The simplest way to start working with EGADS is to run it from the Python command line. To load EGADS into the Python name-space, simply import it:
>>> import egads
You may then begin working with any of the algorithms and functions contained in EGADS.
There are several useful methods to explore the routines contained in EGADS.
The first is using the Python built-in dir()
command:
>>> dir(egads)
returns all the classes and subpackages contained in EGADS. EGADS follows the naming conventions from the Python Style Guide (http://www.python.org/dev/peps/pep-0008), so classes are always MixedCase
, functions and modules are generally lowercase
or lowercase_with_underscores
. As a further example,
>>> dir(egads.input)
would returns all the classes and subpackages of the egads.input
module.
Another way to explore EGADS is by using tab completion, if supported by your Python installation. Typing
>>> egads.
then hitting TAB
will return a list of all available options.
Python has built-in methods to display documentation on any function known as docstrings.
The easiest way to access them is using the help()
function:
>>> help(egads.input.NetCdf)
or
>>> egads.input.NetCdf?
will return all methods and their associated documentation for the :class:`~.NetCdf` class.
To have a list of file in a directory, use the following function:
>>> egads.input.get_file_list('path/to/all/netcdf/files/*.nc')
The :class:`~.EgadsData` class
At the core of the EGADS package is a data class intended to handle data and associated metadata in a consistent way between files, algorithms and within the framework. This ensures that important metadata is not lost when combining data form various sources in EGADS.
Additionally, by subclassing the Quantities and Numpy packages, :class:`~.EgadsData` incorporates unit comprehension to reduce unit-conversion errors during calculation, and supports broad array manipulation capabilities. This section describes how to employ the :class:`~.EgadsData` class in the EGADS program scope.
Creating :class:`~.EgadsData` instances
The :class:`~.EgadsData` class takes four basic arguments:
- value
- Value to assign to :class:`~.EgadsData` instance. Can be scalar, array, or other :class:`~.EgadsData` instance.
- units
- Units to assign to :class:`~.EgadsData` instance. Should be string representation of units, and can be a compound units type such as 'g/kg', 'm/s^2', 'feet/second', etc.
- variable metadata
- An instance of the :class:`~.VariableMetadata` type or dictionary, containing keywords and values of any metadata to be associated with this :class:`~.EgadsData` instance.
- other attributes
- Any other attributes added to the class are automatically stored in the :class:`~.VariableMetadata` instance associated with the :class:`~.EgadsData` instance.
The following are examples of creating :class:`~.EgadsData` instances:
>>> x = egads.EgadsData([1,2,3], 'm') >>> a = [1,2,3,4] >>> b = egads.EgadsData(a, 'km', b_metadata) >>> c = egads.EgadsData(28, 'degC', long_name="current temperature")
If, during the call to :class:`~.EgadsData`, no units are provided, but a variable metadata instance is provided with a units property, this will then be used to define the :class:`~.EgadsData` units:
>>> x_metadata = egads.core.metadata.VariableMetadata({'units':'m', 'long_name':'Test Variable'}) >>> x = egads.EgadsData([1,2,3], x_metadata) >>> print x.units m >>> print x.metadata {'units': 'm', 'long_name': 'Test Variable'}
The :class:`~.EgadsData` is a subclass of the Quantities and Numpy packages. Thus it allows different kind of operations like addition, substraction, slicing, and many more. For each of those operations, a new :class:`~.EgadsData` class is created with all their attributes.
Note
With mathematical operands, as metadata define an :class:`~.EgadsData` before an operation, and may not reflect the new :class:`~.EgadsData`, it has been decided to not keep the metadata attribute. It's the responsability of the user to add a new :class:`~.VariableMetadata` instance or a dictionary to the new :class:`~.EgadsData` object. It is not true if a user wants to slice an :class:`~.EgadsData`. In that case, metadata are automatically attributed to the new :class:`~.EgadsData`.
The metadata object used by :class:`~.EgadsData` is an instance of :class:`~.VariableMetadata`, a dictionary object containing methods to recognize, convert and validate known metadata types. It can reference parent metadata objects, such as those from an algorithm or data file, to enable users to track the source of a particular variable.
When reading in data from a supported file type (NetCDF, NASA Ames), or doing calculations with an EGADS algorithm, EGADS will automatically populate the associated metadata and assign it to the output variable. However, when creating an :class:`~.EgadsData` instance manually, the metadata must be user-defined.
