EMD在GNSS时间序列周期项处理中的应用  

Application of EMD to GNSS Time Series Periodic Term Processing

作　　者：刘希康[1,2] 丁志峰[1] 李媛[2] 刘志广[2] LIU Xikang;DING Zhifeng;LI Yuan;LIU Zhiguang(Institute of Geophysics,China Earthquake Administration,Beijing 100081,China;The First Crust Monitoring and Application Center,China Earthquake Administration,Tianjin 300180,China)

机构地区：[1]中国地震局地球物理研究所,北京100081 [2]中国地震局第一监测中心,天津300180

出　　处：《武汉大学学报（信息科学版）》2023年第1期135-145,共11页Geomatics and Information Science of Wuhan University

基　　金：国家重点研发计划(2018YFC1503606);中国地震局青年震情跟踪课题(2020010226,2021010511);国家自然科学基金(41472180)。

摘　　要：全球导航卫星系统(global navigation satellite system,GNSS)位置时间序列蕴含有丰富的构造和非构造变形信息,具有成分复杂、建模困难、非构造信息难以有效分离等特点,利用自适应的经验模态分解(empirical mode decomposition,EMD)方法对川滇地区24个GNSS连续站时间序列作周期项修正。结果表明,周期项的修正十分必要,EMD方法能够根据每个台站信号的自身特性,自适应地提取不同频率、振幅的周期成分,这也更符合实际情况;相较于谐波模型,EMD方法对原始时间序列在N、E、U方向的改正均更加精确有效。使用修正后的连续站时间序列模拟流动观测,发现经过5~6 a/期观测即可得到相对可靠的台站运动速度,并通过距离较近的实际连续观测站对流动观测站周期项改正,验证了EMD方法的稳定性和可靠性,这也为流动GNSS观测实施、周期修正和资料使用提供了参考意见和理论依据。Objectives:The position time series of global navigation satellite system(GNSS)contain rich tectonic and non-tectonic deformation information,and the complex components are difficult to accurately model and effectively separate non-tectonic information.It is very important to remove the non-tectonic deformation information for the accurate and effective application of the observation data.Methods:Empirical mode decomposition(EMD)is an adaptive time-frequency processing method to correct the period term of time series at 24 GNSS continuous stations and 2 mobile stations in Sichuan Province and Yunnan Province.Results:The results show that the correction of the period term is necessary,and the EMD method can extract the different frequency and amplitude period components adaptively according to its own characteristics of signal at each station.Compared with the original time series,the average root mean square error after correction is reduced significantly in N,E and U directions.And EMD method is more accurate and effective than harmonic model modification.We use the modified continuous station time series to simulate the mobile observations,and find that relatively reliable motion velocities can be obtained after an observation period of 5-6 years.The stability and reliability of EMD method are verified by correcting the period term of the mobile station with the actual continuous observation station at a close distance.Conclusions:This paper provides a reference and theoretical basis for the implementation of mobile GNSS observation and the correction of observation data.

关 键 词：全球导航卫星系统 周期运动 流动观测 谐波模型 经验模态分解 

分 类 号：P228[天文地球—大地测量学与测量工程]