A per-profile observation error model for radio occultation using observations

Estimation of uncertainties (random error statistics) of radio occultation (RO) observations is important for their effective assimilation in numerical weather prediction (NWP) models. Average uncertainties can be estimated for large samples of RO observations by various methods, and these statistics are typically used for specifying the observation errors in NWP data assimilation. However, the uncertainties of individual RO observations vary, and so using average uncertainty estimates will overestimate the uncertainties of some observations and underestimate those of others, reducing their overall effectiveness in the assimilation.

Here we present a new error model for RO based on two parameters of the observations: local spectral width (LSW) and STD4060. LSW is a measure of atmospheric multipath and is a good predictor of RO uncertainties below 10 km. Above 30 km, the RO observational uncertainty varies with STD4060, the standard deviation of the RO observation anomalies from an exponential fit to the observations between 40 and 60 km. We show that RO uncertainties estimated by the three-cornered hat method are highly correlated with these parameters below 10 km and above 30 km. Between 10 and 30 km the observational uncertainty of all profiles is well represented by the statistical mean. We describe the generation of this hybrid error model, based on LSW and STD4060, the use of which is described in two other talks at this conference.