This repository contains a dataset of Two Line Element (TLE) data for 15 satellites. Its purpose is to serve as a benchmark dataset for studies examining changes in satellite orbits over larger time scales. It also contains the ground-truth manoeuvre timestamps for these satellites. Highly-accurate positional data from DORIS ground beacons is also included for some satellites.

Please see the paper: Wide-scale Monitoring of Satellite Lifetimes: Pitfalls and a Benchmark Dataset for a detailed description of the dataset.

The following table lists the satellites in the dataset:

The folder processed_files contains the TLE files as well as the manoeuvre timestamps. The TLE files conform to the TLE standard as defined by NORAD. A good starting point for working with them is the Skyfield library

Some brief advice on getting started using Skyfield in conjunction with this data. You can use the tle_file() function to load the TLE file into a list of Skyfield EarthSatellite objects. (see here for some general documentation on loading TLE files). Each EarthSatellite object is associated with the data published for one satellite epoch, which corresponds with a pair of lines in the original TLE file. One can call the .at() function on these objects to get the satellite position in cartesian geocentric coordinates. One can then use the osculating_elements_of() function to convert this position into osculating keplerian elements. Moreover, by providing times other than the published epoch to the at() function, it can also perform propagation of the position and osculating elements to different points in time. If one is interested in monitoring changes in satellite orbits over longer time scales (weeks to years), we suggest that using the at() function is counter-productive, due to the fact that it uses models of orbiting satellites to add in non-Keplerian components of the orbit. This is discussed in detail in the paper Wide-scale Monitoring of Satellite Lifetimes: Pitfalls and a Benchmark Dataset. Rather, we suggest using the mean Keplerian elements that are presented in the raw TLE records. One can still use Skyfield to conveniently load and manipulate the data. The mean Keplerian elements are stored as instance variables of the model variable of each EarthSatellite object. Note that these mean elements are also propagated and updated each time the at() method is called on the EarthSatellite object.

The manoeuvre timestamps are stored in YAML files, also found in the directory processed_files. These files also contain the satellite's SATCAT number as a piece of useful metadata. The timestamps are stored in an item named manoeuvre_timestamps. This is a list of strings of ISO8601 timestamps which contain the UTC time and date of each manoeuvre.

The accurate positional data from the DORIS ground beacons is stored in the DORIS_beacon_positions directory in CSV files.

The directory plotting contains the scripts for creating the plots in the paper Wide-scale Monitoring of Satellite Lifetimes: Pitfalls and a Benchmark Dataset.

• plotting_TLE_propagation_residuals.py creates the plots in figures 2, 11, 12 and 13.
• plotting_osculating_mean_and_DORIS.py creates the plots in figure 3.
• plotting_osculating_and_mean_diffs_distribution.py creates the plots in figure 4.
• plotting_osculating_propagation_to_mean_diff.py creates the plots in figure 5.
• plot_satellite_liftimes_gantt.py creates the plot in figure 6.
• plotting_propagation_accuracy_TLE_to_SP3.py creates the plots in figure 7.
• plotting_DORIS_fourier_analysis.py creates the plots in figures 8, 9 and 10.

The directory utilities contains some useful scripts for inspecting the data.

you will need to change the file paths in params.py.

The directory dataset_formatting contains scripts for manipulating the raw data into the compiled files.