Aeolus + Processes, Technical Note 4

The report presents results of the Work package 4 (WP4) of the ESA Contract
No. 4000139422/22/I-NS which compares the impacts of Aeolus and COSMIC2 observa-
tions in the ECMWF system. Individual and combined effects of the two observing systems
are quantified by comparing three observing system experiments (OSEs) with the control
experiment which assimilated all operationally used data at the ECMWF. The three ad-
ditional OSEs include Aeolus, COSMIC2, and both Aeolus and COSMIC2 assimilated on
top of all observations used in the control experiment.
The results show that Aeolus and COSMIC2 provide complementing effects in almost
all scores and processes discussed in the report. Both datasets reduce the background
departure statistics of other observation systems used operationally at ECMWF. When
the both observation types are assimilated together, a further improvement is produced
in the short-range and medium-range forecasts. The COSMIC2 data produce changes
in the large-scale wind analyses that generally agree with Aeolus effects, and become the
dominant impact on the wind field in the upper stratosphere, where Aeolus is not available.
We showed that stratospheric zonal wind changes produced by COSMIC2 are largely a
linearly balanced zonal wind. There is a clear synergy from the combined assimilation of
the two datasets in the upper troposphere and lower stratosphere; they both impact the
zonal-mean zonal wind by sharpening the vertical shear zones associated with the westerly
phase of the quasi-biennial oscillation (QBO) during the OSE period.

Within the tropical troposphere, effects of COSMIC2 on the wind analyses reduce
compared to the stratosphere, and Aeolus can produce a larger change than COSMIC2, as
illustrated in figure showing the time-averaged analysis difference between the control ex-
periment (CTRL) and COSMIC2 and between CTRL and the experiment with COSMIC2
and Aeolus (Both) data. The experiment with both data types produces a larger effect
on the equatorial waves and momentum flux analyses than either system individually. In
particular, Aeolus effects may be crucial for the correction of random errors at small scales
in the troposphere, an effect hard to filter from the variance spectra dominated by large
scale effects. Although the amplitude of Aeolus impact on high-latitude analyses is smaller
than in the tropics, it is relevant for forecast improvements, especially since COSMIC2
data do not extend poleward from 40o. Both observing systems have a positive impact on
the forecast-error reduction for the vector winds and temperature, mainly in the tropics,
with the larger COSMIC2 impact in the stratosphere. Aeolus tends to provide a more
positive impact in the extra-tropical lower troposphere compared to COSMIC2. As for
the analyses, the combined impact of the two observing systems exceeds their individual
effects for most of the variables and levels assessed.