Estimation and Evaluation of COSMIC Radio Occultation Excess Phase Using
Non-differenced Measurements
Abstract. In the GPS radio occultation technique, the atmospheric excess phase (AEP) can be used to derive the refractivity which is an important quantity in numerical weather prediction. The AEP is conventionally estimated based on GPS double-differenced or single-differenced techniques. These two techniques, however, require the reference link data in the data processing increasing the complexity of computation. In this study, a non-differenced (ND) processing strategy is proposed to estimate the AEP. To begin with, we used PANDA (Positioning and Navigation Data Analyst) software to perform the precise orbit determination (POD) for the COSIMC (The Constellation Observing System for Meteorology, Ionosphere and Climate) satellite to acquire the position and velocity of the center of mass of the satellite and the corresponded receive clock offset. The bending angles, refractivity and dry temperature profiles are derived from the estimated AEP by the ROPP (Radio Occultation Processing Package) software. The ND method is validated by the COSMIC products in typical rising and setting occultation events. Comparison results indicate that RMS (root mean square) errors of relative refractivity differences between ND-derived and "atmPrf" profiles are better than 4 % and 3 % in rising and setting occultation events, respectively. In addition, we also compared the relative refractivity bias between ND-derived and "atmPrf" profiles of globally distributed 200 COSMIC occultation events on December 12, 2013. The statistic results show that the average RMS relative refractivity deviation between ND-derived and COSMIC profile is better than 2 % in the rising occultation event, and it is better than 1.7 % in setting occultation event. Moreover, the observed COSMIC refractivity profiles from ND processing strategy are further validated using European Centre for Medium-Range Weather Forecasts (ECMWF) analyses data, and the results indicate that non-differencing reduces the noise level on the excess phase paths in the lower troposphere compared to single difference processing strategy.