GPS-derived Density Data for the Swarm Satellites During the Declining Phase of the Solar Cycle

After the detection of many anomalies in the accelerometer data, the development and production of GPS-derived acceleration and thermosphere density products for the three Swarm satellites has been intensified. In order to convert the range and phase information in the Swarm GPS measurements into accelerations, a precise orbit determination approach needs to be used, in which gravitational accelerations are modelled with a very high fidelity, but in which the non-gravitational accelerations are part of the parameters to be estimated. After initial tests with both batch least-squares and Kalman filter orbit determination approaches, a Kalman filter approach was tuned and selected for computing the acceleration data. The resulting GPS-derived accelerations currently serve as a baseline for the correction of Swarm C alongtrack accelerometer data. In addition, the GPS-derived accelerations for all three Swarm satellites are converted directly to thermosphere neutral density data. This GPS-derived density data can serve as a replacement for the originally planned accelerometer thermosphere density products, albeit at a much lower temporal resolution than the accelerometers would have been able to deliver. The accuracy at which accelerations, and subsequently densities, can be derived from the GPS range and phase observations depends on the parameterisation in the orbit determination process. In principle, a higher accuracy can be traded off against a lower temporal resolution. An additional source of error is the modelling of radiation pressure accelerations, which need to be removed from the estimated signal to arrive at the aerodynamic accelerations, which are used as a source to determine density. We have assessed the impact of the declining solar activity level on the currently available acceleration and density data, as well as the impact of various scenarios for the future evolution of the Swarm orbits. Most of the currently available data contains a significant signal well above 2 cycles per orbit revolution at high solar activity. At low solar activity this maximum significant frequency is reduced. A complicating factor is that it would not be very useful for the interpretation of the data to estimate the accelerations and densities above 1 but below 2 cycles per revolution. Currently, the Swarm satellites are still in relatively high orbits, while solar activity is getting lower. Our conclusion is that with the current level of error sources, and keeping the orbits at the current altitude or higher, as proposed in some scenarios, will make it very difficult to resolve latitudinal density variability at solar minimum, using Swarm GPS data.