Characterizing noise in daily GPS position time series with overlapping Hadamard variance and maximum likelihood estimation

Abstract

We assess the noise characteristics of daily position time series (Jet Propulsion Laboratory global solutions) from 12 International GNSS Service sites located in China, using both overlapping Hadamard variance (OHVAR) and maximum likelihood estimation (MLE). The result shows that OHVAR is computationally cheap, and the dominate power-law noise inferred is in good agreement with the ones inferred by comparing MLE values. However, OHVAR is insufficient to assess the long-period noise characteristics [e.g., random walk noise (RW)]. From the MLE analysis, the RW contribution is likely to be present in 14% of the time series, while the noise of most sites are characterized by a combination of white and flicker (or power law) noise (i.e., FL + WH or PL + WH). Moreover, sites near the northern crest of the equatorial ionization anomaly have larger amplitudes of WH in the vertical component, while no clear latitude dependency is observed in FL amplitudes.