Detection of Hazardous Ground Movements with Instantaneous Velocity Estimates by GNSS

Abstract

The motivation for this work is to rapidly assess whether a GNSS (Global Navigation Satellite System) station is moving hazardously or not. The GNSS stations of interest are located on moving land masses – like rock glaciers – in alpine regions in Switzerland. The purpose is to support real-time natural hazard early warning systems with information on ground movements. One of the main problems with geodetic high-precision GNSS positioning algorithms is the convergence time needed until a reliable and accurate solution is obtained; furthermore a data stream with corrections from GNSS reference stations or reference networks is needed. This problem is overcome here by inferring movements from high-precision and high-rate (≥ 1 Hz) estimates of the instantaneous station velocity, which is obtained based on the GNSS Doppler shift. This approach works with numerical derivatives of GNSS carrier phase measurements, and only needs data from the satellite broadcast message. Thus it can be used for stand-alone, autonomous GNSS stations. At each time instant, statistical hypothesis tests are applied to the resulting velocity estimates, in order to test their significance and to detect a movement. Furthermore a global decision criterion is introduced, which incorporates the movement information collected over several epochs – the goal is to reduce the number of false alarms. The proposed algorithm was tested on movement data from two field experiments: For the first tests, data from a shake table experiment was used, and the second test was carried out with an industrial robot. Besides testing the algorithms’ performance, another question was how small a movement can be, in order to still be detectable with the algorithm. Furthermore, the method was tested on GNSS measurements from a strong earthquake in central Italy, with data for more than 40 autonomous GNSS stations available. It was found that the proposed algorithm has the potential to resolve movements at the millimeter-per-second level; even reaching the sub-mm/s level is possible by including measurements to multiple GNSS. Reliable movement information can be provided within seconds. It can be concluded that the proposed algorithm can give an important contribution to early warning systems for natural hazards, as e.g. for landslides or earthquakes.