Velocity estimation of GPS base stations considering the coloured noises

Abstract

The optimal velocity estimation of the base sta- tions relies on the optimal or at least realistic function model and stochastic model of the noises in the GPS monitoring position series. In this paper, we first analyse the theoreti- cal effects of periodic signals and different noises on veloc- ity estimates, and numerically analyse the impacts of com- mon mode errors on the different noises. Then we numer- ically analyse the effects of the different types of noises on the velocity estimation using the GPS monitoring posi- tion series of base stations in China. The result shows the importance of the correct noise composition that is specified by the correct stochastic model. We therefore determine the optimal noise composition and its stochastic model by using variance component estimation and w-test theory on the daily monitoring position series with removed common mode errors. The result indicates that the noise of daily monitoring position series consists of both white noise and flicker noise rather than only the white noise in statistics. As a result, the realistic station velocities are computed with the associated noise stochastic model by using a 7-year GPS monitoring position series of 24 base stations in China, comparing with the results obtained using the pure white noise model. The results show that the root mean squared error of velocity derived from the realistic stochastic model is over 6 times larger than that derived from the pure white noise model, which indicates that the pure white model gets a too optimistic velocity estimate. Therefore, from the esti- mated velocities with the realistic noise stochastic model, the real-world tectonic motion trends can be recovered.