CCSDS Navigation Data Messages (NDM) Read/Write


The Consultative Committee for Space Data Systems (CCSDS) develops communications and data systems standards for spaceflight. CCSDS Navigation Data Messages (NDM) is the set of file standards to define common data types such as trajectory, orbit, attitude and conjunction events. These data types are routinely generated and exchanged within and between spacecraft operators, space agencies, researchers, amateurs and commercial companies. As such, accurate definition and common interpretation of the data is crucial (and sometimes mission-critical).

The standard description for each data type is encapsulated in an XML Schema file. This project ccsds-ndm aims to be the reference open-source Python implementation to read and write the NDM XML and KVN files, through an object tree API, auto-generated by these schema files. It supports the up-to-date NDM XML 2.0.0 standard.

The source code is on Github.

Current functionality:




All NDM Types

All NDM Types


All except NDM Combined Instantiation

All except NDM Combined Instantiation


Not specified in CCSDS Standards

Not specified in CCSDS Standards

Usage and Examples

There are two use cases:

  • The ccsds-ndm library reads the NDM file, fills an object tree and offers it to the users. The users will then have to fill their own attitude, orbit or trajectory objects used in their libraries.

  • The user fills an object tree from their own attitude, orbit or trajectory object. The ccsds-ndm library writes the NDM file using this object tree.

For the first use case, reading an OEM file from xml_read_path is as simple as:

>>> cdm = NdmIo().from_path(xml_read_path)

Note that file format (XML or KVN) and data type (e.g. CDM or NDM) are inferred automatically. The output cdm is the object tree for a Conjunction Data Message (CDM). The contents can then be reached going deeper in the object tree as specified in the corresponding NDM Standard. This example shows how to reach the orbit normal position component of the relative state vector:

>>> print(cdm.body.relative_metadata_data.relative_state_vector.relative_position_n)

The data can sometimes be of type NDM Combined Instantiation. This means that there are multiple NDM data bodies (e.g. 2 AEMs, 3 OEMs and one OMM) within a single file. The file reader supports these types as well and they are kept within the Ndm object as individual lists for each of the file types. The following example retrieves the second Omm object in the NDM file and then continues to dive deeper into the object tree to retrieve the eccentricity value.

>>> print(ndm.omm[1]

If the file is of the type NDM Combined Instantiation but there is only a single data (e.g. OMM) in it, the ndm tags are stripped and only the single data is presented to the user.

Filling the objects with data properly requires some care. As the standard is understandably strict, the object tree derived from the XSD files are also rather exacting in how they accept data. Out of these three different ways of inserting data, the third one is invalid:

>>> # This is valid but missing units information
>>> cdm.body.relative_metadata_data.relative_state_vector.relative_position_r = LengthType(700)
>>> # This is the proper way to write data into the object
>>> cdm.body.relative_metadata_data.relative_state_vector.relative_position_t = LengthType(Decimal(800), LengthUnits.M)
>>> # This is invalid
>>> # cdm.body.relative_metadata_data.relative_state_vector.relative_position_r = 600

The output of the above two methods is different on the XML file - the latter is properly transmitting important unit information:

>>>      <relativeStateVector>

Therefore, care must be taken (and standard documents must be kept as a reference) when mapping the user objects into the object tree. Valuable information on units, models and methods can be found there to correctly interpret the data. Also, comments can also be used to provide supplementary information on how this data is generated.

Finally, once filled with the relevant data, the cdm object can be written to xml_write_path in XML data format as:

>>> NdmIo().to_file(cdm, NDMFileFormats.XML, xml_write_path)

The ndm object trees are not very user friendly and most probably will have to be filled by the users’ own equivalent objects (trajectory, orbit, attitude etc.).

Installing ccsds-ndm

The ccsds-ndm package is on PyPI and you can install it simply by running:

pip install ccsds_ndm

You can also install it via conda-forge:

conda install -c conda-forge ccsds_ndm

Do not install ccsds-ndm using sudo.


  • xsData is used to read and write XML files (and also to generate the object tree)

  • lxml to support XML object creation


Please use the DOI for citations. This is the latest version:


Known Issues

  • Some browsers (at least Firefox 82.0 in Dec 2020) strip the namespace information from the XML file when the file is viewed. However, when the file is downloaded directly (e.g. through wget), the namespace information is kept intact. While neither case causes issues with ccsds-ndm, the former case may result in the XML viewers to report namespace errors. (Hat tip to Juan Luis Cano Rodríguez for spotting this)

Indices and Tables