EGU2020-21260
https://doi.org/10.5194/egusphere-egu2020-21260
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Spatial variations of stochastic noise properties of GNSS time series

Rui Fernandes1,2, Xiaoxing He3,4, Jean-Philippe Montillet1,5, Machiel Bos1,2, Tim Melbourne6, Weiping Jiang4, and Feng Zhou7
Rui Fernandes et al.
  • 1University of Beira Interior, Covilhã, Portugal
  • 2Instituto Dom Luiz (IDL), Faculdade de Ciências, Universidade de Lisboa, Lisboa, Portugal
  • 3School of Civil Engineering and Architecture, East China Jiaotong University, Nan Chang
  • 4GNSS Research Center, Wuhan University, Wuhan, China
  • 5Institute of Earth Surface Dynamics University of Lausanne, Lausanne, Switzerland
  • 6Pacific Northwest Geodetic Array, Dept. of Geological Sciences, Central Washington University, U.S.A.
  • 7College of Geomatics, Shandong University of Science and Technology, Qingdao 266590, China

The analysis of daily position Global Navigation Satellite System (GNSS) time series provides information about various geophysical processes that are shaping the Earth’s crust. The goodness of fit of a trajectory model to these observations is an indication of our understanding of these phenomena. However, the fit also depends on the noise levels in the time series and in this study we investigate for 568 GNSS stations across North America the noise properties, its relation with the choice of trajectory model and if there exists a relationship with the type of monuments. We use the time series of two processing centers, namely the Central Washington University (CWU) and the New Mexico Tech (NMT), which process the data using two different complete processing strategies.

We demonstrate that mismodelling slow slip events within the geodetic time series increases the percentage of selecting the Random-Walk + Flicker + White noise (RW+FN+WN) as the optimal noise model for the horizontal components, especially when the Akaike Information Criterion is used. Furthermore, the analysis of the spatial distribution of the RW component (in the FN+WN+RW) around North America takes place at stations mostly localised around tectonic active areas such as the Cascadia subduction zone (Pacific Northwest) or the San Andreas fault (South California) and coastal areas. It is in these areas that most shallow and drilled-braced monuments are also located. Therefore, the comparison of monument type with observed noise level should also take into account its location which mostly has been neglected in previous studies. In addition, the General Gauss-Markov (GGM) with white noise (GGM+WN) is often selected for the Concrete Pier monument especially on the Up component which indicates that the very long time series are experiencing this flattening of the power spectrum at low frequency. Finally, the amplitude of the white noise is larger for the Roof-Top/Chimney (RTC) type than for the other monument’s types. With a varying seasonal signal computed using a Wiener filter, the results show that RTC monuments have larger values in the East and North components, whereas the deep-drilled brace monuments have larger values on the vertical component.

How to cite: Fernandes, R., He, X., Montillet, J.-P., Bos, M., Melbourne, T., Jiang, W., and Zhou, F.: Spatial variations of stochastic noise properties of GNSS time series, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21260, https://doi.org/10.5194/egusphere-egu2020-21260, 2020

How to cite: Fernandes, R., He, X., Montillet, J.-P., Bos, M., Melbourne, T., Jiang, W., and Zhou, F.: Spatial variations of stochastic noise properties of GNSS time series, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-21260, https://doi.org/10.5194/egusphere-egu2020-21260, 2020