On filtering ionospheric effects in GPS observations using the Matérn covariance family and its impact on orbit determination of Swarm satellites

Abstract

With three near-polar low earth orbiter satellites, the Swarm mission contributes to bridging the gap between Earth gravity field missions. However, observations from on-board GPS receivers are strongly impacted by ionospheric scintillations, degrading the kinematic orbit solutions of the satellites and thus the gravity field, which is derived from them. Besides the elimination of parts of the GPS observations strongly impacted by scintillations, an alternative is a physically based filtering of these measurements. Based on the previous empirical and theoretical works, a Matérn covariance function, whose parameters are to be chosen adequately, is an answer to mitigate ionospheric effects from the GPS time series. An optimal parameter set corresponding to a smoothness of 1 and a correlation factor of 1.5 is physically plausible and provides an adequate filtering of the ionospheric scintillations. The detection of noisy parts is carried out with an easy to use algorithm to obtain filtered time series with a homogeneous standard deviation. The orbit solution computed with filtered observations does not exhibit ionospheric artifacts compared with the non-filtered solution. At the same time, the high-frequency slope of the orbit power spectral density is similar to the one obtained for GPS carrier phase observations with low ionospheric scintillation errors. The proposed methodology is independent of the processing used, whether double differences, Observed Minus Computed or raw carrier phase observations.