This dissertation has been more in the nature of a survey than a focused attack on a particular well-defined problem. The conclusions are therefore varied. I present them in this chapter, arranged into categories corresponding to the various lines of investigation. I finish with an anticipation of future lines of research, especially the exciting potential of the Galileo Jupiter orbiter mission, currently expected to reach Jupiter in the mid-90's.

Shape of Thermal Emission Spectra

The emission spectra are smooth and featureless (Chapter 4). This may constrain surface composition and texture but not without more laboratory and theoretical work. However the smoothness of the Callisto and Ganymede spectra, compared to those from the Earth's moon, shows that, if these objects have segregated surfaces (in which case most of the emission is from the dark component), the dark material does not have lunar-like silicate composition and texture.
Perhaps the most important conclusions of the analysis of the IRIS spectra concern the variations in brightness temperature with wavelength that are seen in the local thermal emission from the Galilean satellites. This dissertation demonstrates for the first time that the shape of the thermal emission spectrum of an airless body, and the variation in shape across the surface, is a useful probe of surface physical conditions.
The spectrum slopes observed by IRIS are real, and are probably due to local temperature contrasts, not spatially resolved, with amplitudes of tens of degrees. The different patterns of variation of spectrum slope on Europa, Ganymede and Callisto imply that the slopes have more than one cause, different causes being dominant on different bodies (Chapter 4).
On Callisto, the slopes of the daytime spectra appear to be due largely to topography, and become more pronounced towards the terminator as shadows lengthen and topographic temperature contrasts increase. A roughness comparable to the Moon's may be sufficient to produce the observed slopes, though better topographic temperature models are required to confirm this (Chapter 6). Callisto near-subsolar spectrum slopes are smaller than on Ganymede, but a little steeper than predicted by current topographic models. This may indicate the presence of separated dark and bright materials on the surface, similar to the situation inferred on Ganymede (see below), but with one component (the dark, low thermal inertia one) dominant, occupying 80%--90% of the surface (Chapter 7). Near-subsolar spectrum slopes are steeper in the Voyager 2 data (late morning, trailing/farside quadrant) than Voyager 1 (early afternoon, leading/nearside quadrant). The slopes on the trailing hemisphere are thus closer to those on Ganymede, which is interesting as this hemisphere is the more Ganymede-like in its photometry and polarimetry also (Chapter 4).
The nighttime Callisto spectra are also quite steep, steeper than those on Ganymede, and require temperature contrasts with amplitudes of around 30oK. Whether this is due to `remnant' topographic effects cannot be checked without much more complex thermophysical topographic models. Alternatively, if the possible minor bright component on the surface has high thermal inertia, this might also explain the steep nighttime slopes on Callisto (Chapter 7).
The Ganymede spectrum slopes are probably not due to topography, because they do not increase towards the terminator, and are in fact remarkably uniform across the satellite. The shallowness of the near-terminator daytime spectra indicates a smoother surface than Callisto at an undetermined lengthscale: the higher albedo of Ganymede cannot explain the apparently smaller topographic temperature contrasts than on Callisto (Chapter 6). The slopes are most easily explained by a surface with similar coverage of horizontally separated dark, low thermal inertia and bright, high thermal inertia components (Chapter 7). The very bright ejecta blanket of the crater Osiris shows unusually small slopes, which is consistent with the 2-component interpretation as one component (the bright one) would be expected to be dominant in this region, resulting in reduced temperature contrasts. By far the steepest spectrum slopes seen on Ganymede occur near the south pole, for reasons that are not yet clear.
The spectrum slopes on Europa are smaller than on either Ganymede or Callisto, and require local temperature contrasts of 20oK or less (Chapter 4). The available coverage is too limited to constrain the origin of the slopes. However their small amplitude is consistent with the impression, obtained from imaging and reflectance spectroscopy, that Europa's surface is topographically smooth and compositionally homogeneous.