As mentioned, :class:`~.VariableMetadata` is a dictionary object, thus all metadata are stored as keyword:value pairs. To create metadata manually, simply pass in a dictionary object containing the desired metadata:
>>> var_metadata_dict = {'long_name':'test metadata object', '_FillValue':-9999} >>> var_metadata = egads.core.metadata.VariableMetadata(var_metadata_dict)
To take advantage of its metadata recognition capabilities, a conventions
keyword can be passed with the variable metadata to give a context to these metadata.
>>> var_metadata = egads.core.metadata.VariableMetadata(var_metadata_dict, conventions='CF-1.0')
If a particular :class:`~.VariableMetadata` object comes from a file or algorithm, the class attempts to assign the conventions
automatically. While reading from a file, for example, the class attempts to discover the conventions used based on the "Conventions" keyword, if present.
:class:`~.EgadsData` subclasses Quantities, thus all of the latter's unit comprehension methods are available when using :class:`~.EgadsData`. This section will outline the basics of unit comprehension. A more detailed tutorial of the unit comprehension capabilities can be found at http://packages.python.org/quantities/
In general, units are assigned to :class:`~.EgadsData` instances when they are being created.
>>> a = egads.EgadsData([1,2,3], 'm') >>> b = egads.EgadsData([4,5,6], 'meters/second')
Once a unit type has been assigned to an :class:`~.EgadsData` instance, it will remain that class of unit and can only be converted between other types of that same unit. The rescale
method can be used to convert between similar units, but will give an error if an attempt is made to convert to non-compatible units.
>>> a = egads.EgadsData([1,2,3], 'm') >>> a_km = a.rescale('km') >>> print a_km ['EgadsData', array([0.001, 0.002, 0.003]), 'km'] >>> a_grams = a.rescale('g') ValueError: Unable to convert between units of "m" and "g"
Likewise, arithmetic operations between :class:`~.EgadsData` instances are handled using the unit comprehension provided by Quantities, and behave . For example adding units of a similar type is permitted:
>>> a = egads.EgadsData(10, 'm') >>> b = egads.EgadsData(5, 'km') >>> a + b ['EgadsData', array(5010.0), 'm']
But, non-compatible types cannot be added. They can, however, be multiplied or divided:
>>> distance = egads.EgadsData(10, 'm') >>> time = egads.EgadsData(1, 's') >>> distance + time ValueError: Unable to convert between units of "s" and "m" >>> distance/time ['EgadsData', array(10), 'm/s']
EGADS provides the :class:`egads.input.text_file_io.EgadsFile` class as a simple wrapper for interacting with generic text files. :class:`~.EgadsFile` can read, write and display data from text files, but does not have any tools for automatically formatting input or output data.
To open a text file the :class:`~.EgadsFile` class, use the :meth:`~.EgadsFile.open(filename, permissions)` method:
>>> import egads >>> f = egads.input.EgadsFile() >>> f.open('/pathname/filename.txt','r')
Valid values for permissions are:
r
-- Read: opens file for reading only. Default value if nothing is provided.w
-- Write: opens file for writing, and overwrites data in file.a
-- Append: opens file for appending data.r+
-- Read and write: opens file for both reading and writing.
The following methods are available to control the current position in the file and display more information about the file.
f.display_file()
-- Prints contents of file out to standard output.f.get_position()
-- Returns current position in file as integer.f.seek(location, from_where)
-- Seeks to specified location in file.location
is an integer specifying how far to seek. Valid options forfrom_where
areb
to seek from beginning of file,c
to seek from current position in file ande
to seek from the end of the file.f.reset()
-- Resets position to beginning of file.
Reading data is done using the read(size)
method on a file that has been opened with r
or r+
permissions:
>>> import egads >>> f = egads.input.EgadsFile() >>> f.open('myfile.txt','r') >>> single_char_value = f.read() >>> multiple_chars = f.read(10)
If the size
parameter is not specified, the read()
function will input a single character from the open file. Providing an integer value n as the size
parameter to read(size)
will return n characters from the open file.
