Filtering of GPS Time Series Using Geophysical Models and Common Mode Error Analysis

Abstract

In the previous chapters we have discussed various methods to estimate the parameters of the trajectory models for geodetic time series. The observations were written as the sum of a signal plus noise and we emphasized in particular the modelling of the temporal correlated noise in these analyses. In most cases we are interested in the secular motion which is modelled by a linear trend. However, the observations can contain other geophysical signals which need to be included in the trajectory model as well. In this chapter we explain the most common ones such as offsets, seasonal variations and post-seismic relaxation. In addition, in many situations it is beneficial to pre-process the time series before the analysis is performed. We show how the output of various surface loading models can be used to reduce the scattering of the time series. Furthermore, Common Mode and Principal Component Analysis may be applied which again causes a further reduction of the noise and in this way could produce a more accurate estimate of the trajectory model parameters.