Item

Abstract

Pervasive seasonal signals have been found in GPS site position time series. However, the artificial seasonal signals, such as the GPS draconitic year (frequency of 1.04 cycle per year (cpy)), are often present in the GPS time series, and bias the annual signal (1.00 cpy) estimation in conventional sinusoidal model fitting. A more rigorous approach should remove the assumption of constant phase and let the data themselves reveal the time-varying seasonal signals in time series. In this study, we focus on the identification of such anomalous harmonics in GPS station position time series. We applied the Independent Component Analysis (ICA) and Singular Spectrum Analysis (SSA) to extract GPS draconitic year in GPS coordinate timeseries in Yunnan Province, China. The results show that ICA can effectively identify the GPS draconitic errors and annual signals. Then, we remove the systematic errors from the GNSS observations, and the results show that ~20% power of observed annual signals reduce in E/N/U directions. The effects of removing those errors on velocity estimations are also investigated, and the largest velocity difference of ~0.2 mm/yr are obtained in E/N/U directions. We also apply the SSA to extract the modulated seasonal signals in the region. The first two recovered principal components were treated as the modulated seasonal signals according to the spectra analysis, which explain ~50% of data variance. Our results demonstrate that SSA can effectively extract modulated oscillations in Yunnan in 2019, which is highly correlated with the precipitation anomalies.