Data can be read line-by-line from text files using read_line()
:
>>> line_in = f.read_line()
To write data to a file, use the write(data)
method on a file that has been opened with w
, a
or r+
permissions:
>>> import egads >>> f = egads.input.EgadsFile() >>> f.open('myfile.txt','a') >>> data = 'Testing output data to a file.\n This text will appear on the 2nd line.' >>> f.write(data)
To close a file, simply call the close()
method:
>>> f.close()
Here is a basic ASCII file, created by EGADS:
.. literalinclude:: example_files/main_raw_file.dat
This file has been created with the following commands:
import EGADS module:
>>> import egads
create two main variables, following the official EGADS convention:
>>> data1 = egads.EgadsData(value=[5.0,2.0,-2.0,0.5,4.0], units='mm', name='sea level', scale_factor=1., add_offset=0., _FillValue=-9999) >>> data2 = egads.EgadsData(value=[1.0,3.0,-1.0,2.5,6.0], units='mm', name='corr sea level', scale_factor=1., add_offset=0., _FillValue=-9999)
create an independant variable, still by following the official EGADS convention:
>>> time = egads.EgadsData(value=[1.0,2.0,3.0,4.0,5.0], units='seconds since 19700101T00:00:00', name='time')
create a new EgadsFile instance:
>>> f = egads.input.EgadsFile()
use the following function to open a new file:
>>> f.open('main_raw_file.dat', 'w')
prepare the headers if necessary:
>>> headers = '# The current file has been created with EGADS\n# Institution: My Institution\n# Author(s): John Doe\n' >>> headers += time.metadata["long_name"] + ' ' + data1.metadata["long_name"] + ' ' + data2.metadata["long_name"] + '\n''My institution')
prepare an object to receive all data:
>>> data = '' >>> for i, _ in enumerate(time.value): ... data += str(time.value[i]) + ' ' + str(data1.value[i]) + ' ' + str(data2.value[i]) + '\n'
write the headers and data into the file
>>> f.write(headers) >>> f.write(data)
and do not forget to close the file:
>>> f.close()
:class:`egads.input.text_file_io.EgadsCsv` is designed to easily input or output data in CSV format. Data input using :class:`~.EgadsCsv` is separated into a list of arrays, which each column a separate array in the list.
To open a text file the :class:`~.EgadsCsv` class, use the open(pathname, permissions, delimiter, quotechar)
method:
>>> import egads >>> f = egads.input.EgadsCsv() >>> f.open('/pathname/filename.txt','r',',','"')
Valid values for permissions are:
r
-- Read: opens file for reading only. Default value if nothing is provided.w
-- Write: opens file for writing, and overwrites data in file.a
-- Append: opens file for appending data.r+
-- Read and write: opens file for both reading and writing.
The delimiter
argument is a one-character string specifying the character used to separate fields in the CSV file to be read; the default value is ,
. The quotechar
argument is a one-character string specifying the character used to quote fields containing special characters in the CSV file to to be read; the default value is "
.
The following methods are available to control the current position in the file and display more information about the file.
f.display_file()
-- Prints contents of file out to standard output.f.get_position()
-- Returns current position in file as integer.f.seek(location, from_where)
-- Seeks to specified location in file.location
is an integer specifying how far to seek. Valid options forfrom_where
areb
to seek from beginning of file,c
to seek from current position in file ande
to seek from the end of the file.f.reset()
-- Resets position to beginning of file.
Reading data is done using the read(lines, format)
method on a file that has been opened with r
or r+
permissions:
>>> import egads >>> f = egads.input.EgadsCsv() >>> f.open('mycsvfile.csv','r') >>> single_line_as_list = f.read(1) >>> all_lines_as_list = f.read()
read(lines, format)
returns a list of the items read in from the CSV file. The arguments lines
and format
are optional. If lines
is provided, read(lines, format)
will read in the specified number of lines, otherwise it will read the whole file. format
is an optional list of characters used to decompose the elements read in from the CSV files to their proper types. Options are:
i
-- intf
-- floatl
-- longs
-- string
Thus to read in the line:
FGBTM,20050105T143523,1.5,21,25
the command to input with proper formatting would look like this:
>>> data = f.read(1, ['s','s','f','f'])
To write data to a file, use the write(data)
method on a file that has been opened with w
, a
or r+
permissions:
>>> import egads >>> f = egads.input.EgadsCsv() >>> f.open('mycsvfile.csv','a') >>> titles = ['Aircraft ID','Timestamp','Value1','Value2','Value3'] >>> f.write(titles)
where the titles
parameter is a list of strings. This list will be output to the CSV, with each strings separated by the delimiter specified when the file was opened (default is ,
).
