Jupiter The clouds of Jupiter Fred Taylor and Patrick Irwin examine the Galileo mission’s contribution to our understanding of Jupiter’s colourful chemical cloak. he highly organized and brightly coloured cloud structure on the nearest and largest gas giant planet Jupiter has been explored by the Galileo orbiter/probe project, which completed its nominal mission in December 1997. At least four and possibly as many as six distinct layers of haze or cloud, of different composition and at different depths, appear to contribute to the external appearance of the planet at low and mid-latitudes. A model of the properties of these clouds has been developed from the various data and theoretical constraints. Aspects of the global and time variability of the cloud structure, and its coupling with dynamical systems like the Great Red Spot, are also becoming clearer, allowing speculation about their nature and origins. Analyses of the full four-year data set, some of which is still to be acquired, will add further details of the meteorological behaviour of Jupiter’s atmosphere. T June 1999 Vol 40 I t has been known for a long time that the visible face of Jupiter consists of clouds, with no surface in sight. In the last few decades it has become clear that there is no surface at all in the usual sense, just a very deep atmosphere which becomes denser with depth, eventually condenses, and deeper still, becomes metallic. This point was controversial for a while, mainly because the clouds have quasipermanent structure associated with them which, for some, implied a shallow atmosphere pheric structure, dynamics and meteorology. One of the most curious aspects is the bright and varied colours (figure 1). They suggest that we are observing, not just one single type of cloud, but several, probably with different chemical compositions. The colours also hint at complex chemistry in the clouds, perhaps even with exobiological implications, although the latter remains pure speculation at this stage. The other remarkable feature is the morphology; Jovian clouds form broad bands which retain their overall organization for centuries. These bands are studded with giant oval eddies, some of which are long-lived and tend to be solitary (the GRS being the most famous example), while others are less persistent and sometimes occur in strings. There are also remarkable, chaotic-looking clouds, made up of strands which seem disinclined to mix with their surroundings, although this is at least partly an illusion which occurs because of their grand scale. Some of the dark regions within the belts appear very bright in thermal emission, especially near wavelengths of 5 microns where the gaseous components – primarily hydrogen, helium, methane and ammonia – are free of absorption bands. It follows that these regions, known as “hot spots”, are relatively cloudfree: emission from levels as deep as 8 bars in pressure, where the temperature is around 300 K, or close to that on the surface of the Earth, reaches space from the clearest regions. High-resolution spectroscopy in these hot spots can sound levels warm enough for species such as water and phosphine to be present as vapour in appreciable amounts. Possibilities with Galileo 1 Jupiter at visible wavelengths, with the Earth on the same scale for comparison (NASA). overlying a solid surface. Many of the theories of the origin of the Great Red Spot (GRS), for example, required the atmosphere to be flowing over some rigid topographical feature. Previous knowledge and key questions A quick look at a photograph of Jupiter reveals that the clouds are complex (figure 1). Longstanding questions include: How many layers of cloud are we seeing? What are they composed of? Where do they typically lie in pressure, temperature, and altitude? How are they produced and dissipated? What are the colours due to? What are the giant white ovals, the brown barges, the Great Red Spot? Eventually we should like to understand every aspect of the chemistry and microstructure of Jovian cloud systems, and how they relate to atmos- The Galileo mission started in December 1995 with the insertion of an entry probe, and has continued with remote sensing from the orbiter. Prior to studies of Jupiter by the Galileo mission, we lacked even basic knowledge of the Jovian clouds and the related issues of atmospheric composition and its variability, the atmospheric general circulation, and the dynamical nature of the various kinds of giant eddies. Galileo offers considerable advantages over earlier investigations. First of all, the probe data set is unique, since it is the first acquired inside any of the outer planet atmospheres. However, care is required in its interpretation since the probe sampled only one point in what is obviously a very inhomogeneous system. Next, the orbiting spacecraft is close enough to the planet that it can resolve features such as hot spots not only with its camera, but also with infrared remote sensing instruments. This offers the prospect of relatively cloud-free remote sensing of the atmosphere to a considerable depth, perhaps down to the levels where