Thesis title: NOMAD/TGO and PFS/MEx joint analysis for the retrieval of Trace Gases in the Atmosphere of Mars
NOMAD and PFS are two infrared spectrometers for atmospheric analysis respectively onboard Esa’s Trace Gas Orbiter and Mars Express satellites. PFS has two channels: the Long-Wavelength channel (LWC), which works in the (5.5, 45) μm spectral range, and the Short Wavelength channel (SWC), which works in the (1.2,5.7) μm spectral range. NOMAD has instead two nearly twin channels, the SO and the LNO, which work in the (2.3, 4.3) μm spectral range in solar occultation and nadir geometries, respectively, plus a UVIS channel in the ultraviolet. The LNO, and the two channels of PFS have been used to obtain a global and extensive climatology of the main trace gases of climatological interest in the Martian atmosphere, i.e. water vapour and carbon monoxide. Then, the joint observations of these two instruments have been exploited to obtain cross-information and validation of the two datasets.
A first effort has been done in developing an algorithm for the computation of the absorption coefficients of water vapour and carbon monoxide in an atmosphere of carbon dioxide. The approach used is the line-by-line with a Voigt profile, using the parameters from the HITRAN 2016 database (Gordon et al., 2017) and from Gamache et al. (2016) and Brown et al. (2007). For each species, the absorption coefficiens have been computed for a set of 15 atmospheric profiles that cover all the possible pressure-temperature conditions one can find on Mars in order to build a reference database to be used for later interpolation on the specific temperature-pressure profiles associated to the various measurements. These absorption coefficients have been used as key input for computation of synthetic spectra to best-fit the observed LNO and PFS spectra. The other input atmospheric parameters (Temperature-Pressure profiles, integrated dust and ice opacities) are extracted from the general circulation model MCD v5.3 (Forget et al., 2017) for the LNO dataset, and retrieved from the LWC (Grassi et al., 2005) for the PFS dataset. The forward model used in this analysis is provided by the radiative transfer software ARS developed by Ignatiev et al. (2005). Water vapour has been retrieved from the PFS LWC using the Levenberg-Marquardt algorithm in the 310.4-500.0 cm−1 (20-32.21 μm) spectral range, and from the NOMAD LNO orders 167 (3752.96 cm−1,3782.91 cm−1), 168 (3775.89 cm−1, 3806.03 cm−1), and 169 (3797.90 cm−1, 3828.20 cm−1) using the Optimal Estimation Method with a Bayesian approach. The carbon monoxide has been retrieved by applying the Optimal Estimation Method to both instruments’ database, using the PFS SWC in the 2000.0-2220.0 cm−1 (4.51-5 μm) spectral range and the NOMAD LNO orders 189 (4247.77 cm−1, 4281.67 cm−1) and 190 (4270.17 cm−1, 4304.26 cm−1). The codes have been written in Interactive Data Language (IDL), and run on a Rack Server with 8 Intel Xeon Gold processors running at 2.4 GHz with a total of 144 cores and 288 threads fully dedicated to the retrievals. In order to optimise the consumption of time, the retrievals ran with a parallelization scheme which takes advantage of the whole set of available cores. The retrieval is able to obtain the water integrated abundance and carbon monoxide volume mixing ratio with an uncertainty typically lower than 15% for PFS and 20% for LNO retrievals.
The whole dataset of PFS has been processed, obtaining a global climatology of the two trace gases for about 10 Martian Years between MY 26 and MY 35, which is the most extended and continuous Martian atmospheric dataset to date. The LNO has a much shorter dataset, obtained between the end of MY 34 and the beginning of MY 36, but it allowed to analyse the water vapour for more than one Martian year for the first time using this instrument. Finally, simultaneous observations of the two instruments have been exploited for the first time. Joint observations allow a direct and precise comparison of the abundances of the trace gases which can be used to cross-validate the results, obtained with instruments which are very different and operates at different spectral regions. The main climatological features have been recognised both for H2O and for CO, analysed from a qualitative and quantitative point of view, and compared to model predictions and previous results providing climatology maps, spatial maps, and comparison maps.
