Data Containers

sbpy relies heavily on the use of DataClass data containers that are used to encapsulate data and to propagate them through your workflow.

sbpy provides data containers for orbital elements (Orbit), ephemerides (Ephem), observational data (Obs), and physical properties (Phys) data.

What are Data Containers?

DataClass - and hence all the data containers presented here - uses a QTable object under the hood. You can think of those as tables - consisting of fields (or columns) and rows - that have units attached to them, allowing you to propagate these units through your programs. We strongly urge the user to make use of units in the definition of data containers in order to minimize confusion and tap the full potential of sbpy. Finally, DataClass objects have a meta attribute that enables the user to label these objects with unstructured meta data.

The user is free to add any fields they want to a DataClass object. However, in order to enable the seamless use of sbpy functions and methods, we require the user to pick among a few common field names for different properties as listed here. DataClass objects are able to identify alternative field names as listed in this document, as well as to perform transformations between a few field names - see below for more details.

A DataClass object can hold as many data rows as you want. All rows can refer to a single object, or each row can refer to a separate object - this is usually up to the user, restrictions exist only in a few cases as detailed in this documentation.

Why are there different Data Container Types?

Although the data container classes Ephem, Orbit, Obs and Phys look very similar and are based on the same base class (DataClass), there are subtle differences. The container classes have been introduced to minimize confusion between properties of different natures, i.e., to avoid mixing apples with oranges. For instance, Ephem objects deal with ephemerides (e.g., RA and DEC and other properties that change as a function of time); the diameter of an object, however, does usually not vary with time. Adding a “diameter” field to an Ephem object would hence be inefficient: this object can contain hundreds of rows describing a target’s ephemerides, while the diameter would be the same in each of these rows. Furthermore, providing separate data containers for different properties enables the implementation of specifically designed methods to query and modify the data held by these classes.

Data Container Type Overview


Ephem has been designed to hold ephemerides, i.e., properties that vary with time.

Ephem currently provides convenience functions to query ephemerides from the JPL Horizons system (from_horizons) the Minor Planet Center (from_mpc), IMCCE’s Miriade system (from_miriade) as well as a convenience function to derive ephemerides from an Orbit object using pyoorb.


Obs is tailored to holding observational data, e.g., magnitudes as a function of time. The Obs class is the only data container that is not directly derived from DataClass, but from Ephem, providing the same functionality as the latter. from_mpc enables you to query observations reported to the MPC for a specific target and supplement queries one of the ephemerides service to supplement your observation data.


Orbit should be used to hold orbital elements of one or several bodies. Elements can be retrieved using the convenience function from_horizons, propagated using oo_propagate, and transformed into other frames using oo_transform.


Phys objects are meant to hold physical properties that do not change over time. Known physical properties can currently be queried from the JPL Small-Body Database Browser system using from_sbdb.


Names objects are somewhat different from the other data containers, as they don’t hold properties but only object names. These names can be used to identify object nature (asteroid_or_comet) and they can be parsed to extract individual identifier components (parse_asteroid and parse_comet).

How to use Data Containers

All of the data objects dealt with in share the same common base class: DataClass defines the basic functionality and makes sure that all objects can used in the exact same way.

In plain words, this means that in the following examples you can replace DataClass, Ephem, Orbit, Obs, and Phys object with each other. In order to show some useful use cases, we will iterate between these types, but keep in mind: they all work the exact same way.

Building Data Containers

While Ephem, Orbit, Obs, and Phys provide a range of convenience functions to build objects containing data, for instance from online data archives, it is easily possible to build these objects from scratch. This can be done for input data stored in dictionaries (from_dict), lists or arrays (from_columns and from_rows), Table objects (from_table), or from data files (from_file).

