|Title||Spread of the ensemble of data assimilations in radiance space|
|Publication Type||Technical memorandum|
|Secondary Title||ECMWF Technical Memoranda|
|Authors||Bormann, N, Bonavita, M|
This memorandum evaluates the spread of the EDA in radiance space against observation departures, with the aim to validate the use of the EDA spread as an estimate of the expected size of the background errors for use in the analysis quality control decisions. The investigated approach replaces similar estimates previously derived with a randomisation method. It is found that the EDA spread shows good skill in identifying regions of larger observation departures, but it appears to be under-dispersive especially in the extra-tropics. Some inflation of the EDA spread in radiance space is hence required to account for this. After this calibration step, the EDA-based background error estimates and observation departure statistics show very consistent characteristics. The temporal evolution of the EDA spread over the subsequent assimilation window has also been investigated, with indications that the EDA spread does not grow as quickly as observation departures do. The use of the EDA-based background error estimates in FG-check decisions has been investigated in assimilation experiments. Together with a re-tuning of the FG-check limit, the approach allows the use of more observations that were previously rejected.