摘要
为了进一步研究青藏块体东北缘全球定位系统(GPS)测站垂向位移序列的最佳噪声模型及对测站垂向速度估计的影响,提出一种青藏块体东北缘GPS垂向时间序列噪声分析方法:选取青藏块体东北缘27个中国大陆构造环境监测网络(CMONOC)GPS测站,利用不同的噪声模型对测站垂向位移时间序列进行噪声分析;并探讨赤池信息准则(AIC)、贝叶斯信息准则(BIC)、时间优化贝叶斯信息准则(BICtp)和修正贝叶斯信息准则(BICc)确定的最优噪声模型之间的差异及对测站垂向速度估计的影响。结果表明,AIC确定的最佳噪声模型主要是广义高斯马尔可夫噪声(GGM)+自回归滑动平均噪声(ARMA1)、GGM+白噪声(WN)和WN+闪烁噪声(FN),分别占总测站数的29.6%、22.2%和18.5%,而BIC、BICtp和BICc准则确定的最佳噪声模型主要为WN+FN,分别占总测站数的74.1%、66.7%和70.4%,其次是GGM,分别占测站数的22.2%、22.2%和14.8%;BIC、BICtp、BICc在确定最佳噪声模型过程中,对GGM+ARMA1和GGM+WN噪声模型的影响较大,并且BICc对随机漫步噪声的敏感程度高于BIC、BICtp;不同信息准则确定的青藏块体东北缘垂向速度及不确定度存在一定的差异,但整体垂向趋势基本相同,西南部测站呈现下降趋势,东北边缘测站呈现上升趋势,但上升或下降的幅度相对较小。
In order to futher study the optimal noise model for global positioning system(GPS)station vertical displacement sequences at the northeast margin of Qinghai-Xizang block and the effect on station vertical velocity estimation,the paper proposed time series noise analysis method of GPS vertical observations in northeast margin of Qinghai-Xizang block:the GPS coordinate time series from 27 crustal movement observation network of China(CMONOC)reference stations in the northeast margin of Qinghai-Xizang block were selected,and different noise models were used to analyze the vertical displacement time series of the stations;then,the differences among the optimal noise models determined by Akaike information criterion(AIC),Bayesian information criterion(BIC),Bayesian information criterion with time penalty(BICtp)and Bayesian information criterion with correction(BICc)and their effects on vertical velocity estimation were discussed.Results showed that the optimal noise models determined by AIC would be mainly:generalized Gaussian Markov noise(GGM)+auto-regressive moving average of order 1(ARMA1),GGM+white noise(WN)and WN+flicker noise(FN),accounting for 29.6%,22.2%and 18.5%of the total number of stations,respectively,while the optimal noise models determined by BIC,BICtp and BICc criteria would be mainly WN+FN,accounted for 74.1%,66.7%and 70.4%of the total number of stations,respectively,followed by GGM,accounting for 22.2%,22.2%and 14.8%of the total number of stations,respectively;moreover,it was found that BIC,BICtp and BICc would have greater influence on GGM+ARMA1 and GGM+WN models in the process of determining the optimal noise model,and BICc would be more sensitive to random walk noise than BIC and BICtp;furthermore,there would be some differences in the drooping speeds and uncertainties at the northeastern margins of the Qinghai-Tibet block determined by different information criteria,but the overall trend of drooping could be basically the same,with a decreasing trend at the southwestern stations and an increasing trend at the northeastern edge stations,meanwhile the magnitude of the increase or decrease would be relatively small.
作者
蔡晓军
张子哲
马涛
CAI Xiaojun;ZHANG Zizhe;MA Tao(Gansu CSCEC Municipal Engineering Investigation and Design Institute Co.Ltd.,Lanzhou 730000,China)
出处
《导航定位学报》
2024年第6期128-137,共10页
Journal of Navigation and Positioning
关键词
全球定位系统(GPS)
时间序列
噪声模型
极大似然估计
信息准则
global positioning system(GPS)
time series
noise model
maximum likelihood estimation(MLE)
information criteria