aqueous clouds are expected to lie, where the pressure is several bars. It 3.21 Jupiter also offers some unique observing geometry, including infrared coverage of the night side of Jupiter which is not accessible to Earth-based telescopes, and high-resolution feature tracking over a range of zenith and phase angles to obtain new information on the scattering properties of the cloud particles. The science payload of Galileo includes the first imaging spectrometer to be used on a mission to the outer planets. This device, called the Near Infrared Mapping Spectrometer, NIMS (Carlson et al. 1992), has allowed imaging of Jupiter at 408 wavelengths from 0.7 to 5.2 µm (figure 2). Spectra at each point can be inverted to retrieve vertical profiles of cloud opacity and species abundances on Jupiter, with spatial mapping of cloud-top temperature and morphology within belts and zones, as well as localized features such as the Great Red Spot. The solid-state imaging system (Belton et al. 1996) has a series of filters in the visible and near-infrared which has been used to probe the properties of the highest cloud and haze layers (Banfield et al. 1998). The purpose of this paper is to synthesize the latest results with earlier observations and theoretical models to give an updated overall picture of the Jovian cloud systems. As we shall see, there seem to be at least four layers of different condensed materials contributing to the visual appearance of Jupiter, all highly variable across the face of the planet. It is not possible at this stage to characterize such a complex arrangement fully or in detail, but the nature of Jupiter’s clouds to first order is much clearer than it was before Galileo. Theoretical basis Theoretical arguments provide a strong underpinning of the overall picture, by defining our expectations of the overall elemental composition of Jupiter and hence of the abundances of cloud-forming materials. The basis of current thinking is that when the solar system formed, the core of Jupiter formed by the initial accretion of planetesimals was large and cool enough to retain all the material in the protosolar cloud, including the lightest elements, hydrogen and helium. Thus, it should have the same mix of elements as the Sun, modified, possibly substantially, by the later accrual of planetesimals and smaller solid debris in the form of meteorites and comets. A ball of gas the size of Jupiter will release heat from its interior, due to fractionation of heavier elements in the interior and overall contraction. The energy budget of the planet has been measured by airborne and spaceborne radiometers (Hanel et al. 1981). The internal power turns out to be a large fraction, about 70%, of the heat radiation which Jupiter receives from the Sun. The release of such large amounts of internal heat makes Jupiter deeply 3.22 2 Simultaneous images of Jupiter at four infrared wavelengths (left to right, top to bottom: 1.61, 2.17, 3.01, and 4.99 µm), obtained by the Near Infrared Mapping Spectrometer (NIMS) on Galileo’s second orbit of the planet in 1996. The difference in the appearance of the clouds is due primarily to the wavelength dependence of the absorption properties of methane and ammonia in the atmosphere, and to the transition from primarily reflected solar energy at short wavelengths to Jovian thermal emission at longer wavelengths. convective. Air generally rises in the cloudy zones and descends in the dark belts, as discussed further below. The temperature as a function of radial distance from the centre of Jupiter can be calculated, and the expected composition in terms of stable compounds can also be worked out from the known mixture of elements. To first order, one expects that the abundant elements will be present mostly in their fully hydrogenated form (nitrogen as NH3, carbon as CH4, and so on). It is straightforward to consider reactions between them (H2S tends to combine with NH3 to form NH4SH, for example) and calculate where on the profile these compounds will condense. The most recent study of this kind (CameronSmith 1998) confirms that the highest substantial condensate cloud on Jupiter should be crystals of frozen ammonia, overlying solid particles of condensed ammonium hydrosulphide, with a primarily aqueous cloud deeper still. Such results are interesting, but do not provide a definite solution since, among other limitations, the method makes no allowance for atmospheric motions – which move cloudforming materials in all directions, most critically upwards – for non-equilibrium chemistry, or for heterogeneous processes. Still, they pro- vide a framework in which to understand experimental Jovian cloud studies. Results from Galileo The Galileo mission is not over, and much of the data from the 16 orbits which had been completed at the time of writing has still to be analysed. However, preliminary results from the probe and from the orbiter were published nearly two years ago, and the full set of probe results, plus the first papers on fully mature orbiter analyses, have recently appeared (Journal of Geophysical Research, 25 September 1998). Thus, it