To write multiple lines out to a file, writerows(data)
is used:
>>> data = [['FGBTM','20050105T143523',1.5,21,25],['FGBTM','20050105T143524',1.6,20,25.6]] >>> f.writerows(data)
To close a file, simply call the close()
method:
>>> f.close()
Here is a basic CSV file, created by EGADS:
.. literalinclude:: example_files/main_csv_file.csv
This file has been created with the following commands:
import EGADS module:
>>> import egads
create two main variables, following the official EGADS convention:
>>> data1 = egads.EgadsData(value=[5.0,2.0,-2.0,0.5,4.0], units='mm', name='sea level', scale_factor=1., add_offset=0., _FillValue=-9999) >>> data2 = egads.EgadsData(value=[1.0,3.0,-1.0,2.5,6.0], units='mm', name='corr sea level', scale_factor=1., add_offset=0., _FillValue=-9999)
create an independant variable, still by following the official EGADS convention:
>>> time = egads.EgadsData(value=[1.0,2.0,3.0,4.0,5.0], units='seconds since 19700101T00:00:00', name='time')
create a new EgadsFile instance:
>>> f = egads.input.EgadsCsv()
use the following function to open a new file:
>>> f.open('main_csv_file.csv','w',',','"')
prepare the headers if necessary:
>>> headers = ['time', 'sea level', 'corrected sea level']
prepare an object to receive all data:
>>> data = [time.value, data1.value, data2.value]
write the headers and data into the file
>>> f.write(headers) >>> f.write(data)
and do not forget to close the file:
>>> f.close()
EGADS provides two classes to work with NetCDF files. The simplest, :class:`egads.input.netcdf.NetCdf`, allows simple read/write operations to NetCDF files. The other, :class:`egads.input.netcdf.EgadsNetCdf`, is designed to interface with NetCDF files conforming to the EUFAR Standards & Protocols data and metadata regulations. This class directly reads or writes NetCDF data using instances of the :class:`~.EgadsData` class.
To open a NetCDF file, simply create a :class:`.NetCdf()` instance and then use the open(pathname, permissions)
command:
>>> import egads >>> f = egads.input.NetCdf() >>> f.open('/pathname/filename.nc','r')
Valid values for permissions are:
r
-- Read: opens file for reading only. Default value if nothing is provided.w
-- Write: opens file for writing, and overwrites data in file.a
-- Append: opens file for appending data.r+
-- Same asa
.
f.get_dimension_list()
-- returns a list of all dimensions and their sizesf.get_dimension_list(var_name)
--var_name
is optional and if provided, the function returns a list of all dimensions and their sizes attached tovar_name
f.get_attribute_list()
-- returns a list of all top-level attributesf.get_attribute_list(var_name)
--var_name
is optional and if provided, the function returns a list of all attributes attached tovar_name
f.get_variable_list()
-- returns list of all variablesf.get_filename()
-- returns filename for currently opened filef.get_perms()
-- returns the current permissions on the file that is open
To read data from a file, use the read_variable()
function:
>>> data = f.read_variable(var_name, input_range)
where var_name
is the name of the variable to read in, and input_range
(optional) is a list of min/max values.
If using the egads.input.NetCdf()
class, an array of values contained in var_name
will be returned. If using the egads.input.EgadsNetCdf()
class, an instance of the :class:`~.EgadsData` class will be returned containing the values and attributes of var_name
.
The following describe how to add dimensions or attributes to a file.
f.add_dim(dim_name, dim_size)
-- add dimension to filef.add_attribute(attr_name, attr_value)
-- add attribute to filef.add_attribute(attr_name, attr_value, var_name)
--var_name
is optional and if provided, the function add attribute tovar_name
Data can be output to variables using the write_variable()
function as follows:
>>> f.write_variable(data, var_name, dims, type)
where var_name
is a string for the variable name to output, dims
is a tuple of dimension names (not needed if the variable already exists), and type
is the data type of the variable. The default value is double, other valid options are float, int, short, char and byte.
If using :class:`~.NetCdf`, values for data
passed into write_variable
must be scalar or array. Otherwise, if using :class:`~.EgadsNetCdf`, an instance of :class:`~.EgadsData` must be passed into write_variable
. In this case, any attributes that are contained within the :class:`~.EgadsData` instance are applied to the NetCDF variable as well.
The conversion is only possible on opened NetCDF files. If modifications have been made and haven't been saved, the conversion won't take into account those modifications. Actually, the only File Format Index supported by the conversion in the NASA/Ames format is 1001. Consequently, if variables depend on multiple independant variables (e.g. data
is function of time
, longitude
and latitude
), the file won't be converted and the function will raise an exception. On the contrary, if multiple independant variables (or dimensions) exist, and if each variable depend on only one independant variable (e.g. data
is only function of time
), the file will be converted and the function will generate one file per independant variable. If the user needs to convert a complex file with variables depending on multiple independant variables, the conversion should be done manually by creating a NASA/Ames instance and a NASA/Ames dictionary, by populating the dictionary and by saving the file.
