The Influence of the Signal-to-Noise Ratio upon Radio Occultation Retrievals

Abstract

We study the dependence of radio occultation (RO) inversion statistics on the signal-to-noise ratio (SNR). We use observations from four missions: COSMIC, COSMIC-2, METOP-B, and Spire. All data are processed identically using the same software with the same settings for the retrieval of bending angles, which are compared with reference analyses of the National Oceanic and Atmospheric Administration (NOAA) Global Forecast System. We evaluate the bias, the standard deviation, and the penetration characterized by the fraction of events reaching a specific height. In order to compare SNRs from the different RO missions, we use the results of our previous study, which defined two types of SNR. The statically normalized SNR is defined in terms of the most probable value of the noise floor for the specific mission and global navigation satellite system. The dynamically normalized SNR uses the noise floor value for the specific profile. This study is based on the dynamical normalization. We also evaluate the latitudinal distributions of occultations for different missions. We show that the dependence of the retrieval statistics on the SNR is not very strong, and it is mostly defined by the variations of latitudinal distributions for different SNR. For Spire, these variations are the smallest, and here, the bias and standard deviation reach saturated values for a relatively low SNR.