A NEW MODEL FOR MARS ATMOSPHERIC DUST BASED UPON ANALYSIS OF ULTRAVIOLET THROUGH INFRARED OBSERVATIONS FROM MARINER-9, VIKING, AND PHOBOS

We propose key modifications to the Toon et al. (1977) model of the pa rticle size distribution and composition of Mars atmospheric dust, bas ed on a variety of spacecraft and wavelength observations of the dust. A much broader (r(eff)variance similar to 0.8 mu m), smaller particle size (r(mode)similar to 0.02 mu m) distribution coupled with a ''pala gonite-like'' composition is argued to fit the complete ultraviolet-to -30-mu m absorption properties of the dust better than the montmorillo nite-basalt, r(eff)variance = 0.4 mu m, r(mode) = 0.40 dust model of T oon et al. Mariner 9 (infrared interferometer spectrometer) IRIS spect ra of high atmospheric dust opacities during the 1971-1972 Mars global dust storm are analyzed in terms of the Toon et al. dust model, and a Hawaiian palagonite sample (Roush et al., 1991) with two different si ze distribution models incorporating smaller dust particle sizes. Viki ng Infrared Thermal Mapper (IRTM) emission-phase-function (EPF) observ ations at 9 mu m are analyzed to retrieve 9-mu m dust opacities coinci dent with solar band dust opacities obtained from the same EPF sequenc es (Clancy and Lee, 1991). These EPF dust opacities provide an indepen dent measurement of the visible/9-mu m extinction opacity ratio (great er than or equal to 2) for Mars atmospheric dust, which is consistent with a previous measurement by Martin (1986). Model values for the vis ible/9-mu m opacity ratio and the ultraviolet and visible single-scatt ering albedos are calculated for the palagonite model with the smaller particle size distributions and compared to the same properties for t he Toon ct al. model of dust. The montmorillonite model of the dust is found to fit the detailed shape of the dust 9-mu m absorption well. H owever, it predicts structured, deep absorptions at 20 mu m which are not observed and requires a separate ultraviolet-visible absorbing com ponent to match the observed behavior of the dust in this wavelength r egion. The modeled palagonite does not match the 8- to 9-mu m absorpti on presented by the dust in the IRIS spectra, probably due to its low SiO2 content (31%). However, it does provide consistent levels of ultr aviolet/visible absorption, 9- to 12-mu m absorption, and a lack of st ructured absorption at 20 mu m. The ratios of dust extinction opacitie s at visible, 9 mu m, and 30 mu m are strongly affected by the dust pa rticle size distribution. The Toon et al. dust size distribution (r(mo de) = 0.40, r(eff)variance = 0.4 mu m, r(cw mu) = 2.7 mu m) predicts t he correct ratio of the 9- to 30-mu m opacity, but underpredicts the v isible/9-mu m opacity ratio considerably (1 versus greater than or equ al to 2). A similar particle distribution width with smaller particle sizes (r(mode) = 0.17, r(eff)variance = 0.4 mu m, r(cw mu) = mu m) wil l fit the observed visible/9-mu m opacity ratio, but overpredicts the observed 9-mu m/30-mu m opacity ratio. A smaller and much broader part icle size distribution (r(mode) = 0.02, r(eff)variance = 0.8 mu m, r(c w mu) = 1.8 mu m) can fit both dust opacity ratios. Overall, the nanoc rystalline structure of palagonite coupled with a smaller, broader dis tribution of dust particle sizes provides a more consistent fit than t he Toon et al. model of the dust to the IRIS spectra, the observed vis ible/9-mu m dust opacity ratio, the Phobos occultation measurements of dust particle sizes (Chassefiere et al., 1992), and the weakness of s urface near IR absorptions expected for clay minerals (Clark, 1992; Be ll and Crisp, 1993).