Assessment of high-rate GPS using a single-axis shake table

The developments in GNSS receiver and antenna technologies, especially the increased sampling rate up to 100 sps, open up the possibility to measure high-rate earthquake ground motions with GNSS. In this paper we focus on the GPS errors in the frequency band above 1 Hz. The dominant error sources are mainly the carrier phase jitter caused by thermal noise and the stress error caused by the dynamics, e.g. antenna motions. To generate a large set of different motions, we used a single-axis shake table, where a GNSS antenna and a strong motion seismometer were mounted with a well-known ground truth. The generated motions were recorded with three different GNSS receivers with sampling rates up to 100 sps and different receiver baseband parameters. The baseband parameters directly dictate the carrier phase jitter and the correlations between subsequent epochs. A narrow loop filter bandwidth keeps the carrier phase jitter on a low level, but has an extreme impact on the receiver response for motions above 1 Hz. The amplitudes above 3 Hz are overestimated up to 50 % or reduced by well over half. The corresponding phase errors are between 30 and 90 degrees. Compared to the GNSS receiver response, the strong motion seismometer measurements do not show any amplitude or phase variations for the frequency range from 1 to 20 Hz. Due to the large errors for dynamic GNSS measurements, it is essential to account for the baseband parameters of the GNSS receivers if high-rate GNSS is to become a valuable tool for seismic displacement measurements above 1 Hz. Fortunately, the receiver response can be corrected by an inverse filter if the baseband parameters are known.