面向导航增强的低轨卫星钟差确定及预报方法研究

A study of LEO satellite clock determination and prediction for navigation augmentation

低轨导航增强是未来导航发展的重要趋势,而高精度低轨卫星钟差是实现低轨导航增强的必要条件。基于Sentinel-6A卫星,对低轨卫星钟差特性进行了分析,给出了钟差确定方法及影响因素,介绍了顾及钟差特性的低轨卫星钟差预报方法。实验表明,低轨卫星钟差含有多个周期项,给低轨卫星建模和预报带来了困难。与使用运动学定轨模型相比,基于简化动力学的定轨模型可显著提升低轨卫星钟差精度;当基于运动学模型确定低轨卫星钟差时,相较于使用GPS单系统数据,多GNSS观测数据可提升低轨卫星钟差精度。研究表明,基于GPS和Galileo观测的Sen-tinel-6A卫星钟差精度相较于GPS单系统钟差精度改善了36%,同时,所使用的GNSS产品精度与低轨卫星钟差精度密切相关。利用顾及卫星钟差特性的低轨卫星钟差预报方法,当预报时长小于1min,低轨卫星钟差预报精度(预报与解算值之差的RMSE)在0.1ns之内,当预报时长小于5min,预报精度在0.3ns之内,随着预报时长的增长,预报精度显著下降。

Augmentation with low Earth orbit(LEO)satellites has become a hot topic in future positioning,navigation and timing services,and high-precision LEO satellite clock product is an essential condition for the realization the LEO-augmentation.Using the data from the Sentinel-6A satellite,this paper analyzed its clock characteristics,explained the clock determination method and influencing factors,and introduced an LEO satellite clock prediction method that considers the clock characteristics.Experiments show that the LEO satellite clocks contain multiple periodic terms,which brings difficulties to clock modeling and prediction.Compared with the case using the kinematic precise orbit determination(POD)model,the reduced-dynamic POD model can significantly improve the LEO satellite clock precision.In the case of using the kinematic POD model,the multi-GNSS scenario delivers better LEO satellite clock precision than the GPS-only scenario.The research result shows that Sentinel-6A satellite clock accuracy based on GPS and Galileo observation is improved by 36%compared with GPS single system clock error accuracy.The precision of the LEO satellite clocks is also found to be closely related to that of the GNSS products used.Using the LEO satellite clock prediction method that considers the satellite clock characteristics,the RMSE of the predicted and estimated clocks is within 0.1 ns and 0.3 ns,respectively,for a prediction time shorter than 1 min and 5 min.The prediction precision degrades significantly with the increasing prediction time.