Optimierung der Parameterauswahl für ein operationelles off-line-Retrieval von Spurengasvertikalprofilen aus MIPAS-Daten

“Optimisation of the choice of parameters for an operational
off-line retrieval of vertical profiles of trace gases from MIPAS
data”. Since March 2002 the environment satellite ENVISAT is
operational. ENVISAT carries a series of instruments used for
remote sensing the earth atmosphere. MIPAS, as one of those
instruments is, due to its conception as a limb sounding infrared
spectrometer, intended to provide new insights in the chemical
processes of the middle atmosphere. To achieve this MIPAS measures
emissions of trace gases like H2O, O3, CH4, NO2, N2O, HNO3.
Additionally MIPAS acquires global pressure and temperature
profiles. Vertical profiles of trace gases are computed from
calibrated measurements in the processing chain by an iterative
calculating scheme (retrieval). To this measurements are compared
with simulated spectra from a forward model. At each comparison
modelling parameters are systematically adjusted until an agreement
in the compared quantities is reached, which lies within the
boundaries of some previously defined criteria. Improvements in the
retrieval scheme are possible in three categories: first,
adjustments in the side constraints (regularisation), which divide
the mathematical possible solutions from the physical reasonable,
are conceivable. Second, adjustments in the handling of certain
systematic errors are necessary and finally, one is interested in
eliminating measurements which have become useless due to cloud
influences. To obtain a measure for the improvements, some criteria
are defined and discussed in this work, with which it is possible
to compare differently parameterised retrievals with regard to
their quality. These criteria are: height resolution (averaging
kernel), information content and degrees of freedom, as well as
common quantities like root mean square. The subject regularisation
is being discussed by the means of a comparison between a newly
developed approach (HoMa), a not yet with operational trace gas
retrievals used maximum entropy method (MEM2) and the standard
methods Optimal Estimation and Tikhonov. It can be shown, that the
alternative methods have massive advantages compared with the
standard methods in terms of the above defined criteria. This also
is true for aspects like transparency, diagnostic options and
quality of the retrieved profiles. Especially the results of MEM2
have to be pointed out, because it does not use a priori knowledge.
The treatment of systematic errors shows, that this is an essential
part of every retrieval. The results of computations with simulated
data while using the corrections show tremendous improvements
compared to results of not corrected computations. The part which
deals with the consideration of cloud influences in measurements
shows the advancement of an established idea. A detailed analysis
of cloud indices enables to rescue measurements which otherwise
would have been parcelled out for security reasons. This may lead
to gather additionally knowledge about tropospheric regions from
MIPAS measurements. A comparison of results of the work with
independent results from the ROSE-model delivers promising
consistence.