An analysis of 108,058 atmospheric refractivity profiles observed by CHAMP during 2002 and 2003 reveals a negative bias compared to ECMWF meteorological fields at altitudes below 5 km. The bias is most pronounced in the tropical Pacific, Central Africa, Indonesia, and India with values reaching -4%. In order to separate bias contributions caused by critical refraction from contributions induced by the receiver tracking process a comprehensive end-to-end simulation study was performed using radio sonde profiles obtained regularly by Alfred Wegener Institute aboard research vessel POLARSTERN since 1982. Within a subset of 2917 profiles recorded between 60N and 60S on the Atlantic ocean between 29 December 1982 and 14 November 2003, 40.2% (1172 profiles) indicate the presence of critical refraction with vertical refractivity gradients below -157 km^-1. Layers exceeding the critical refractivity value are mainly located between 1 to 2 km altitude, above 3 km the occurrence of critical refraction can be disregarded. Simulations including a receiver signal tracking model and using these 2917 sonde observations confirm that four quadrant carrier phase extraction outperforms the arctangent method currently implemented on CHAMP. Within regions of low signal-to-noise ratio an interesting alternative to 'fly-wheeling' and open-loop tracking methods is carrier loop band width reduction. Changing the band width from 30 to 10 Hz improves data yield at 0.5 km altitude by about 16%.

---

Beyerle, G.; König-Langlo, G.; Wickert, J.; Schmidt, T.; Heise, S.; Kaschenz, J. (2004): An analysis of the negative refractivity bias detected in GPS radio occultation data: Results from simulation studies, aerological soundings and CHAMP observations. Joint CHAMP/GRACE Science Meeting (Potsdam 2004).