Data association experiments using real radar data
Abstract
The monitoring of the space object population, especially the space debris population, is an important aspect of space safety to enable the safe operation of near-Earth space missions. One of the main components of such a space surveillance system is the maintenance of a space object database, commonly called catalogue, which can be used for e.g. conjunction assessment and re-entry casualty risk estimation. The build-up and maintenance of such a database requires the continuous, automated processing of observations to create new database entries and maintain the accuracy of existing ones.
To support the initialisation of new objects into the database, it is advantageous to use multiple passes of an object to have a more reliable orbit compared to that from a single pass. To identify such associated passes, previous work by the authors include the development of a method to test whether two measured passes, called tracklets, originate from the same object in which case they are called to be correlated. This method was tested on simulated measurement data and showed promising results. The present paper uses real radar measurements from two LeoLabs radars in the United States to further validate and characterise the method, mainly for objects in low earth orbit. The first tests confirm that the method’s initial orbit determination which includes the secular J2 -perturbation works for the maximum time span of circa 24 days between two tracklets. The correlation which also requires the selection of the correct number of revolutions between the two tracklets works reliably for 1-2 days before the selection of a wrong number of revolutions increases, which is also dependent on the geometry of the two passes. Further experiments test the sensitivity of the method to e.g. distributed stations, atmospheric drag and manoeuvres. Finally, different approaches to a cold-start of a database are tested using either least squares orbit determination or a graph network for object confirmation. Overall results show that the developed methods works well with real radar data and can contribute to the build-up of a space object database.
To support the initialisation of new objects into the database, it is advantageous to use multiple passes of an object to have a more reliable orbit compared to that from a single pass. To identify such associated passes, previous work by the authors include the development of a method to test whether two measured passes, called tracklets, originate from the same object in which case they are called to be correlated. This method was tested on simulated measurement data and showed promising results. The present paper uses real radar measurements from two LeoLabs radars in the United States to further validate and characterise the method, mainly for objects in low earth orbit. The first tests confirm that the method’s initial orbit determination which includes the secular J2 -perturbation works for the maximum time span of circa 24 days between two tracklets. The correlation which also requires the selection of the correct number of revolutions between the two tracklets works reliably for 1-2 days before the selection of a wrong number of revolutions increases, which is also dependent on the geometry of the two passes. Further experiments test the sensitivity of the method to e.g. distributed stations, atmospheric drag and manoeuvres. Finally, different approaches to a cold-start of a database are tested using either least squares orbit determination or a graph network for object confirmation. Overall results show that the developed methods works well with real radar data and can contribute to the build-up of a space object database.
Date Issued
2020
Publication Type
Conference Item
Subject(s)
Language(s)
en
Additional Credits
Title of Event
Access(Rights)
restricted