Application of Attributables to the Correlation of Surveillance Radar Measurements

TL;DR

This paper investigates how fitting attributables to radar surveillance data affects the correlation of tracklets for space object cataloging, introducing a new coordinate system and filters to improve accuracy and reduce false positives.

Contribution

It introduces a singularity-free coordinate system for attributable fitting and demonstrates its effectiveness in correlating radar tracklets with enhanced accuracy.

Findings

Improved attributable fitting results with the new coordinate system.

Successful correlation of tracklets up to three minutes long.

Effective filters to eliminate false positive correlations.

Abstract

Space surveillance by radar is especially used for the low Earth orbit to maintain a database, also called catalogue, of objects on orbit. Among others, surveillance radars which are constantly scanning a region of interest in the sky are used for this purpose. The detections from such a radar which cannot be assigned to an already known catalogue object might not contain enough information to obtain a reliable initial orbit for a new catalogue entry from a single measured pass, also called tracklet. Instead, two tracklets can be combined to improve the quality of the initial orbit which leads to the correlation problem. This means that it has to be tested whether two tracklets belong to the same object and an initial orbit has to be derived by combining the tracklets. A common approach to condense the information in the tracklet is fitting them with so-called attributables. Because radar observations include different types of observables, the fitting of these attributables has to be considered as an important part of the entire correlation process. This paper analyses the effect of the attributable fitting considering the achieved accuracy and influence on the tracklet correlation. A new singularity-free coordinate system is introduced, which improves the results of the fitting and correlation. Finally, a test on a simulated survey scenario introduces two additional filters to remove false positive correlations. It is shown that the attributable-based approach can be applied successfully to tracklets of up to three minutes length with different detection frequencies.