Anisotropy of Thermal Emission

As far as can be determined from the Voyager IRIS data, the falloff in thermal emission from a given point on Ganymede or Callisto with increasing solar phase angle is comparable to that on the Earth's Moon (Chapter 5 and Appendix A).

Global Temperature Distributions

Equatorial surface temperatures are significantly below equilibrium (zero thermal inertia) values. The maximum daytime temperatures on Ganymede and Callisto are about 158 and 147oK respectively, approximately 10oK and 5oK respectively below equilibrium. Nighttime temperatures are warmest on Ganymede, which is 15oK warmer in the early part of the night than Europa or Callisto (Chapter 5). This is due to Ganymede's higher thermal inertia and faster rotation than Callisto, and its much lower albedo that Europa (Chapter 7). Anomalously cold regions, not associated with high surface albedos, were seen by Voyager 1 on Ganymede at
20o N, 10o W (in the early afternoon), and at 0o N 250o W (just before midnight). Large anomalies in thermal properties are required to explain these (Chapter 7). Similarly, temperatures around 20o N, 330o W on Callisto (in the late afternoon) are also anomalously cold.
The 2-layer surface models of Morrison and Cruikshank (1973) and Hansen (1973) explain both the eclipse and diurnal thermal behaviors of Ganymede and Callisto, but only when the top-layer thermal inertias that they used are increased by about 50%. The diurnal profiles constrain the thermal inertia of the lower layer of this model to be a factor of several lower than solid­ ice values. The eclipse and diurnal behavior of Ganymede, however, are also consistent with a 2-component surface with equal surface coverage of a bright high thermal inertia and a dark low thermal inertia component (Chapter 7).

Ice Mobility

Thermal segregation of water ice on the icy Galilean satellites is a potentially powerful process that may dominate the appearance of the surfaces at sub-kilometer length scales. Ganymede and Callisto, geologically dead at the scale of the Voyager images, may show features at small scales that reflect equilibrium with a process that operates on a timescale of decades or less (Chapter 8). The competing ice redistribution processes of ion sputtering and micrometeorite bombardment are unlikely to prevent segregation except possibly in polar regions and perhaps on Europa (Chapter 9), especially the trailing hemisphere where ion sputtering is intense. Evidence for segregation includes direct imaging by Voyager at Callistoan high latitudes, the apparent lack of the kilometers of poleward ice migration that would be expected on Ganymede and especially Callisto if the ice were unsegregated and warm, and the preservation of local albedo contrasts at low latitudes on Ganymede and Calliso. The steep subsolar IRIS spectrum slopes on Ganymede may also require a segregated surface (Chapter 7). The reflectance spectra of Ganymede and Callisto may be consistent with segregated surfaces, though Europa's spectrum probably indicates a homogeneous surface (Chapter 10).
If the surfaces of Ganymede and Callisto are influenced by thermal segregation, the dark material on their surfaces may be mostly in the form of thin lag deposits, and the substrate composition is likely to be more ice-rich than the visible surface. Segregated surfaces will have lower ice mobility and possibly much lower overlying atmospheric pressure (Chapter 11).

