Height Time Series Analysis Of Continuous GPS In Tianshan

Early1990s, with the establishment of IGS, GPS global positioning system began to be used in the work of Geophysical Research. Geoscience research institutions of various countries in the world began to establish GPS continuous tracking station in various countries and regions. With a large number of continuous GPS stations came along, we have accumulated a large amount of data, makes GPS a substantial increase in the feasibility and reliability of Geophysical Research. The UNAVCO Facility established a large number of continuous GPS tracking station worldwide. In this thesis, numbers of consecutive stations observation data in Tianshan region were used.First half of this article describes the basic theory of time series analysis and time series models, introduces a variety of commonly used time series analysis method, such as least-squares linear fit, the calculation of the power spectral density and the wavelet multi-resolution analysis.Later in this article, we screened11continuous tracking stations which have better continuity of data. The Bernese software has been used to get the original elevation coordinate time series of those continuous stations about ten years. Then used the method of least squares linear fit to extracted the linear velocity from the original coordinates time sequences, and removed the linear movement trend from the original sequences to get the residual sequences without linear trend. Through above calculations, we found that the slopes of the linear fit of11stations are positive, showing a good regional consistency in size, which illustrated that there is a overall uplift trend in the region in the past ten years.Subsequently, we did spectral analysis to the residual series, and used Welch method to calculate the indexes of power spectrums. The analysis results showed that, except a small number of stations (3) whose noise model can be expressed by white noise+flicker noise, the others need to use white noise+flicker noise+random walk model to describes the noise. Finally, we did wavelet multi-resolution analysis of time series, and used Coiflets filter to extract the annual cycle and semi-annual periodic terms of each tracking stations, and counted of the amplitude and the maximum month of the anniversary of the entry of each station. The analysis revealed that each tracking station has obvious seasonal change, and this change showed some regional characteristics. This seasonal trend is caused by a variety of geophysical factors. Thus when making high-precision monitoring of crustal deformation, we should choose the same season of GPS observations to exclude the impact of seasonal error.