To convert a NetCDF file, simply use:
f.convert_to_nasa_ames()
-- convert the currently opened NetCDF file to NASA/Ames file formatf.convert_to_nasa_ames(na_file, requested_ffi, float_format, delimiter, annotation, no_header)
--na_file
,requested_ffi
,float_format
,delimiter
,annotation
andno_header
are optional parameters ;na_file
is the name of the output file once it has been converted, by default the name of the NetCDF file will be used with the extension .na ;requested_ffi
is not used actually, but will be functional in a next version of EGADS ;float_format
is the formatting string used for formatting floats when writing to output file, by default%g
;delimiter
is a character or a sequence of character for use between data items in the data file, by default ' ' (four spaces) ; ifannotation
is set toTrue
, write the output file with an additional left-hand column describing the contents of each header line, by defaultFalse
; ifno_header
is set toTrue
, then only the data blocks are written to file, by defaultFalse
To convert a NetCDF file to NASA/Ames CSV format, simply use:
f.convert_to_csv()
-- convert the currently opened NetCDF file to NASA/Ames CSV formatf.convert_to_csv(csv_file, float_format, annotation, no_header)
--csv_file
,float_format
,annotation
andno_header
are optional parameters ;csv_file
is the name of the output file once it has been converted, by default the name of the NetCDF file will be used with the extension .csv ;float_format
is the formatting string used for formatting floats when writing to output file, by default%g
; ifannotation
is set toTrue
, write the output file with an additional left-hand column describing the contents of each header line, by defaultFalse
; ifno_header
is set toTrue
, then only the data blocks are written to file, by defaultFalse
f.get_attribute_value(attr_name)
-- returns the value of a global attributef.get_attribute_value(attr_name, var_name)
--var_name
is optional and if provided, the function returns the value of an attribute attached tovar_name
f.change_variable_name(var_name, new_name)
-- change the variable name in currently opened NetCDF file
To close a file, simply use the close()
method:
>>> f.close()
Note
The EGADS :class:`~.NetCdf` and :class:`~.EgadsNetCdf` use the official NetCDF I/O routines, therefore, as described in the NetCDF documentation, it is not possible to remove a variable or more, and to modify the values of a variable. As attributes, global and those linked to a variable, are more dynamic, it is possible to remove, rename, or replace them.
Here is a NetCDF file, created by EGADS, and viewed by the command ncdump -h ....
:
.. literalinclude:: example_files/ncdump_example_file.txt
This file has been created with the following commands:
import EGADS module:
>>> import egads
create two main variables, following the official EGADS convention:
>>> data1 = egads.EgadsData(value=[5.0,2.0,-2.0,0.5,4.0], units='mm', name='sea level', scale_factor=1., add_offset=0., _FillValue=-9999) >>> data2 = egads.EgadsData(value=[1.0,3.0,-1.0,2.5,6.0], units='mm', name='corr sea level', scale_factor=1., add_offset=0., _FillValue=-9999)
create an independant variable, still by following the official EGADS convention:
>>> time = egads.EgadsData(value=[1.0,2.0,3.0,4.0,5.0], units='seconds since 19700101T00:00:00', name='time')
create a new EgadsNetCdf instance with a file name:
>>> f = egads.input.EgadsNetCdf('main_netcdf_file.nc', 'w')
add the global attributes to the NetCDF file:
>>> f.add_attribute('Conventions', 'CF-1.0') >>> f.add_attribute('history', 'the netcdf file has been created by EGADS') >>> f.add_attribute('comments', 'no comments on the netcdf file') >>> f.add_attribute('institution', 'My institution')
add the dimension(s) of your variable(s), here it is
time
:>>> f.add_dim('time', len(time))
write the variable(s), it is a good practice to write at the first place the independant variable
time
:>>> f.write_variable(time, 'time', ('time',), 'double') >>> f.write_variable(data1, 'sea_level', ('time',), 'double') >>> f.write_variable(data2, 'corrected_sea_level', ('time',), 'double')
and do not forget to close the file:
>>> f.close()
To work with NASA Ames files, EGADS incorporates the NAPpy library developed by Ag Stephens of BADC. Information about NAPpy can be found at http://proj.badc.rl.ac.uk/cows/wiki/CowsSupport/Nappy
In EGADS, the NAPpy API has been adapted to match the other EGADS file access methods. Thus, from EGADS, NASA Ames files can be accessed via the :class:`egads.input.nasa_ames_io.NasaAmes` class. Actually, only the FFI 1001 has been interfaced with EGADS.
