Signal and noise separation in time series of DORIS station coordinates using wavelet and singular spectrum analysis

Abstract

The aim of this article is to assess the noise in Doppler Orbitography and Radio-positioning Integrated by Satellite (DORIS) stations coordinate time series in order to better apprehend the spectral content of the non-linear station motion. In the frame of this work, we have used two different approaches to assess the noise, the wavelet transform in the frequency space, and the Singular Spectrum Analysis (SSA) in the phase space. The data used are the weekly solutions of coordinate residuals of 18 stations in STCD format (solution ign09wd01), computed by Institut géographique national/Jet Propulsion Laboratory (IGN, France/JPL, USA) using the GIPSY-OASIS II software package, referred to ITRF2005 and expressed in the local geodetic reference frame. The SSA has properly identified the non-linear trends and the seasonal signals (annual and semi-annual, 120, 59 and 20.3 days) in the analysed time series. The results show that the dominant signal present in the horizontal components (North and East) is mainly of a geophysical nature (plate tectonics) with a variance over more than 90% in most cases, and the remaining noise at 120 days was reduced in the ign09wd01 solution (w.r.t. the previous ign07wd01 solution). Using the wavelet transform method, we also identify, for the CADB station, a vertical discontinuity of 13mm on 3 June 2008. For the noise level determination, both approaches converge to the same results. The standard deviation of the noise is in the range of 7–13, 8–24 and 7–14mm in, respectively, North, East and Vertical components. The noise level is the highest in the east direction, probably related to the high orbit inclination of some DORIS satellites, and is lower in the high latitude stations which provide more observations from the DORIS satellites.