The water vapour is substantially absent in the polar regions during both northern and southern winter, reaching values about 0 pr-μm, whilst it has its maxima during polar summer where it reaches values up to 64 ± 9.6 pr-μm (PFS) or 67 ± 13 pr-μm (LNO) in the northern hemisphere and 30± 4.5 pr-μm (PFS) or 33 ± 7 pr-μm (LNO) in the southern hemisphere. After the northern summer peak the water vapour is transported equatorwards, and a specular but much less intense feature is present at Ls about 240°. This feature is present both in PFS and LNO data, but from a quantitative point of view, LNO retrievals show a little water increase with respect to PFS retrievals, up to 9 pr-μm. The difference is not present at midlatitudes, where the mean water abundance is always 10 ± 1.5 pr-μm when retrieved with PFS and 10 ± 2 pr-μm when retrieved with LNO. The comparison between PFS and LNO results has been further investigated using the 184 joint measurements of the two instruments. Compared to NOMAD, PFS retrieves slight less water vapour, by about 2-3 pr-μm on average. This difference, which is systematic and non negligible regardless of the input used (MCD or PFS itself) shows a trend, with the difference increasing as the absolute abundance increases, and this means that the difference is due to a multiplicative factor of about 20% rather than to a systematic summation term. Nevertheless, PFS and LNO retrieval datasets should be definitely considered in good agreement within the experimental uncertainties, both from a qualitative and quantitative point of view. The results have also been compared to those obtained by other instruments, including TES/MGS (Smith, 2004), CRISM/MRO (Smith et al., 2018), and SPICAM/MEx (Trokhimovskiy et al., 2015), obtaining a general good agreement in trends and abundances with minor differences due, for example, to a different treatment of the scattering regimes. In any case and for every instrument, the water vapour extracted from MCD at the main summer peak is much higher than the retrieved abundance, suggesting that the model overestimates the rate or the overall amount of H2O sublimation in northern polar summer.
As far as CO is concerned, a characteristic pattern with two equatorial peaks and two summer minima of mixing ratio has been obtained with both instruments. Main equatorial peak is up to 1010 ± 153 ppmv (PFS) or 1054 ± 210 ppmv (LNO) at the end of northern summer, whilst the secondary peak is up to 850 ± 127 ppmv (PFS) or 858 ± 172 ppmv (LNO). The two seasonal minima are down to 375 ± 56 ppmv (PFS) or 400 ± 80 ppmv (LNO) during southern summer, and 600 ± 90 ppmv (PFS) or 660 ± 132 ppmv (LNO) during northern summer. The southern winter maximum predicted by the models has also been observed for the first time in the PFS dataset. The general spatial agreement between PFS and LNO is good, but there are some differences. Both instruments see strong seasonal and spatial gradients which are not correlated with topography. The deepest depletions of CO abundance are observed during northern spring and southern spring/summer, but generally the depletion is stronger in the south than in the north. At southern latitudes the longitudinal variation is very weak during that season, whilst it is stronger in the northern hemisphere, where the extension of the minimum varies more longitudinally. The depletion in the northern hemisphere is latitudinally more spread than in the southern hemisphere, but the two instruments do not see the same extension. Using PFS the southern CO depletion is confined between 60°S and the pole, whilst in the north it extends down to 30°N in Vastitas borealis. With LNO the abundance is systematically higher than PFS at this latitudes by about 50 ppmv during the summer low. The global average is about 800 ppmv, which is consistent with previous literature. The only instrument which retrieved carbon monoxide extensively is CRISM (Smith et al., 2018), and there is a geneneral very good agreement between PFS, LNO, and CRISM results. Actually all the CO results of PFS and LNO are in extremely good agreement, well within the estimated uncertainty, and this is confirmed also by the 33 joint observations here analysed, with no particular trends or bias due to the different inputs. An additional comparison between PFS and the results from Mars Science Laboratory (Trainer et al., 2019), confirmed this results, showing a similar trend for CO. Moreover, LNO, PFS and MSL see the main peak position which is 15° later in solar longitude than the model MCD, suggesting slower physical mixing of air masses than predicted by the models.