To open a NASA Ames file, simply create a :class:`.NasaAmes()` instance and then use the open(pathname, permissions)
command:
>>> import egads >>> f = egads.input.NasaAmes() >>> f.open('/pathname/filename.na','r')
Valid values for permissions are:
r
-- Read: opens file for reading only. Default value if nothing is provided.w
-- Write: opens file for writing, and overwrites data in file.a
-- Append: opens file for appending data.r+
-- Same asa
.
Once a file has been opened, a dictionary of NASA/Ames format elements is loaded into memory. That dictionary will be used to overwrite the file or to save to a new file.
f.get_attribute_list()
-- returns a list of all top-level attributesf.get_attribute_list(var_name, var_type, na_dict)
--var_name
is optional and if provided, the function returns list of all attributes attached tovar_name
; ifvar_type
is provided the function will search in the variable typevar_type
by default ;na_dict
is optional if provided, will return a list of all top-level attributes, or allvar_name
attributes, in the NASA/Ames dictionaryna_dict
f.get_attribute_value(attr_name)
-- returns the value of a global attribute namedattr_name
f.get_attribute_value(attr_name, var_name, var_type, na_dict)
--var_name
,var_type
andna_dict
are optional ; ifvar_name
is provided, returns the value of an attribute namedattr_name
attached to a variable namedvar_name
; ifvar_type
is provided, the function will search in the variable typevar_type
by default ; ifna_dict
is provided, returns the attribute value from the NASA/Ames dictionaryna_dict
f.get_dimension_list()
-- returns a list of all variable dimensionsf.get_dimension_list(na_dict, var_type)
--var_type
is optional, if provided, the function returns a list of all variable dimensions based on thevar_type
by default ;na_dict
is optional and will returns the dimension list from the NASA/Ames dictionaryna_dict
;f.get_variable_list()
-- returns list of all variables ;f.get_variable_list(na_dict)
--na_dict
is optional and if provided, will return a list of all variables in the NASA/Ames dictionaryna_dict
f.get_filename()
-- returns filename for currently opened file
To read data from a file, use the read_variable()
function:
>>> data = f.read_variable(var_name)
where var_name
is the name of the variable to read in. The data will be read in to an instance of the :class:`~.EgadsData` class, containing the values and attributes of var_name
.
To write data to the current file or to a new file, the user must save a dictionary of NASA/Ames elements. Few functions are available to help him to prepare the dictionary:
f.create_na_dict
-- create a new dictionary populated with standard NASA/Ames keys.f.write_attribute_value(attr_name, attr_value)
-- write or replace a specific attribute (from the official NASA/Ames attribute list) in the currently opened dictionaryf.write_attribute_value(attr_name, attr_value, var_name, var_type, na_dict)
--var_name
andvar_type
are optional, if provided, write or replace a specific attribute linked to the variablevar_name
(var_type
is by default equal to 'main') in the currently opened dictionary ; ccepted attributes for a variable are 'name', 'units', '_FillValue' and 'scale_factor', other attributes will be refused and should be passed as 'special comments' ;na_dict
is optional and if provided the function will write the attribute in the NASA/Ames dictionaryna_dict
f.write_variable(data, var_name)
-- write or replace a variable ; the function will search ifdata
is already in the dictionary by comparingvarname
with other variable names in the dictionary, if it is found,data
will replace the old variable, if notdata
is considered as a new variable ;data
can be an :class:`~.EgadsData` or a vector/matrix.f.write_variable(data, var_name, var_type, attr_dict, na_dict)
--var_type
,attr_dict
andna_dict
are optional ;attr_dict
(a dictionary of standard NASA/ames variable attributes: 'name', 'units', '_FillValue' and 'scale_factor') must be provided ifdata
is not an :class:`~.EgadsData` (in that case, variable attributes are retrieve from the :class:`~.EgadsData`.metadata dictionary) ; ifna_dict
is provided, the function saves the variable in the NASA/Ames dictionaryna_dict
Once a dictionary is ready, use the save_na_file()
function to save the file:
>>> data = f.save_na_file(file_name, na_dict, float_format):
where file_name
is the name of the new file or the name of the current file, na_dict
the name of the dictionary to be saved (optional, if not provided, the current dictionary will be used), and float_format
the format of the floating numbers in the file (by deffault, two decimal places).
When a NASA/Ames file is opened, all metadata and data are read and stored in memory in a dedicated dictionary. The conversion will convert that dictionary to generate a NetCDF file. If modifications are made to the dictionary, the conversion will take into account those modifications. Actually, the only File Format Index supported by the conversion in the NASA/Ames format is 1001. Consequently, if variables depend on multiple independant variables (e.g. data
is function of time
, longitude
and latitude
), the file won't be converted and the function will raise an exception. If the user needs to convert a complex file with variables depending on multiple independant variables, the conversion should be done manually by creating a NetCDF instance and by populating the NetCDF files with NASA/Ames data and metadata.
