The influence of different types of noise on the velocity uncertainties in GPS time series analysis

Abstract

Due to the development of GPS technology, nowadays we are able to determine geodynamic plate motion around the world. Using an appropriate noise technique and understanding all the stochastic effects we are able to do a proper GPS time series analysis in which the sources of noise can be classified as: white noise, flicker noise and random walk. We study the area from the Caribbean Sea, taking the data from two GPS stations for a period of 7.5 years. We use spectral analysis and Maximum Likelihood Estimation. In the first part of the analysis we simultaneously estimate the velocity and amplitudes of the noise with integer spectral index and in the second part we estimate the spectral index. The noise model that presented the higher values of the log likelihood is a combination between power law and white noise which best describes the noise characteristics of all three components. In all the cases the noise amplitudes presented higher values for the vertical component. Also, the rate uncertainties for the power law plus white were higher by a factor of 10, then by using only white noise model.