Session A4b: LEO PNT 2: Precise Orbit Determination and Interference

Preliminary Results of Precise Orbit Determination for Collision Probability Estimation of LEO Satellites: A COSMIC-2 Case Study
Binson Shih-Pin Lee, Loren Chang, National Central University
Location: Prince Edward Room
Date/Time: Wednesday, Apr. 15, 3:39 p.m.

Accurate Conjunction Assessment (CA) and collision probability (Pc) estimation are critical for Space Situational Awareness (SSA). While frameworks such as NASA’s CARA (Conjunction Assessment Risk Analysis) provide standard methodologies for evaluating conjunction threats, the reliability of these metrics is highly dependent on the fidelity of the orbital state covariances. This study investigates how Precise Orbit Determination (POD) utilizing space-borne GNSS measurements impacts Pc estimation during extreme space weather events, specifically comparing robust EKF solutions against conventional Two-Line Element (TLE) derived baselines for satellites in Low Earth Orbit (LEO).

The methodology focuses on processing FORMOSAT-7/COSMIC-2 microsatellite raw observations through a robust Extended Kalman Filter (EKF) equipped with a pseudo-measurement soft constraint. This approach is designed to bound orbital uncertainty during GNSS signal dropouts caused by intense ionospheric scintillation. We analyze a 48-hour period encompassing the historical May 2024 G5 geomagnetic storm, which serves as a severe stress test for LEO navigation. By mitigating the effects of ionospheric signal degradation, this filtering strategy ensures that state error covariances remain stable and bounded even when measurement updates are temporarily unavailable.

Our results demonstrate that unconstrained dynamic propagation during a storm-induced data gap leads to a peak position error of approximately 7.5 km. This massive uncertainty induces a dangerous Probability Dilution effect, artificially suppressing the Pc to near 10?15 and creating a catastrophic false sense of security. In contrast, the proposed EKF method maintains a stable average 3D accuracy of 99.7 meters. Even during peak storm intensity, transient error peaks are successfully bounded within 840 meters. By preventing probability dilution, the robust EKF restores the dynamic sensitivity of the risk assessment, effectively capturing threat peaks that approach the actionable 10?4 threshold. These findings establish a validated framework for autonomous navigation and risk assessment, significantly enhancing the survivability of LEO satellites in the increasingly congested orbital environment.