| In the past two decades,with the development of space geodetic technology,GNSS crustal movement observation network has been gradually established,and continuous station coordinate time series data can be obtained.At the same time,as early as the1990 s,the Japanese Academy of land and geography began to establish a continuous observation network geonet.As of 2013,1273 observation stations have been built in Japan,with an average distance of about 20 kilometers.At the same time,the data of each station is easy to obtain and is an ideal area for research,which has attracted many scholars to study this area.Taking GEONET network as the research object,this paper obtains the data of GNSS reference station in Japan for six consecutive years from 2005 to 2010,and studies and analyzes the coordinate time series.The specific research contents are as follows:(1)This paper discusses the principle and method of time series analysis,applies principal component analysis to the extraction of common mode error,and summarizes the process of principal component analysis.Firstly,the coordinate time series are preprocessed,including gross error elimination,various discontinuous and step correction methods,interpolation of missing data and least square fitting modeling of coordinate time series.Finally,the residual coordinate time series are obtained.(2)The principal component analysis method is used to process the station coordinate time series.According to the F test and the consistency of the principal component station response,the first principal component is selected as the common mode error,in which the contribution rates of the first principal component in E,N and U directions are 35.24%,40.87% and 28.54% respectively.At the same time,the temporal and spatial distribution of common mode error is analyzed.The analysis shows that the spatial response of the first principal component in E,N and U directions of each station is consistent in all stations,and the average spatial response is 77.53%,77.68% and 73.24% respectively.Then the common mode error is eliminated from the station coordinate time series,and the values of various parameters in the coordinate time series before and after spatial filtering are estimated again by Hector software.The results showed that the standard deviations of E,N and U directions were improved by17.25%,19.40% and 5.40% on average;At the same time,it can be seen that the uncertainty of annual amplitude and half annual amplitude of all stations has decreased,and the uncertainty of annual term amplitude in E,N and U directions has decreased by27.48%,24.92% and 10.13% respectively.The half year term amplitude uncertainty in E,N and U directions decreased by 26.88%,25.65% and 9.53% respectively.It shows that the error of the periodic term can be greatly reduced after the common mode error is eliminated,the accuracy is improved,and the signal originally submerged in the error is more obvious.(3)Finally,the influence of eliminating common mode error on the combination of optimal noise model and velocity uncertainty is analyzed.It can be seen that the optimal noise models of E,N and U before eliminating common mode error are WN+FN and WN+PLN;After eliminating the common mode error,WN+FN in E direction decreased from 66.67% to 35.56%,and WN+RWN+FOGM increased from8.89% to 22.22%;In the N direction,WN+FN decreased from 60% to 42.22%,and WN+RWN+FOGM increased from 6.67% to 22.22%;In U direction,WN+FN increased from 40% to 57.78%,and WN+ PLN decreased from 44.44% to 26.67%;The E directions optimal noise model becomes WN+FN and WN+RWN+FOGM;At the same time,it can be obtained that most of the velocity changes in E,N and U directions are 0.05mm/a;The velocity uncertainty in E,N and U directions decreased by 28.11%,26.14% and 10.83% respectively;In the horizontal direction,the movement rate direction of most stations is southeast;(16 pictures,10 tables,60 references)... |
PDF Full Text Request