To convert a NASA/Ames file, simply use:
f.convert_to_netcdf()
-- convert the currently opened NASA/Ames file to NetCDF format.f.convert_to_netcdf(nc_file)
--nc_file
is an optional parameter ;na_file
is the name of the output file once it has been converted, by default the name of the NASA/Ames file will be used with the extension .nc
f.read_na_dict()
-- returns a deep copy of the current opened file dictionaryf.na_format_information()
-- returns a text explaining the structure of a NASA/Ames file to help the user to modify or to create his own dictionary
To close a file, simply use the close()
method:
>>> f.close()
Here is a NASA/Ames file:
.. literalinclude:: example_files/na_example_file.na
This file has been created with the following commands:
- import EGADS module:
>>> import egads
create two main variables, following the official EGADS convention:
>>> data1 = egads.EgadsData(value=[5.0,2.0,-2.0,0.5,4.0], units='mm', name='sea level', scale_factor=1, _FillValue=-9999) >>> data2 = egads.EgadsData(value=[1.0,3.0,-1.0,2.5,6.0], units='mm', name='corr sea level', scale_factor=1, _FillValue=-9999)
create an independant variable, still by following the official EGADS convention:
>>> time = egads.EgadsData(value=[1.0,2.0,3.0,4.0,5.0], units='seconds since 19700101T00:00:00', name='time')
create a new NASA/Ames empty instance:
>>> f = egads.input.NasaAmes()
initialize a new NASA/Ames dictionary:
>>> na_dict = f.create_na_dict()
prepare the normal and special comments if needed, in a list, one cell for each line:
>>> scom = ['========SPECIAL COMMENTS===========','this file has been created with egads','=========END========='] >>> ncom = ['========NORMAL COMMENTS===========','headers:','time sea level corrected sea level','=========END=========']
populate the main NASA/Ames attributes:
>>> f.write_attribute_value('ONAME', 'John Doe', na_dict = na_dict) # ONAME is the name of the author(s) >>> f.write_attribute_value('ORG', 'An institution', na_dict = na_dict) # ORG is tne name of the organization responsible for the data >>> f.write_attribute_value('SNAME', 'tide gauge', na_dict = na_dict) # SNAME is the source of data (instrument, observation, platform, ...) >>> f.write_attribute_value('MNAME', 'ATESTPROJECT', na_dict = na_dict) # MNAME is the name of the mission, campaign, programme, project dedicated to data >>> f.write_attribute_value('DATE', [2017, 1, 30], na_dict = na_dict) # DATE is the date at which the data recorded in this file begin (YYYY MM DD) >>> f.write_attribute_value('NIV', 1, na_dict = na_dict) # NIV is the number of independent variables >>> f.write_attribute_value('NSCOML', 3, na_dict = na_dict) # NSCOML is the number of special comments lines or the number of elements in the SCOM list >>> f.write_attribute_value('NNCOML', 4, na_dict = na_dict) # NNCOML is the number of special comments lines or the number of elements in the NCOM list >>> f.write_attribute_value('SCOM', scom, na_dict = na_dict) # SCOM is the special comments attribute >>> f.write_attribute_value('NCOM', ncom, na_dict = na_dict) # NCOM is the normal comments attribute
write each variable in the dictionary:
>>> f.write_variable(time, 'time', vartype="independant", na_dict = na_dict) >>> f.write_variable(data1, 'sea level', vartype="main", na_dict = na_dict) >>> f.write_variable(data2, 'corrected sea level', vartype="main", na_dict = na_dict)
and finally, save the dictionary to a NASA/Ames file:
>>> f.save_na_file('na_example_file.na', na_dict)
Since the first version of EGADS, the direct conversion was possible with the NAPpy library with the help of CDMS. As CDMS is not compatible with windows, that possibility has been dropped. However, two functions have been introduced to allow a conversion from NetCDF to NASA/Ames format, and from NASA/Ames format to NetCDF. Please read the section about NetCDF and NASA/Ames file handling to learn how to convert between those formats.
