Derivation of Martian meteorological parameters using ground-based telescopes and forward-modelling

Despite the increasing number of spacecraft sent to Mars in recent times, many properties of the Martian atmosphere remain relatively unconstrained (Forget et al., 1999). Indeed, the Martian surface pressure has only been continuously monitored for any length of time at a total of three surface locations (Leovy, 1979; Golombek et al., 1999). A more complete set of meteorological data is important both for the characterisation of the atmosphere and as input for general circulation models (GCMs) used in the prediction of atmospheric parameters for future Mars missions. Near-infrared spectroscopy provides a method of obtaining much information about the atmospheric state of a planet. In the 1-2.5 μm region spectral lines of CO₂, H ₂O, CO, O₂, O₃ and many others are present to varying degrees. These spectral lines give information not only about the composition of the atmosphere but also about its temperature and pressure via spectral band shapes. Many aerosols, such as Martian dust, also strongly influence the radiative output of Mars in this region. The use of high-resolution (R - 40000) groundbased spectroscopy (with NASA IRTF/CSHELL) has allowed us to develop a method to retrieve Martian atmospheric parameters, such as surface temperature and pressure, across the entire Martian disk by way of forward-modelling. Our input spectra are observations taken close to the 2005 Martian opposition (Oct 25-27, 2005) at a number of different wavelengths: 1.597 μm, 1.603 μm, 1.607 μm, 2.073 μm and 2.332 μm. 'Initial guess' values are taken from the Mars Climate Database (MCD) and iteration proceeds using a Levenberg-Marquardt update rule. Radiative transfer modelling is performed by the SMART package of tools. Resulting spectra typically match our observed spectra very closely. Introduction: Their lower spatial resolution notwithstanding, ground-based observations offer a number of advantages over observations taken on orbital platforms. The large distance from Earth allows the atmospheric state of an entire Martian hemisphere to be characterised using a single set of spectra. Observations can also be obtained at significantly higher spectral resolution than is currently possible with instruments like PFS (Formisano, 2005). However, the use of ground-based spectroscopy does introduce the not-insignificant problem of the removal of telluric lines from the observed spectra. Typically, telluric spectral features are removed from astronomical spectra by making near-simultaneousobservations of a 'featureless' standard star and dividing the planet's spectrum by this. Data reduction done in this manner generally makes the assumption that, in the case of ideal observations, the spectrum of the planet (as seen above the atmosphere) can be reproduced perfectly; however, this is not the case. Planetary atmospheres with molecular components will have narrow infrared spectral features which will remain unresolved when using lowresolution (R < ∼20000) spectroscopy. Consider a low-resolution telluric region of (for the sake of example) 50% transmission calculated using observations of a standard star. If a narrow, unresolved spectral line in the planetary spectrum coincides with a narrow, unresolved band of high transmission in the Earth's atmosphere then the strength of that band will be exaggerated compared to the value obtained by correcting using the 50% value. Conversely, if that line coincides with an unresolved region of strong absorption then using the standard star value will produce an over-correction. This situation is illustrated in Figure 1. For spectra which are uncorrelated with the Earth's atmospheric spectrum (i.e. they have few similar spectral lines) then the net result is that most of these unresolved differences cancel out and the standard star method gives generally accurate results. However, in the case of observations of an atmosphere with similar constituents to our own (e.g. Mars), many of the narrow molecular lines match up, effectively meaning that the standard star method will overcorrect for these lines. To quantify these effects, we generated modelled spectra of the atmospheres of both Earth and Mars using the SMART atmospheric modelling tool (as described below). To simulate the standard star calculation, we used the following procedure: 1. High-resolution atmospheric spectra of both Earth and Mars were generated using a model solar spectrum. 2. The Mars spectrum as seen at the top of the Earth's atmosphere was convolved with a gaussian of appropriate properties to simulate a spectrograph at different spectral resolutions. 3. The Earth spectrum as seen at the surface was convolved with the same gaussian to simulate low-resolution observations of a standard star. 4. The high-resolution Mars spectrum was passed through the Earth model, and the resulting spectrum as seen on the ground was convolved with the gaussian to simulate low-resolution observations of Mars. 5. The low-resolution Mars observations in (4) were divided by the low-resolution transmission derived from (3) and compared to the low-resolution Mars spectrum in (2). A diagram of this simulation is shown in Figure 2; the results are shown in Figure 3. As can be seen, the difference between 'standard star reduced' and 'actual' values is as much as 50% around strong CO₂ features. At higher spectral resolutions, a similar problem arises (particularly in the case of Mars) due to the presence of saturated lines. If an observed spectral line is saturated, the linearity of transmission is not preserved - it becomes impossible to directly determine how much of the absorption is due to the observed planet and how much is telluric. To a greater or lesser degree, some form of modelling approach is needed to derive meteorological parameters from these data.