Evaluation of strategies for the ultra-rapid orbit prediction of BDS GEO satellites

The quality of BeiDou Navigation Satellite System (BDS) Geostationary Earth Orbit (GEO) ultra-rapid products is unsatisfactory because GEO satellites are nearly stationary relative to ground stations. To optimize the quality of these ultra-rapid orbit products, we investigated the effects of the fitting arc length, an a priori Solar-Radiation Pressure (SRP) model, and the along-track empirical acceleration on the prediction of BDS GEO satellite orbits. The predicted orbit arcs of 24-h were evaluated through comparisons with the corresponding observed orbit arc and Satellite Laser Ranging (SLR) observations. In both eclipse and non-eclipse seasons, accuracy of the orbit predictions obtained using a 48-h fitting arc length were better than those obtained using 24-h and 72-h fitting arc lengths. Although the overlapping precision of predicted orbits exhibited no obvious improvement when an a priori SRP model was employed, the systematic bias in the SLR residuals was significantly reduced. Specifically, the mean value of SLR residuals decreased from -0.248 m to -0.024 m during non-eclipse seasons and from -0.333 m to -0.041 m during eclipse seasons, respectively. In addition, when an empirical acceleration in the along-track direction was introduced, the three-Dimensional Root-Mean-Square (3D RMS) of overlapping orbits during eclipse seasons decreased from 2.964 to 1.080 m, which is comparable to that during non-eclipse seasons. Furthermore, the Standard Deviation (STD) of SLR residuals decreased from 0.419 to 0.221 m during eclipse seasons. The analysis of SRP estimates shows that the stability of SRP parameters was significantly enhanced after the introduction of along-track empirical acceleration in eclipse seasons. The optimal BDS GEO ultra-rapid orbit prediction products were yielded by using a 48-h fitting arc length, an a priori SRP model and an along-track empirical acceleration.