Depending on how your input data are organized, you can use different options in different cases:

Building a Data Container from a Dictionary

Assume that you want to build an Orbit object to propagate this orbit and obtain ephemerides. Since you are dealing with a single orbit, the most convenient solution might be to use a dictionary to build your object:

>>> from import Orbit
>>> from astropy.time import Time
>>> import astropy.units as u
>>> elements = {'a':1.234*, 'e':0.1234, 'i':12.34*u.deg,
...             'argper': 123.4*u.deg, 'node': 45.2*u.deg,
...             'epoch': Time(2451200.5, format='jd'), 'true_anom':23.1*u.deg}
>>> orb = Orbit.from_dict(elements)
>>> orb  
<QTable length=1>
   a       e       i     argper   node    epoch   true_anom
   AU             deg     deg     deg      Time      deg
float64 float64 float64 float64 float64  float64   float64
------- ------- ------- ------- ------- --------- ---------
  1.234  0.1234   12.34   123.4    45.2 2451200.5      23.1

One quick note on building DataClass objects from dictionaries: dictionaries have no intrinsic order. In dictionary elements as defined here, there is no guarantee that 'a' will always be located before 'e' when reading out the dictionary item by item, which happens when the data table is built in the background. Hence, the order of the resulting data table columns has to be considered random. If you want to force a specific order on the columns in your data table, you can use an OrderedDict instead of a simple dictionary. The order of elements in an OrderedDict will be the same as the order of the data table columns.

For details on how to build objects from dictionaries, see from_dict.

Building a Data Container from Columns

Now assume that you want to build an Obs object holding RA, Dec, and observation midtime for some target that you observed. In this case, you can use from_columns as shown here:

>>> from import Obs
>>> import astropy.units as u
>>> from astropy.time import Time
>>> from numpy import array
>>> ra = [10.223423, 10.233453, 10.243452]*u.deg
>>> dec = [-12.42123, -12.41562, -12.40435]*u.deg
>>> epoch = Time(2451523.5 + array([0.1234, 0.2345, 0.3525]), format='jd')
>>> obs = Obs.from_columns([ra, dec, epoch], names=['ra', 'dec', 't'])
>>> obs  
<QTable length=3>
    ra       dec         t
   deg       deg
 float64   float64      Time
--------- --------- ------------
10.223423 -12.42123 2451523.6234
10.233453 -12.41562 2451523.7345
10.243452 -12.40435 2451523.8525

Note how epoch is handled differently: it is provided to Obs.from_column as a Time object (see Design Principles - The Zen of sbpy for a discussion).

For details on how to build objects from lists or arrays, see from_columns and also from_rows, depending on whether your data is represented as rows or columns. Note that you could also use from_dict by providing column data to the different fields.

Building a Data Container from a Table

If your data are already available as a Table or QTable, you can simply convert it into a DataClass object using from_table.

Building a Data Container from a File

You can also read in the data from a file that should be properly formatted using from_file. This function merely serves as a wrapper for and uses the same parameters as the latter function; please refer to this document for a review.

As an example, you can read in a properly formatted ASCII file using the following lines:

>>> from import Ephem
>>> data = Ephem.from_file('data.txt', format='ascii') 

Please note that the file formats available (see here for a list of available formats) provide varying support for units and meta data. For instance, basic, csv, html, and latex do not provide unit or meta data information. However, fits, cds, daophot, ecsv, and ipac do support units and meta data.

Building a Data Container from an Online Query

Most DataClass data containers offer convenience functions to query data from online service. Please refer to the corresponding classes for information and examples for querying data.

A Note on Field Names

In order for sbpy to properly identify the fields that might be necessary for calculations, default column names should be used to name these fields. For instance, a column of Right Ascensions should be named 'RA' or 'ra'. For a list of acceptable field names, please refer to the list of sbpy Field Names.

Also note that sbpy is able to use alternative field names, but only those that are listed in the list of sbpy Field Names.

Accessing data

In order to obtain a list of field names in a DataClass object, you can use field_names:

>>> obs.field_names
['ra', 'dec', 't']

Each of these columns can be accessed easily, for instance:

>>> obs['ra']  
[10.223423 10.233453 10.243452] deg

which will return an Quantity object if that column has a Unit attached to it or a Column otherwise.

