2.2 Retrieval algorithm
For this study, we use measurements of diffraction order 49
(6318–6387 cm–1), order 54
(6960–7040 cm–1) and diffraction order 56 (7217-7300
cm–1) (Fig.1A-D). Order 49 was initially used to
retrieve the temperature and CO2 density using the 1.57
μm CO2 band and CO mixing ratio using a weaker
overlapping band (Fedorova et al., 2020) (Fig.1B). Order 56 contains the
strongest lines of the 1.38 μm water vapor band (Fig.5D). Order 54,
firstly used here for NIR analysis, contains the strongest (in the NIR
spectral range) the 1.43 μm CO2 band. Together with
order 49 it was used to retrieve the temperature and CO2density especially at altitudes above 80 km where the 1.57 μm
CO2 band is weaker.
The retrieval follows the same methodology as the previous temperature
and water vapor retrieval with this instrument (Fedorova et al., 2020).
A forward model of transmission is computed using a look-up-table of
absorption cross-sections (as a function of pressure and temperature)
for a corresponding number of atmospheric layers (40 to 130 depending on
orbit), using the spectral line parameters from HITRAN 2016 (Gordon et
al., 2017) with a correction coefficient of 1.7 for the
H2O broadening in CO2-dominated
atmosphere, as suggested by Gamache et al. (1995) and self-broadening in
the case of CO2.
The first step is a retrieval of temperature and pressure profiles and
the CO mixing ratio based from order 49 and separately from order 54.
The retrieval is performed using a Levenberg–Marquardt iterative scheme
with Tikhonov regularization applied, customary for vertical inversions
in order to smooth the profile and minimize the errors. To constrain
simultaneous retrieval of temperature and pressure, we assume
hydrostatic equilibrium as described in Fedorova et al. (2020). Results
are presented in Figure 1 for temperature (5E) and density (5F)
profiles. Both orders have shown very close results and to minimize the
uncertainties for future study we use the weight averaged profiles
between two orders.
In the next step, the H2O number density and v.m.r. are
retrieved, applying a similar retrieval procedure to the spectra in the
diffraction order 56 (7217–7302 cm–1). Only one free
parameter vector is retrieved (the H2O vmr ) with
the pressure and temperature profiles obtained in the previous step. The
self-consistent retrieval of the H2O profile, along with
that of CO2 and the temperature profile, is a more
reliable approach than using Climate model predictions of temperature.
These predictions are shown to have a bias compared to observations (as
shown in Fedorova et al. 2020). Figure 1 (B,C,D) shows the example of
spectra with the best-fit synthetic spectra demonstrating the good
quality of the data.
The uncertainty on the retrieved quantities is given by the covariance
matrix of the solution. In case of water vapor, we account for the
Jacobian errors due to the uncertainties in the retrieved T and P. For
further analysis, we removed all data with H2O
uncertainties exceeding 100% or with temperature error exceeding 12 K.
ACS NIR measured CO2 and H2O vertical
profiles together with several instruments onboard TGO. The validation
of NIR temperature profiles against ACS MIR results obtained from the
2.7 μm band showed a good agreement for a wide set of simultaneous
measurements (Alday et al., 2019; 2021; Fedorova et al., 2020; Belyaev
et al., 2021). The water profiles were validated with the ACS MIR
measurements in 2.6 μm water band (Alday et al., 2021; Belyaev et al.,
2021).