Enhancing real-time LEO kinematic precise orbit determination during ionospheric disturbances by optimizing stochastic model and cycle-slip detection

Abstract

Real-time kinematic precise orbit determination (KPOD) is inherently a precise point positioning (PPP)-based method for estimating the precise position of low Earth orbit (LEO) satellites, which relies solely on the onboard global navigation satellite system (GNSS) observations without considering the dynamic model. The high precision and continuous tracking capability of GNSS enable excellent KPOD of LEO satellites. However, the performance of KPOD is severely degraded during ionospheric disturbances, as the geometry-free (GF) cycle-slip detection becomes less reliable and the standard observation stochastic model proves inadequate for characterizing scintillation signals. To address these challenges, two new metrics, the acceleration of total electron content (AOT) and the AOT index (AOTI) are defined in this study. AOT represents the second-order time derivative of the total electron content (TEC), while AOTI is the standard deviation of AOT. AOT and AOTI are proposed to address the unreliable cycle-slip detection and the inadequate stochastic model, respectively, during ionospheric disturbances. By eliminating the first-order variation of TEC, AOT-based GF cycle-slip detection method can prevent the over-detection of spurious cycle slips without increasing the detection threshold. When AOT remains constant over a short interval (e.g., 10 s), AOTI serves as a proportional proxy for phase observation noise, permitting its real-time, on-the-fly estimation. The implementation of AOT and AOTI yields a significant improvement in real-time KPOD of the Swarm-A satellite. During ionospheric disturbances over the last five years, the average improvements in the root mean square (RMS) values computed from orbit errors below the 95th percentile are 21.4% (A), 17.2% (C), 31.6% (R), and 26.1% (3D). Corresponding improvements in the 95th percentile values are 29.7% (A), 22.4% (C), 42.4% (R), and 37.0% (3D). It validates the efficacy of AOT and AOTI as robust metrics for enhancing GNSS-based KPOD of LEO satellites during ionospheric disturbances. Besides, a more substantial reduction in the 95th percentile value compared to the RMS value indicates that the method specifically targets and mitigates the larger orbit errors.