Radio occultation soundings of the neutral atmosphere exhibit some (by now well known) features: Raw measurements are, at least in principle, traceable to SI standards; there is no need for measurement calibration; and no known instrument drifts exist. Due to this unique combination of measurement characteristics, radio occultation soundings are not only used succesfully in operational Numerical Weather Prediction (NWP) systems, but also widely recognized as possible climate benchmark measurements.

Early intercomparisons between co-located measurements from the US-Taiwanese COSMIC/FORMOSAT-3 constellation on one hand and the European GNSS Receiver for Atmospheric Sounding (GRAS) instrument on the other show small, but systematic differences even in the mid and lower stratosphere, where radio occultation measurements are thought to exhibit their highest accuracy. These findings may raise concern about how well the said measurement characteristics are indeed realized by present radio occultation observing systems. One possible cause for the observed systematic differences might be differences in the receivers themselves ("Inter-instrument uncertainty"). An alternative explanation might be differences in the implementation of various algorithms in the respective processing systems leading to systematic differences in retrieved quantities ("Structural uncertainty").

In order to obtain some insight into these issues we have processed both low level COSMIC data (excess carrier phase measurements) and raw GRAS data with the same radio occultation processing system. In this talk we will provide initial results of this exercise, and discuss them in the context of inter-instrument and structural uncertainty.