Research On Analysis Of Noise Characteristics And Its Influence On Station Motion For GEONET GNSS Coordinate Time Series

In the 1990s,to carry out precise positioning and crustal deformation monitoring,the Geospatial Information Authority of Japan established the GNSS Earth Observation Network System(GEONET).After more than 20 years of development,GEONET has become one of the most dense continuous GNSS observation networks in the world.GEONET GNSS coordinate time series contains abundant geophysical signals.These signals and noise intersect and fuse with each other to form a complex surface movement.Therefore,the research of GEONET GNSS coordinate time series is great significance for maintaining the dynamic earth reference frame and studying crustal deformation.The key of signal modeling is the nature of noise.Different from other signal modeling,GNSS coordinate time series contains colored noise.Colored noise will cause serious interference to signal extraction and accuracy evaluation.How to accurately extract signals from GNSS coordinate time series polluted by colored noise is the difficulty of research.Taking GEONET GNSS coordinate time series as the research object,this paper mainly explores the time-frequency characteristics of colored noise,noise model of GNSS coordinate time series,function model of GNSS coordinate time series,co-seismic and post-seismic deformation estimation considering colored noise.The main achievements of this paper are as follows:(1)The standard linear trajectory model(SLTM)was used to model 152 high-quality GEONET GNSS coordinate time series before the earthquake,its parameters are estimated using MLE method and the best noise model is identified according to the conservative estimation criterion.The results show that the E and N components are mainly white noise(WN)+ flicker noise(FN)and WN + power-law noise(PL),which accounting for 89% of the total horizontal component noise;The U component is mainly represented by WN + PL,accounting for 84% of the total vertical component noise;Ignoring colored noise will make the uncertainties of velocities estimated too optimistic,and there is a time-varying difference in the ratio of uncertainties of velocity estimated by different noise models,that is,the ratio of uncertainty of velocity increases with the increase of time series span.(2)Considering the effect of large earthquakes on GEONET GNSS coordinate time series,a method of constructing GEONET GNSS coordinate time series function model by combining MLE and nonlinear least square(NLS)algorithm is proposed.Firstly,MLE is used to analyze the noise characteristics of GNSS coordinate time series and construct a stochastic model.Then,NLS based on colored noise is used to estimate the characteristic timescale of post-seismic transient,so as to construct function model of the GNSS coordinate time series.Finally,MLE is used to estimate the coefficient parameters for the function model.The results show that compared with the traditional method,this method improves the convergence of NLS algorithm,reduces the number of iterations by 90%,and ensures the reliability of the function model.(3)The coordinate time series functional models of 83 GEONET GNSS reference stations in Tohoku are constructed using the combining MLE and NLS algorithm,and the noise characteristics before and after the earthquake are identified using Bayesian information criteria(BIC).The results show that before the 2011 Tohoku-Oki earthquake,the noise characteristics of GEONET GNSS coordinate time series can be described as WN + FN and WN + PL;After the earthquake,17% of the time series noise characteristics can be described as WN + FN + random walk noise(RW).The effect of colored noise on co-seismic displacement shows an obvious spatial pattern.In the northern of Tohoku,considering white noise model alone will underestimate the co-seismic displacement by 20 mm;Colored noise generally has little effect on co-seismic displacement in the south of Tohoku.Different from co-seismic displacement,the effect of colored noise on post-seismic displacement shows strong randomness in spatial domain.Considering only white noise model,the post-seismic displacement deviation caused by white noise model is as high as 0.5 m,and the deviation greater than 0.2 m is randomly distributed in the spatial domain.(4)Based on the optimal noise model,the station motion parameters before and after the earthquake are estimated for Tohoku.By comparing the station motion parameters before and after the earthquake,the effect of the 2011 Tohoku-Oki earthquake on the movement characteristics of GEONET GNSS reference station is analyzed.The 2011 Tohoku-Oki earthquake has no significant effect on the GEONET site velocities.95% of the velocity residuals(the difference of site velocities before and after the earthquake)are less than its uncertainty,and the rest residuals are basically distributed in the northern region with the smallest deformation displacement.After the 2011 Tohoku-Oki earthquake,the annual movement of GEONET GNSS reference station has no obvious change,and the semiannual movement has weakened.Among them,the amplitude of semiannual periodic signal of E and N component decreases by 0.5 mm and 0.3 mm respectively.The co-seismic displacement decreases sharply with the increase of the distance from the epicenter.The high value area is2.5 ～ 4 m,the medium value area is 1.5 ～ 2.5 m,the low value area is 0.5 ～ 1.5 m,and the average direction is ES30°.Different from the co-seismic displacement,the post-seismic displacement decreases slowly with the increase of the distance from the epicenter,the magnitude is about 0.5 ～ 1.7 m,and the average direction is ES26°.