顾及空间相关性的GNSS坐标序列插值比较

Comparison of interpolation methods for GNSS coordinate time series considering spatial correlation

针对常规的全球卫星导航系统(GNSS)坐标序列缺失数据插值方法多基于单站,且只适用于缺失数据较少的情形,提出1种适用于数据缺失率高,尤其是数据连续缺失时的数据插值方法:利用基于多站的正则期望最大化算法(RegEM)和克里金卡尔曼滤波(KKF)算法,对实测数据和不同比例连续缺失的模拟数据进行插值;同时与多通道奇异谱分析(MSSA)的插值结果进行比较,这3种方法均顾及了空间相关性。实验结果表明,对于含有连续缺失的GNSS坐标序列,RegEM的插值效果最好,且能够较好地还原GNSS坐标序列的噪声水平,保留的方差也最大,在细节方面处理得更好,而KKF插值效果次之,MSSA插值结果最差。

Aiming at the problems that traditional interpolation methods of GNSS coordinate time series missing data are usually based on single station,and only applicable to the cases with few missing data,the paper proposed an interpolation method suitable for high missing rate,especially continuously missing data:the measured data and the simulated data with continuously missing data of different proportions were interpolated by using algorithms of RegEM and KKF based on multi-station,and the results were compared with that of MSSA interpolation.The spatial correlation was considered by all the three algorithms.Experimental results showed that:for GNSS coordinate time series with continuous absence,the interpolation effect of RegEM method would be better than that of other two methods,and the former could restore the noise amplitude of GNSS coordinate time series well,and retain the largest variance with best handling the details among the three methods;while the interpolation effect of KKF method would be the second,and the MSSA the third.