Research Of The Application Of Continuously Operating Reference Systerm In Ground Deformation Monitoring

In December 20,2010, the concept of geographical condition monitoring was put forward in our country. GPS observation technology existing in geophysical research for many years has been accumulating a large number of GPS data. These data was successfully applied in the study of horizontal crustal movement, providing important material for studying crustal movement and earthquake prediction in China. Continuously Operating Reference System can constantly obtain observation data and has overwhelming advantage in the earth surface deformation monitoring.However, noise of time series in GPS elevation direction is larger than that in the horizontal direction, and the factors which have influence on vertical movement are also more complicated. So the application of Continuously Operating Reference System in crustal vertical movement is under some restrictions.This paper aims to analysis the periodic motion changes and rules of CORS sites elevation coordinate, then the real information of ground surface deformation can be extracted. And research the causes qualitatively in the paper. The following research was done in the paper:(1)Studied the methods in interpolating for the continuous missing data of the CORS sites.Compare BP neural network method with Lagrange method and cubic spline method in interpolating for missing data. The result indicates that the BP neural network method provides time series data with a better solution for the continuous more missing point.(2) Analyzed the periodic variation rule of time series in GPS elevation direction.Using power spectrum analysis and wavelet transform analysis method to study the periodic variation rule of time series in GPS elevation direction. Pick up the periodic and linear trend of time series and use the smooth time series to build a model. Through the analysis found that most CORS sites exists in semi-annual periodic motion and annual periodic motion mainly, especially with the annual periodic motion.(3) Established prediction models of CORS sites at elevation direction.At first predict the change of GPS elevation coordinate by using time series model and polynomial fitting model, and then combined BP neural network with time series model to build a new prediction model. The result shows that the new fusion model has higher precision.(4) Researched the factors which have influence on the periodic motivation of GPS elevation time series. Analyze the impact of ocean tides, atmospheric pressure and bedrock thermal expansion qualitatively.