numpy datatype 'SN' or 'UN', where
N is the number of characters in each string. Will be converted
to an array of characters (datatype 'S1' or 'U1') of shape
a.shape + (N,).</li>
</ul></dd>
<dt>Returns:</dt>
<dd>A numpy character array with datatype 'S1' or 'U1' and shape
a.shape + (N,), where N is the length of each string in a.</dd>
</dl>
</td></tr></table>
</div>
<br />
<!-- ==================== VARIABLES DETAILS ==================== -->
<a name="section-VariablesDetails"></a>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr bgcolor="#70b0f0" class="table-header">
<td align="left" colspan="2" class="table-header">
<span class="table-header">Variables Details</span></td>
</tr>
</table>
<a name="default_fillvals"></a>
<div>
<table class="details" border="1" cellpadding="3"
cellspacing="0" width="100%" bgcolor="white">
<tr><td>
<h3 class="epydoc">default_fillvals</h3>
<dl class="fields">
</dl>
<dl class="fields">
<dt>Value:</dt>
<dd><table><tr><td><pre class="variable">
<code class="variable-group">{</code><code class="variable-quote">'</code><code class="variable-string">S1</code><code class="variable-quote">'</code><code class="variable-op">: </code><code class="variable-quote">'</code><code class="variable-string">\x00</code><code class="variable-quote">'</code><code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">U1</code><code class="variable-quote">'</code><code class="variable-op">: </code><code class="variable-quote">'</code><code class="variable-string">\x00</code><code class="variable-quote">'</code><code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">f4</code><code class="variable-quote">'</code><code class="variable-op">: </code>9.96920996839e+36<code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">f8</code><code class="variable-quote">'</code><code class="variable-op">: </code>9.96920996839e+36<code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">i1</code><code class="variable-quote">'</code><code class="variable-op">: </code>-127<code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">i2</code><code class="variable-quote">'</code><code class="variable-op">: </code>-32767<code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">i4</code><code class="variable-quote">'</code><code class="variable-op">: </code>-2147483647<code class="variable-op">,</code>
<code class="variable-quote">'</code><code class="variable-string">i8</code><code class="variable-quote">'</code><code class="variable-op">: </code>-9223372036854775806<code class="variable-op">,</code>
<code class="variable-ellipsis">...</code>
</pre></td></tr></table>
</dd>
</dl>
</td></tr></table>
</div>
<br />
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Home link -->
<th bgcolor="#70b0f0" class="navbar-select"
> Home </th>
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<th class="navbar" width="100%"></th>
</tr>
</table>
<table border="0" cellpadding="0" cellspacing="0" width="100%%">
<tr>
<td align="left" class="footer">
Generated by Epydoc 3.0.1 on Wed Mar 18 10:55:13 2015
</td>
<td align="right" class="footer">
<a target="mainFrame" href="http://epydoc.sourceforge.net"
>http://epydoc.sourceforge.net</a>
</td>
</tr>
</table>
<script type="text/javascript">
<!--
// Private objects are initially displayed (because if
// javascript is turned off then we want them to be
// visible); but by default, we want to hide them. So hide
// them unless we have a cookie that says to show them.
checkCookie();
// -->
</script>
</body>
</html>