Similarly, if you are interested in the first set of observations in obs, you can use:

>>> obs[0]  
    ra       dec         t
   deg       deg        Time
--------- --------- ------------
10.223423 -12.42123 2451523.6234

which returns you a new instance of the same class as your original objet with only the requested subset of the data. In order to retrieve RA from the second observation, you can combine both examples and do:

>>> obs[1]['ra'] 
10.233453 deg

Just like in any Table or QTable object, you can use slicing to obtain subset tables from your data, for instance:

>>> obs['ra', 'dec']  
<QTable length=3>
    ra       dec
   deg       deg
--------- ---------
10.223423 -12.42123
10.233453 -12.41562
10.243452 -12.40435
>>> obs[:2] 
    ra       dec         t
   deg       deg        Time
--------- --------- ------------
10.223423 -12.42123 2451523.6234
10.233453 -12.41562 2451523.7345
>>> obs[obs['ra'] <= 10.233453*u.deg] 
    ra       dec         t
   deg       deg        Time
--------- --------- ------------
10.223423 -12.42123 2451523.6234
10.233453 -12.41562 2451523.7345

The results of these examples will be of the same data type as obs (or really just any type derived from DataClass, e.g., Ephem, Orbit, …) The latter example shown here uses a condition to filter data (only those observations with RA less than or equal to 10.233453 degrees; note that it is necessary here to apply u.deg to the value that all the RAs are compared against) but selects all the columns in the original table.

If you ever need to access the actual QTable object that is inside each DataClass object, you can access it as obs.table.

Modifying an object

Individual elements, entire rows, and columns can be modified by directly addressing them:

>>> obs['ra'] 
[10.223423 10.233453 10.243452 10.25546  10.265425 10.25546  10.4545
 10.5656  ] deg
>>> obs['ra'] = obs['ra'] + 0.1*u.deg
>>> obs['ra'] 
[10.323423 10.333453 10.343452 10.35546  10.365425 10.35546  10.5545
 10.6656  ] deg

More complex data table modifications are possible by directly accessing the underlying QTable object as shown below.

DataClass provides a direct interface to the table modification functions provided by Table: add_row, add_column, add_columns, etc. For instance, it is trivial to add additional rows and columns to these objects.

Let’s assume you want to add some more observations to your obs object:

>>> obs.table.add_row([10.255460*u.deg, -12.39460*u.deg, 2451523.94653*u.d])
>>> obs  
<QTable length=4>
    ra       dec          t
   deg       deg
 float64   float64      Time
--------- --------- -------------
10.323423 -12.42123  2451523.6234
10.333453 -12.41562  2451523.7345
10.343452 -12.40435  2451523.8525
 10.25546  -12.3946 2451523.94653

or if you want to add a column to your object:

>>> from astropy.table import Column
>>> obs.table.add_column(Column(['V', 'V', 'R', 'i'], name='filter'))
>>> obs  
<QTable length=4>
    ra       dec          t       filter
   deg       deg
 float64   float64       Time      str1
--------- --------- ------------- ------
10.223423 -12.42123  2451523.6234      V
10.233453 -12.41562  2451523.7345      V
10.243452 -12.40435  2451523.8525      R
 10.25546  -12.3946 2451523.94653      i

The same result can be achieved using the following syntax:

>>> obs['filter2'] = ['V', 'V', 'R', 'i']  
>>> obs  
<QTable length=4>
    ra       dec          t       filter filter2
   deg       deg
 float64   float64       Time      str1    str1
--------- --------- ------------- ------ -------
10.223423 -12.42123  2451523.6234      V       V
10.233453 -12.41562  2451523.7345      V       V
10.243452 -12.40435  2451523.8525      R       R
 10.25546  -12.3946 2451523.94653      i       i

Similarly, existing columns can be modified using:

>>> obs['filter'] = ['g', 'i', 'R', 'V']  

Note how the add_column and add_row functions are called from obs.table. table is a property that exposes the underlying QTable object so that the user can directly interact with it. Please refer to the Table reference and [documentation]( for more information on how to modify QTable objects.

