Quantifying residual ionospheric errors in GNSS radio occultation bending angles based on ensembles of profiles from end-to-end simulations

摘要

The radio occultation (RO) technique using signals from the GlobalNavigation Satellite System (GNSS), in particular from the GlobalPositioning System (GPS) so far, is currently widely used to observe theatmosphere for applications such as numerical weather prediction and globalclimate monitoring. The ionosphere is a major error source in ROmeasurements at stratospheric altitudes, and a linear ionospheric correctionof dual-frequency RO bending angles is commonly used to remove thefirst-order ionospheric effect. However, the residual ionospheric error (RIE)can still be significant so that it needs to be further mitigated for high-accuracy applications, especially above about 30 km altitude where the RIEis most relevant compared to the magnitude of the neutral atmosphericbending angle. Quantification and careful analyses for better understandingof the RIE is therefore important for enabling benchmark-qualitystratospheric RO retrievals. Here we present such an analysis of bendingangle RIEs covering the stratosphere and mesosphere, using quasi-realisticend-to-end simulations for a full-day ensemble of RO events. Based on theensemble simulations we assessed the variation of bending angle RIEs, bothbiases and standard deviations, with solar activity, latitudinal region andwith or without the assumption of ionospheric spherical symmetry andco-existing observing system errors. We find that the bending angle RIEbiases in the upper stratosphere and mesosphere, and in all latitudinalzones from low to high latitudes, have a clear negative tendency and amagnitude increasing with solar activity, which is in line with recent empiricalstudies based on real RO data although we find smaller bias magnitudes,deserving further study in the future. The maximum RIE biases are found at lowlatitudes during daytime, where they amount to within −0.03 to −0.05 μrad, the smallest at high latitudes (0 to −0.01 μrad; quietspace weather and winter conditions). Ionospheric spherical symmetry orasymmetries about the RO event location have only a minor influence on RIEbiases. The RIE standard deviations are markedly increased both byionospheric asymmetries and increasing solar activity and amount to about0.3 to 0.7 μrad in the upper stratosphere and mesosphere. Taking also intoaccount the realistic observation errors of a modern RO receiving system,amounting globally to about 0.4 μrad (unbiased; standard deviation),shows that the random RIEs are typically comparable to the total observingsystem error. The results help to inform future RIE mitigation schemes thatwill improve upon the use of the linear ionospheric correction of bendingangles and also provide explicit uncertainty estimates.