is a good time to consider how Galileo has revised our basic picture of the Jovian clouds and what questions remain. The cloud structure on Jupiter is complicated. The easiest way to consider it further is in terms of five different aspects, initially considered separately. These are: (i) The nature of the hazes near the tropopause and above, which form the uppermost layer or layers. (ii) The main cloud decks which lie at pressures of around 0.7 and 1.5 bars respectively. These probably consist of NH3 and NH4SH and are the main opacity sources accounting for the variable appearance of Jupiter in the visible and infrared. (iii) The deep cloud layer, which is near the depth limit of detectability for June 1999 Vol 40 Jupiter 0.01 stratospheric haze pressure (bars) 0.10 haze, r = 0.5 µm NH3, r = 0.75 µm NH4SH, r = 1.00 10.00 100 5 µm contribution function 150 200 250 300 temperature (K) remote sensing, and which is predominantly aqueous. (iv) The horizontal variability in all of the above, and finally, as a special case of (iv), (v) the cloud structure in compact features like the giant eddies, especially the GRS. Hazes The term “haze” is used to describe particulate layers which are extensive horizontally, and often vertically as well. Haze layers are generally relatively featureless compared to clouds, with more clear space between the droplets or ice particles, and have less total optical depth. Jupiter is hazy at heights well above the main cloud decks. This is not unexpected since the atmosphere contains a substantial amount of methane, which at high levels will be subjected to dissociation by solar ultraviolet radiation, followed by recombination into higher hydrocarbons which condense into liquid droplets. This type of process is thought to go on in the upper atmospheres of all the giant planets, and something similar, but involving more nitrogen compounds, is probably responsible for the deep haze on Titan which tends to obscure the surface of the satellite and gives rise to its orange coloration. The term “tholins” (meaning muddy) is sometimes used to describe the composition of this range of hydrocarbon and other condensates of indeterminate composition, and of the material which is produced in the laboratory by simulating the atmospheric processes with UV lamps or electrical discharges. A preliminary analysis (Banfield et al. 1998) of the Galileo SSI multispectral observations at visible and near-infrared wavelengths suggests that the haze on Jupiter is made up of two separate layers, one high in the stratosphere, identified with the methane-derived tholin haze described above, and another in the upper troposphere. The visible optical depth of the latter has been estimated by the SSI investigators to be in the range from 2 to 6, as opposed to only about 0.1 for the stratospheric haze. These numbers are probably too high, because the SSI analysis assumed perfectly conservative June 1999 Vol 40 0.45 µm –50 µm 350 400 3 Vertical temperature profile and cloud locations on Jupiter, as retrieved from a NIMS spectrum at a single location in the North Equatorial Belt, near where the Galileo probe entered some six months earlier (after Irwin et al. 1998). scattering by the haze particles. The actual value of the single scattering albedo is currently unknown, but is likely to be less than 100%, in which case the optical depth estimates would decrease. It can be speculated that this lower haze is made up of products of the photolysis of ammonia rather than methane, since it lies in the stably stratified region just above the top of the tropospheric convection cells and just above the thick ammonia cloud. The ammonia gas required to produce it could penetrate into the stratified region as a result of breakthrough convection similar to that which injects wet tropospheric air into the lower stratosphere on the Earth. It has been argued on chemical grounds that the principal component of the processed and recondensed ammonia would probably be hydrazine (Strobel 1973). However, hydrazine exhibits in the laboratory the strong spectral signature near 3 µm which is also characteristic of ammonia itself. Galileo infrared spectra indeed show this feature in the main ammonia cloud and from the broadened shape of the absorption feature, a large particle size is inferred. However in the overlying upper tropospheric haze, where small particles are required by the reflectance spectra, small hydrazine or ammonia particles do not seem to be suitable candidates for the primary component since would both contribute sharp, narrow absorption features in the 3 µm region which are not seen. A very recent study (Guillemin et al. 1999) offers a possible explanation in that it suggests that the main product of the combined photochemistry of NH3 and PH3 should be not N2H4 but P2H4. It remains to be shown whether a composition of condensed P2H4 and higherorder products (“pholins”) is consistent with the spectroscopic observations. The visible clouds The white cloud which forms many of the features observed on Jupiter is condensed ammo- nia ice. We know this because it occurs at the level (about 0.7 to 1.0 bar) where the temperature is appropriate for condensation of this species in thermodynamical models such as those discussed above. It also has the right spectral properties (Encrenaz et al. 1996). The ammonia cloud is optically thick in the cloudy zones in the visible and near-infrared, but partially transparent at wavelengths of 5 µm and longer. This is the cloud which is primarily responsible for the visible cloud structure on Jupiter in ground-based and spacecraft imaging experiments, and SSI observations place the features in a narrow height range near 750 mb. The primary evidence that the ammonia cloud is transparent at thermal infrared wavelengths is that the colour temperature of the emission from the cloudy zones, as measured by NIMS, is inconsistent with a source as cool as the ammonia cloud (Drossart et al. 1998). The 5 µm window spectra are typically biased more towards the blue end of the spectrum than the Planck function for the temperature of the ammonia cloud would predict, even on the night side of the planet. Most of the radiance being measured by the spectrometer must therefore have originated below the ammonia cloud, and must propagate through it. In addition, an anticorrelation is observed by NIMS between the brightness in the reflected continuum and in the absorption bands of ammonia, that also suggests that the emission in the belts is being attenuated by the ammonia ice cloud. Histograms of the brightness of Jupiter at 5 µm on the night side (Cameron-Smith 1998) show a narrow peak at 165 K (1.0 bar) and a much broader one at ~240 K (3.5 bar), corresponding essentially to the zones and belts respectively. Day–night differences at 5 µm show that reflected solar as well as emitted thermal radiation contributes to the emission on the day side. Spectral fits to the night side 5 µm flux (Drossart et al. 1998) show that it is due to deep thermal emission attenuated by the cold ammonia cloud and that the transmission of the latter is about 0.002, and its reflectivity about 0.15 at 5 µm. It must lie at p < 0.79 bar or its own emission would become significant enough to destroy the fit. The corresponding values at 4 to 4.5 µm are reflectivity ~0.04 and p < 0.69 bar. The histograms show that these values apply over a wide area on Jupiter. Cameron-Smith (1998) has speculated as to how the cloud opacities in the zones should come to have these preferred values. He showed that the limb darkening in the zones is too flat to correspond to a purely absorbing cloud layer; that would give rise to a dependence of the emitted intensity on the secant of the emission angle, which is not observed. He concluded that the cold (NH3) cloud cannot be a pure absorber but could be strongly scattering or contain “microholes” produced dynamically, perhaps 3.23 Jupiter by the Jovian analogue of “Rayleigh–Benard” convection cells on the Earth. Tb = 165 K Tb = 240 K zone belt Tb = 275 K Deep clouds Deeper in the atmosphere, below the ammonia cloud, thermochemical models lead us to expect that an ammonium hydrosulphide (NH3.H2S or NH4SH) cloud will condense at around 1.5 bars and 180 K, the precise level depending on the deep-atmosphere abundances of NH3 and H2S among other factors. Retrievals of the vertical run of cloud opacity which fits the NIMS 5 µm spectra described above place the lower cloud opacity at about the level the models predict for ammonium hydrosulphide (figure 3). This work also shows that the NH4SH cloud has some large particles, and that it, and not the ammonia ice cloud, is the main variable opacity source on Jupiter at thermal infrared wavelengths. This in turn suggests that the NH3 and NH4SH clouds tend to occur in the same locations, since the features in visible and infrared (5 µm) images are anticorrelated. The water cloud predicted by the thermochemical models would, on the temperature profile measured by the Galileo probe, lie at a pressure of around 5 bar. This is approaching the maximum depth sounded by NIMS, and this is achieved only in hot spots where the water cloud may be reduced or eliminated by strong vertical motions. In one particularly transparent region near the Great Red Spot, however, the SSI team made a rare detection of a very deep cloud roughly 1000 km across and lying at pressures perhaps as great as 4 bar, which they tentatively identify as a water cloud since no other known species on Jupiter would condense at this level. Horizontal variability Discussion of the Jovian clouds has always discriminated between the relatively cloudy zone regions and the more cloud-free belts. As noted above, the reason these exist is because of the release of internal energy, which has to be advected up to levels near the main (ammonia) cloud-tops, where it can radiate to space. The process forms planetary-scale convection cells, which have radial symmetry as a consequence of the rapid rotation of the planet. The reason for the number and spacing of belts and zones, which are sufficiently invariant that the most prominent have received descriptive names (e.g. North Equatorial Belt, South Temperate Zone), is still a matter for conjecture, primarily in the form of detailed fluid dynamical modelling (e.g. Williams 1979). The rising part of each cell consists of air carrying condensates such as water and ammonia which condense at the appropriate temperature and pressure for that species, forming the zone clouds. The descending branch of each cell is relatively depleted in volatiles and is less cloudy 3.24 temp. (K) press. (bars) 110 0.05 ‘tholin haze’ 0.5/0.04 110 hot spot 0.5/0.04 0.5/0.04 0.18 ‘pholin haze’ 135 0.7 20/6 2.2/0.07 2.2/0.07 160 1.4 ?