Formation of a GNSS network in space based on LEO satellite constellations

The number of low Earth orbit (LEO) satellites equipped with Global Navigation Satellite System (GNSS) receivers is rapidly increasing. GNSS observations in space are no longer limited to a small number of Earth observation satellites, but the rapid development of large nanosatellite constellations enables a dense network of GNSS observations around the Earth. An example of this is the Astrocast CubeSat constellation, to which we contribute with our low-cost multi-GNSS payload board. The first 10 satellites of the Astrocast constellation have successfully been launched on 24 January 2021 (5 satellites) and 30 June 2021 (5 satellites). Further Astrocast CubeSats equipped with dual-frequency GNSS receivers will be launched in the coming years, completing a constellation of 100 satellites by 2024.

The formation of a homogeneous and highly dynamic GNSS network in space holds great potential for geodetic Earth observation, as it has some advantages over a ground-based GNSS network and GNSS observations on board single or formation-flying satellites: A space-based GNSS network can be autonomously processed in a double-difference mode without the need for ground observations, thus, GNSS signals are not affected by tropospheric refraction, and it provides a better observation geometry improving the sensitivity to certain geodetic parameters. In this study, we investigate the feasibility of forming such a space-based GNSS network for estimating geodetic parameters, namely the orbit parameters of the LEO and GNSS satellites, the antenna phase center corrections of the GNSS satellites, and the low-degree coefficients of the Earth’s gravity field including the geocenter coordinates.

We consider 3 different constellation scenarios: (1) A LEO constellation of 36 satellites uniformly distributed over 6 orbital planes with an inclination of 55° and (2) the expected configuration of the complete Astrocast constellation, with sun-synchronous polar orbits and equatorial orbits. In both cases (1) and (2), the GNSS observations are simulated with the Bernese GNSS software based on the given orbit specifications. (3) In a third scenario, we use real GNSS observations from various existing Earth observation missions, including GRACE, OSTM/Jason-2 and Swarm, which are combined to a pseudo-constellation.

For each scenario, the number of possible GNSS single- and double-differences and the corresponding baseline lengths will be computed. Based on these observations, we will examine, how well carrier-phase ambiguities can be resolved and how this depends on the constellation configuration. With a network processing of GNSS double-difference observations, we will estimate concrete parameters related to the LEO orbits, the GNSS antenna phase center corrections and the Earth’s gravity field. To estimate the expected accuracy for these parameters, we examine their sensitivity to small errors in the observation data resulting from, e.g., the force model, once-per-revolution parameters, stochastic pulses or small accelerations like ocean tide or Earth albedo effects. Based on this research, we will draw conclusions about the potential of large satellite constellations to complement or replace the existing geodetic Earth observation missions in the future.