Algorithms in EGADS are stored in the :mod:`egads.algorithms` module, and separated into sub-modules by category (microphysics, thermodynamics, radiation, etc). Each algorithm follows a standard naming scheme, using the algorithm's purpose and source:
{CalculatedParameter}{Detail}{Source}
For example, an algorithm which calculates static temperature, which was provided by CNRM would be named:
TempStaticCnrm
There are several methods to get information about each algorithm contained in EGADS. The EGADS Algorithm Handbook is available for easy reference outside of Python. In the handbook, each algorithm is described in detail, including a brief algorithm summary, descriptions of algorithm inputs and outputs, the formula used in the algorithm, algorithm source and links to additional references. The handbook also specifies the exact name of the algorithm as defined in EGADS. The handbook can be found on the EGADS website.
Within Python, usage information on each algorithm can be found using the help()
command:
>>> help(egads.algorithms.thermodynamics.VelocityTasCnrm) >>> Help on class VelocityTasCnrm in module egads.algorithms.thermodynamics. velocity_tas_cnrm: class VelocityTasCnrm(egads.core.egads_core.EgadsAlgorithm) | FILE velocity_tas_cnrm.py | | VERSION $Revision: 104 $ | | CATEGORY Thermodynamics | | PURPOSE Calculate true airspeed | | DESCRIPTION Calculates true airspeed based on static temperature, | static pressure and dynamic pressure using St Venant's | formula. | | INPUT T_s vector K or C static temperature | P_s vector hPa static pressure | dP vector hPa dynamic pressure | cpa coeff. J K-1 kg-1 specific heat of air (dry | air is 1004 J K-1 kg-1) | Racpa coeff. () R_a/c_pa | | OUTPUT V_p vector m s-1 true airspeed | | SOURCE CNRM/GMEI/TRAMM | | REFERENCES "Mecanique des fluides", by S. Candel, Dunod. | | Bulletin NCAR/RAF Nr 23, Feb 87, by D. Lenschow and | P. Spyers-Duran | ...
Algorithms in EGADS generally accept and return arguments of :class:`~.EgadsData` type, unless otherwise noted. This has the advantages of constant typing between algorithms, and allows metadata to be passed along the whole processing chain. Units on parameters being passed in are also checked for consistency, reducing errors in calculations, and rescaled if needed. However, algorithms will accept any normal data type, as well. They can also return non-:class:`~.EgadsData` instances, if desired.
To call an algorithm, simply pass in the required arguments, in the order they are described in the algorithm help function. An algorithm call, using the VelocityTasCnrm
in the previous section as an example, would therefore be the following:
>>> V_p = egads.algorithms.thermodynamics.VelocityTasCnrm().run(T_s, P_s, dP, cpa, Racpa)
where the arguments T_s
, P_s
, dP
, etc are all assumed to be previously defined in the program scope. In this instance, the algorithm returns an :class:`~.EgadsData` instance to V_p
. To run the algorithm, but return a standard data type (scalar or array of doubles), set the return_Egads
flag to false
.
>>> V_p = egads.algorithms.thermodynamics.VelocityTasCnrm(return_Egads=false). run(T_s, P_s, dP, cpa, Racpa)
Note
When injecting a variable in an EgadsAlgorithm, the format of the variable should follow closely the documentation of the algorithm. If the variable is a scalar, and the algorithm needs a vector, the scalar should be surrounded by brackets: 52.123 -> [52.123].
The recommended method for using EGADS is to create script files, which are extremely useful for common or repetitive tasks. This can be done using a text editor of your choice. The example script belows shows the calculation of density for all NetCDF files in a directory.
.. literalinclude:: example_files/example.py
When scripting in Python, there are several important differences from other programming languages to keep in mind. This section outlines a few of these differences.
Python differs from C++ and Fortran in how loops or nested statements are signified. Whereas C++ uses brackets ('{
' and '}
') and FORTRAN uses end
statements to signify the end of a nesting, Python uses white space. Thus, for statements to nest properly, they must be set at the proper depth. As long as the document is consistent, the number of spaces used doesn't matter, however, most conventions call for 4 spaces to be used per level. See below for examples:
FORTRAN:
X = 0 DO I = 1,10 X = X + I PRINT I END DO PRINT X
Python:
x = 0 for i in range(1,10): x = x + i print i print x
Since September 2016, a Graphical User Interface is available at https://github.com/eufarn7sp/egads-gui. It gives the user the possibility to explore data, apply/create algorithms, display and plot data. Still in beta state, the user will have the possibility in the future to work on a batch of file. For now, EGADS GUI comes as a simple python script and need to be launch from the terminal, if EGADS is installed, and once in the EGADS GUI directory:
>>> python egads_gui.py
It will be available soon as a stand alone (imbedding a version of EGADS CORE or using an already installed EGADS package).