/? 2.0/0.07 2.0/0.07 180 1.6 ?/? 1.2/1.3 0.0/0.0 ~275 ~5? ?/? ?/? 0.0/0.0 ascending air ammonia cloud hydrosulphide cloud and haze water cloud descending air 4 A model of the Jovian cloud structure and properties based on the available data and simple physical–chemical models. The approximate temperature and pressure levels where the principal cloud layers occur are shown to the left, while the representative visible (0.7 µm) and infrared (5.0 µm) optical depths are marked on or below the layers themselves. The quantity Tb shown at the top is the typical value of the brightness temperature at 5 µm for each type of region. as a result, forming the belts. To these two kinds of region we must add a third: the hot spots. These are regions where the local meteorological conditions are such that the column is nearly completely free of cloud, at least down to the level where water condenses, and in some cases deeper than that. The detection of a deep water cloud by the Galileo imaging team, while the probe, which descended in a hot spot, found no water cloud at all, implies horizontal inhomogeneity in the deeper clouds also. This is not surprising since the convection cells on Jupiter must extend to a great depth (although not necessarily as single cells; chains of cells could also do the job and may be more stable). Figure 3 shows a five-component model of the cloud system with typical values of the cloud opacity in each vertical component for the three types of region: zone, belt, and hot spot. However, as we have already seen, this is still a gross over-simplification. The atmosphere is very inhomogeneous horizontally as well as vertically, even within a given region. This is illustrated by plotting the centre-to-limb variation in radiance within a zone, for example, as shown in figure 5. This figure shows how futile it is in this case to try to fit centreto-limb scans to a simple model of Jupiter, although this was often tried in the past with lower-resolution Earth-based data. It is better, but more difficult, to use observations of the same location from different points on the orbit, thus varying the scattering angles while viewing the same atmospheric column. The SSI data detected no lateral structure in either of the haze layers, except at high latitudes, over the poles. However, in the NIMS data a wide band of high haze is clearly seen centred over the boundary between the Equatorial Zone and the North Equatorial Belt, about 0 to 9°N (figure 2b). While it is not surprising that haze is thickest over the poles (where the high-altitude air tends to be stagnant and precipitation of energetic particles occurs) and over the equator (where the solar intensity and therefore the production of tholins is greatest), we must again invoke some unknown property of the dynamics of the atmosphere, this time at a relatively high level, to explain why this is not centred over the equator. The giant eddies The most common giant eddies are the white ovals, which are all regions of enhanced cloud. The NH3 cloud is optically thick in these spots, as presumably are the ammonium hydrosulphide and water clouds if we could see them. The clouds themselves are generally several kilometres higher in altitude, and the SSI data shows that the hazes, too, are elevated over the GRS and the white ovals. The best-studied and probably the most interesting dynamical feature on Jupiter, with the possible exception of the basic belt-zone structure, is the Great Red Spot. Figure 6 shows a schematic of a proposed structure for the GRS, June 1999 Vol 40 Jupiter radiance (µWcm–2 Sr–1 µm–1) 0.40 0.30 0.20 χ = 0.01 0.10 χ = 10 χ=3 χ=1 χ = 0.03 χ = 0.3 0.00 0 20 60 40 emission angle (degrees) χ = 0.1 80 5 Jovian limb-darkening curves, showing the thermal emission from Equatorial Zone clouds in a wavelength band from 4.6 to 5.2 µm. The broken curves are NIMS measurements obtained within a few hours of each other at slightly different latitudes, all within the range 2 to 13°N. The solid lines are theoretical limb-darkening curves for a simple cloud model, consisting of a single layer of cold, grey cloud. The large deviation of the measurements from the models, and from each other, show how inhomogeneous the Jovian atmosphere is, even within a region which appears nearly uniform in visible images (Cameron-Smith 1998). 