Analysis of Seasonal Signal in GPS Short-Baseline Time Series

• Published in: Pure and applied geophysics Vol. 175; no. 10; pp. 3485 - 3509
• Main Authors: Wang, Kaihua, Jiang, Weiping, Chen, Hua, An, Xiangdong, Zhou, Xiaohui, Yuan, Peng, Chen, Qusen
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.10.2018; Springer Nature B.V
• Subjects: Amplitude; Amplitudes; Components; Correlation analysis; Data processing; Delay; Direction; Displacement; Earth and Environmental Science; Earth Sciences; Elevation; Flicker; Gaussian process; Geophysics/Geodesy; Global positioning systems; GPS; Height; Horizontal orientation; Markov chains; Modelling; Noise; Quantitative analysis; Random walk; Satellite navigation systems; Signal processing; Solutions; Thermal expansion; Time correlation functions; Time series; Troposphere; White noise; Monument thermal expansion; GPS short-baseline; seasonal signal; noise characteristic
• ISSN: 0033-4553; 1420-9136
• DOI: 10.1007/s00024-018-1871-4

Proper modeling of seasonal signals and their quantitative analysis are of interest in geoscience applications, which are based on position time series of permanent GPS stations. Seasonal signals in GPS short-baseline (< 2 km) time series, if they exist, are mainly related to site-specific effects, such as thermal expansion of the monument (TEM). However, only part of the seasonal signal can be explained by known factors due to the limited data span, the GPS processing strategy and/or the adoption of an imperfect TEM model. In this paper, to better understand the seasonal signal in GPS short-baseline time series, we adopted and processed six different short-baselines with data span that varies from 2 to 14 years and baseline length that varies from 6 to 1100 m. To avoid seasonal signals that are overwhelmed by noise, each of the station pairs is chosen with significant differences in their height (> 5 m) or type of the monument. For comparison, we also processed an approximately zero baseline with a distance of < 1 m and identical monuments. The daily solutions show that there are apparent annual signals with annual amplitude of ~ 1 mm (maximum amplitude of 1.86 ± 0.17 mm) on almost all of the components, which are consistent with the results from previous studies. Semi-annual signal with a maximum amplitude of 0.97 ± 0.25 mm is also present. The analysis of time-correlated noise indicates that instead of flicker (FL) or random walk (RW) noise, band-pass-filtered (BP) noise is valid for approximately 40% of the baseline components, and another 20% of the components can be best modeled by a combination of the first-order Gauss–Markov (FOGM) process plus white noise (WN). The TEM displacements are then modeled by considering the monument height of the building structure beneath the GPS antenna. The median contributions of TEM to the annual amplitude in the vertical direction are 84% and 46% with and without additional parts of the monument, respectively. Obvious annual signals with amplitude > 0.4 mm in the horizontal direction are observed in five short-baselines, and the amplitudes exceed 1 mm in four of them. These horizontal seasonal signals are likely related to the propagation of daily/sub-daily TEM displacement or other signals related to the site environment. Mismodeling of the tropospheric delay may also introduce spurious seasonal signals with annual amplitudes of ~ 5 and ~ 2 mm, respectively, for two short-baselines with elevation differences greater than 100 m. The results suggest that the monument height of the additional part of a typical GPS station should be considered when estimating the TEM displacement and that the tropospheric delay should be modeled cautiously, especially with station pairs with apparent elevation differences. The scheme adopted in this paper is expected to explicate more seasonal signals in GPS coordinate time series, particularly in the vertical direction.