The impact of differences in retrieval algorithms between processing centers on GNSS radio occultation refractivity retrievals in the planetary boundary layer

Abstract

Global Navigation Satellite System radio occultation (GNSS RO) performance in the planetary boundary layer is strongly dependent on retrieval algorithms. In this work, we characterize differences in refractivity retrievals in the planetary boundary layer across three major processing centers of GNSS RO data - NASA JPL, ROM SAF, and UCAR. Using a shared base of occultations from the FORMOSAT-3/COSMIC-1 GNSS RO mission, we identify key differences between the three processing centers that are especially strong in the regions of frequent super-refraction. We find that the minimum penetration height allowed by each processing center is correlated with the amount of super-refraction, resulting in poorer penetration and higher refractivity biases in the tropics. We found JPL to have the most conservative minimum height in this region at 1 km, followed by ROM SAF (640 m) and UCAR (420 m). We identify two key geopotential heights - 0.8 and 2.6 km - to sample the global distribution of inter-center refractivity bias, finding differences of 0.3 %-0.5 % in the tropics. We also find negative refractivity biases of up to -4 % relative to ERA5 reanalysis in regions of persistent high stratocumulus coverage and areas along the descending branch of the Hadley circulation, with negligible bias along the Intertropical Convergence Zone. A comparison to ERA5 also reveals areas of weak (0.2 %-0.5 %) positive refractivity biases in polar regions. We hypothesize potential causes for these biases based on truncation schemes, radio-holographic filtering choices, and quality control and identify findings deserving of further investigation.