Fusion applied to phase altimetry in a GNSS buoy system

It has been shown that the earth surface can be observed using Global Navigation Satellite System (GNSS) signals as signals of opportunity. An important advantage of GNSS in this regard is that it provides a global coverage of the earth thanks to dozens of satellites, projected to be 120 by 2030, distributed in various constellations.

GNSS signals parameters such as the carrier phase and the amplitude can be used for example for soil moisture estimation, sea ice detection or sea surface altimetry, which is an important indicator for studying climate evolution. As the sea level varies in centimeters, sea surface altimeters have to be very precise. This accuracy can be achieved using satellite altimeters, provided the availability of precise validation and calibration techniques and in-situ experiments. The objective of this study is the definition of an original GNSS buoy system for satellite altimeters calibration.

GNSS buoy systems are a cheap and light alternative solution to mareographs and can provide high rate measurements. In our approach using this system, we consider, as observable, the phase difference between incoming GPS-L1 signals at a reference, fixed antenna at ~10m height on the ground and at a buoy antenna on the sea. In an analogy with a GNSS reflectometry system, the buoy can be compared to a specular reflection point, but presents the advantage of collecting the data from all visible satellites at the same location. The signals sensed by both antennas are digitized before processing.

Assuming that the horizontal two-dimensions position of the buoy is accurately known by GNSS positioning (which is more efficient in these dimensions than for estimating the height of the buoy), a new phase observable evolving linearly in the [-π , π] interval as a function of the sine of the satellite elevation can be defined. The slope of this linear evolution is proportional to the height between the two antennas, which is the parameter to estimate. For accuracy and robustness purpose, the estimation of this slope is realized using a circular-linear regression technique that includes the fusion of the data from all visible satellites signals. Indeed, we can show that, using the full span of the sines of visible satellites elevations, centimeter accuracy can be reached for integration times as short as a few milliseconds. The GNSS buoy technique described in this work is evaluated on synthetic and real data.