6 A conceptual model of the structure of the Great Red Spot based on Galileo observations and an analogy with terrestrial hurricanes. which has some structural features analogous to terrestrial hurricanes, in that a relatively narrow column of rising, moist air spreads out into a canopy with spiral arms of cloud extending above the surrounding cloud tops. In the highest resolution NIMS images at 5 µm, the gaps between the cloud spirals can be seen to be bright, due to emission from considerable depths, indicating clear regions of atmosphere beneath the overlying canopy. From the size of the GRS and its unique colour, it seems likely that the rising moist air which forms the Spot originates from a considerable depth, deeper than any of the cloud layers already discussed. The condensed material in the cloud will then be a mixture of ammonia, ammonium hydrosulphide, water and other materials possibly including phosphorus. The last of these is mentioned particularly because, in addition to being present in large amounts on Jupiter, as evidenced by the observed abundance of phosphine, phosphorus has the property of being intrinsically red, and truly red condensate cloud materials are not common in nature. Dry, cloud-free air descends around the spot, and the cloud cap is tilted, presumably as a June 1999 Vol 40 7 The detailed morphology of the Great Red Spot inferred from NIMS observations (Baines 1996), showing the spiral structure in this anticyclonic feature. The plot covers 50° in longitude, 30° in latitude, and approximately 20 km in vertical relief (Baines 1996). result of interactions between the vortex and the prevailing wind in which it is embedded. Figure 7 shows a cloud-top height map inferred from NIMS data, from which it can be seen that the cloud forming the visible top of the GRS is about 10 km higher at one side than the other, a significant distance in terms of vertical structure (about half of one pressure scale height) but small, of course, compared to the mean radius of the GRS, which is somewhat larger than that of the Earth. Outstanding questions Much progress on deciphering the Jovian clouds has been made with Galileo, but some of it is quite speculative still and many questions remain. Perhaps outstanding among these, other than almost everything to do with the atmospheric dynamics, is the fact that we still have no definite explanation for the colours. The white clouds are surely ammonia, but the yellows and browns are harder to account for since pure ammonium hydrosulphide is white. Of course, none of the clouds is likely to be pure, and there are many coloured compounds of sulphur, including allotropes of the element itself, which could account for the colours on Jupiter if they were present due to photochemistry or other nonequilibrium processes. However, we are not at present much further forward on this point than we were a decade or more ago. Another major question remaining is that of the Jovian dynamical meteorology. We see, and have begun to characterize, the massive global variability of cloud features and aspects such as their latitudinal and temporal variations. However, we have no comprehensive theory of the general circulation on Jupiter, nor do we understand the origins and true natures of the giant eddies and the hot spots, to name only the two most conspicuous aspects of Jupiter’s weather. The variability within the cloud structures which we have attempted to categorize here in an averaged sense is very large, some of it appearing organized and some chaotic. The nature, and perhaps the origins, of these fluctuations will become clearer as Galileo extends its coverage in space and time and more of the data is analysed. The Composite Infrared Spectrometer on Cassini will obtain coverage of Jupiter during its fly-by in December 2000. Its spectral resolution and range are both superior to NIMS, being designed to study atmospheric chemistry on Titan, and should help to further unlock the secrets of Jupiter’s colourful cloud chemistry. ● Fred Taylor and Patrick Irwin are at the Dept of Atmospheric, Oceanic & Planetary Physics, Oxford University. The authors acknowledge the contribution of their colleagues on the Galileo Near Infrared Mapping Spectrometer Team, especially Principal Investigator Dr R W Carlson of the Jet Propulsion Laboratory, California Institute of Technology; Dr A L Weir and Dr P Cameron-Smith, Oxford University; Dr K H Baines, JPL; Drs T Encrenaz, P Drossart, and M Roos-Serote, of the Observatoire de Paris, Meudon. They also thank all of the other Galileo scientists and engineers, especially Project Scientist Dr T V Johnson of JPL. References Carlson R W et al. 1992 Space Sci. Rev. 60 457. Irwin P G J et al. 1998 J. Geophys. Res. 103 E10, 23,002. Roos-Serote M et al. 1998 J. Geophys. Res. 103 E10, 23,023. Cameron-Smith P J 1998 D.Phil. Thesis, University of Oxford. Baines K H 1996 Paper presented at the American Astronomical Society, Division for Planetary Sciences, Tucson, Oct 1996. Belton M J S et al. 1996 Science 274, 377. Banfield D et al. 1998 Icarus 135 1, 230. Hanel R et al. 1981 J. Geophys. Res. 86 8705–12. Strobel D 1973 J. Atmos. Sci. 30 1205. Guillemin J-C et al. 1999 in press. Encrenaz Th et al. 1996 Astronomy & Astrophysics 315 317. Drossart P et al. 1998 J. Geophys. Res. 103 E10, 23,043. Williams G P 1979 J. Atmos. Sci. 36 932–968. 3.25
© Copyright 2026 Paperzz