Additional Data Container Concepts

Alternative field names

If you ask 3 different planetary astronomers which field name or variable name they use for the orbital inclination, you will receive 3 different answers. Good candidates might be 'i', 'inc', or 'incl' - it’s a matter of personal taste. The sbpy developers are aware of this ambiguity and hence DataClass provides some flexibility in the use of field name. This functionality is established through a list of acceptable field names that are recognized by sbpy, which is provided in the sbpy Field Names.

As an example, if your Orbit object has a column named 'incl' but you try to get column 'i', the object will internally check if 'i' is a legitimate field name and what its alternatives are, and it will find that a field name 'incl' exists in the object. The corresponding 'incl' column is then returned. If you try to get a field name that is not connected to any existing field name, a KeyError will be raised.

>>> from import Orbit
>>> orb = Orbit.from_dict({'incl': [1, 2, 3]*u.deg})
>>> orb['i']) 
[1. 2. 3.] deg

The definition of alternative field names is done in the file sbpy/data/, using the list fieldnames. This list is automatically tested for potential naming conflicts, i.e., different properties that share the same alternative field names, and a human-readable list is compiled upon building sbpy.

The full list of field names is available here: sbpy Field Names.

Field conversions

There are parameters and properties that can be used synonymously, a good example for which are an object’s radius and diameter. sbpy acknowledges identities like this by providing internal conversions for such properties. Consider the following example:

>>> from import Phys
>>> import astropy.units as u
>>> data = Phys.from_dict({'d': 10*})
>>> print('{:.1f}'.format(data['d'][0]))
10.0 km
>>> print('{:.1f}'.format(data['radius'][0]))
5.0 km

Note that the radius is not explicitly defined in data, but derived internally upon querying it and added to the internal data table:

>>> print(data.field_names)
['d', 'radius']

Epochs and the use of astropy.time

Epochs and data referring to specific points in time have to be provided as Time objects. The advantage of these objects is their flexibility in terms of format and time scale. Time objects can be readily transformed into a wide range of formats; for instance, Time('2019-07-23 10:49').jd can be used to convert an ISO epoch to a Julian Date.

More importantly, Time provides functionality to transform epochs between different time scales. Hence, every Time object comes with a time scale (UTC is used by default) and can be easily transformed into a different time scale. The following example defines an epoch in UTC and as a Julian Date and transforms it to TDB:

>>> from astropy.time import Time
>>> epoch = Time(2451200, format='jd')
>>> epoch
<Time object: scale='utc' format='jd' value=2451200.0>
>>> epoch.tdb
<Time object: scale='tdb' format='jd' value=2451200.000742876>
>>> epoch.tdb.iso
'1999-01-21 12:01:04.184'

Using Time in DataClass objects is straightforward. The following example builds a simple Obs object from a dictionary:

>>> from import Obs
>>> times = ['2018-10-01', '2018-11-01', '2018-12-01']
>>> obs = Obs.from_dict({'epoch': Time(times), 'mag': [10, 12, 14]*u.mag})
>>> obs 
<QTable length=3>
         epoch            mag
         object         float64
----------------------- -------
2018-10-01 00:00:00.000    10.0
2018-11-01 00:00:00.000    12.0
2018-12-01 00:00:00.000    14.0

The 'epoch' column in obs can be used like any other field or Time object. The following example converts the epoch to TAI and Julian Date:

>>> obs['epoch'].tai.jd  
array([2458392.50042824, 2458423.50042824, 2458453.50042824])

Note that different functions in sbpy have different requirements on the time scale of Time objects. Fortunately, Time objects are able to convert most time scales seamlessly. However, that requires that some user-defined time scale might have to be converted to other time scale for compatibility reasons internally, which also means that outpu t epochs usually follow this forced time scale. In order to notify the user that the time scale has been changed, a TimeScaleWarning will be issued.

Writing object data to a file

DataClass objects can be written to files using to_file:

>>> obs.to_file('observations.dat')

By default, the data are written in ASCII format, but other formats are available, too (list of file formats). Please note that not all file formats support units and meta data. For instance, basic, csv, html, and latex do not provide unit or meta data information. However, fits, cds, daophot, ecsv, and ipac do support units and meta data.