Noise characterization of GPS time series from the second IGS reprocessing campaign

Abstract

By late 2013, the Analysis Centers (ACs) of the International Global Navigation Satellite System Service (IGS) begun the second reanalysis of the full history of Global Positioning System (GPS) data collected by the IGS global network since 1994 in a fully consistent way using the latest models and methodology. In this thesis, GPS time series of selected sites from different analysis centers were analysed with the software CATS. The purpose of this thesis is to discuss noise content using maximum likelihood estimation (MLE) and determine a noise model that can best describe the noise process in GPS time series of 20 years or more.

The noise was assumed to be a combination of white noise and power-law noise. In the first part, we used fixed noise models with integer spectral indices for the MLE. The three noise models are: white noise only, white noise plus flicker noise (ƙ = -1) and white noise plus random walk noise (ƙ = -2). The estimated MLE values of different noise models were compared to determine the preferred noise model. Another noise parameter, amplitude, was also estimated. Latitude dependence of the noise amplitudes and the influence of data cleaning on the noise analysis were discussed.

In the second part, the noise model was again assumed to be white noise plus power-law noise, but the spectral index of the power-law noise component was not fixed and was also estimated. The estimated spectral indices were compared with the results of the first part, in order to acquire a more precise description of the realistic noise components.

At last, a comparison between the noise features of GPS time series from different ACs was made. Besides the results of noise analysis, the importance and practical value of noise analysis was also discussed.