Potential for Future Work


The Galileo Jupiter orbiter mission will enormously improve our knowledge of the surfaces of the Galilean satellites.
Thermal observations, comparable to those made by the Voyager IRIS instrument, will be obtained by the Galileo Photopolarimeter-Radiometer (PPR) (see e.g. Hunten et al. 1987). Though designed primarily for Jovian atmospheric observations, (like IRIS), it will also be very suitable for satellite work. Below I compare its characteristics to those of IRIS.
Spectral Resolution. Five broadband flux measurements centered on 17, 21, 27.5, 37, and above 42 microns, compared to the equivalent of about 250 narrowband measurements in the 8--50 microns range for Voyager IRIS.
Angular Resolution. 0.14o diameter circular field (IRIS field is 0.25o diameter).
Spatial Resolution. For typical Galileo satellite flyby altitudes of about 1000 km, much closer than the Voyager flybys, the smallest regions measured will be 2.5 km across, compared to 100s of km for Voyager. Coverage with such high resolution will be very limited but there will be good coverage at much higher spatial resolutions than possible with Voyager.
Acquisition Rate. One set of fluxes every 18 seconds in `cycle mode' (photopolarimetry plus radiometry) compared to a spectrum every 48 seconds from Voyager.
Spatial Coverage. Much more comprehensive than from Voyager, owing to the larger number of flybys. There will be perhaps 11 targeted encounters of the icy Galilean satellites, plus many untargeted encounters which will often provide spatial resolution equal to or better than the Voyager flybys.
So in all respects but spectral resolution Galileo will substantially exceed Voyager's capabilities for measuring thermal radiation from the icy Galilean satellites. The low spectral resolution is not a great handicap, given the smoothness of the thermal spectra revealed by Voyager. If the data are well calibrated, spectrum shapes directly comparable to those obtained here (Chapter 4) will be extractable from the radiometer fluxes.
The Galileo data will allow confirmation of the trends of spectrum shape seen by Voyager, and better characterize the trends on Europa and on the night sides of all satellites. Differences in thermal emission with terrain type will be much more easily visible than with Voyager. Improved absolute pointing accuracy (if only because of the close flybys) and more comprehensive phase coverage should allow characterization of the beaming of thermal emission and correction for it. Complete diurnal temperature profiles should be obtainable for all three satellites. It will be possible to compare different longitudes at similar illumination and viewing geometries on successive flybys, so that the global variations in thermal properties can be better characterized.
There is also the possibility of thermal observations of the Earth's Moon and several asteroids by Galileo on its way to Jupiter. These observations, though limited, will be invaluable for comparison with the Galilean satellite data.
The superior Galileo data set, by allowing better separation and understanding of geographical (global and terrain), diurnal, and geometrical (beaming) effects on the observed thermal emission from the satellites, should provide much tighter constraints on thermal models like those described here in Chapters 6 and 7. It should therefore be possible to determine much more clearly the distribution of surface roughnesses, albedos and thermal inertias on the satellite surfaces.
The other Galileo instruments will also provide a great deal of information that is relevant to the science discussed in this dissertation, especially the possibility of widespread ice segregation. Imaging of the satellite surfaces at very high resolution may detect and characterize ice segregation if it exists, and determine its dependence on latitude and terrain type. The NIMS (near-infrared mapping spectrometer) instrument will also be able to study latitudinal variations in ice distribution on each icy satellite, and will have sufficient resolution, for instance, to obtain separate spectra of the bright and dark segregated patches visible in Fig. 32 at high latitudes on Callisto, thus possibly obtaining a `pure' spectrum of the dark, non-icy surface component. Characterization of the latitudinal variations in ice distribution should allow identification of the dominant redistribution processes and their effects.
The UV stellar occultation technique, used by Voyager to rule out a \mubar oxygen atmosphere on Ganymede, is sensitive enough for direct detection of the water vapor in equilibrium with warm ice on the Galilean satellites (Spencer, 1982). Stellar occultations observed with the Galileo UVS instrument will thus be useful indirect probes of surface ice temperature and mobility. In addition, the plasma instruments and dust detector on board Galileo will greatly improve our understanding of the rates and nature of ice transport processes competing with thermal sublimation.

Other Work

Improved topographic temperature models (Chapter 6) are fairly easy to create, and should allow much better understanding of the effects of topography on thermal spectrum shape and on beaming. With the help of such models should come a better understanding of the spectrum shapes on Callisto and better constraints on the surface roughness on Ganymede and Europa.
Improved eclipse models are possible and should be studied (Chapter 7 and Appendix C). In particular, subsurface penetration of sunlight and oblique eclipse geometry should be investigated. Europa's lack of post-eclipse recovery is a major outstanding problem to be explained. The use of least­ squares fitting techniques to determine simultaneous best-fit thermal models to the eclipse, diurnal, and spectrum slope data may allow a definitive explanation of Ganymede's thermal emission behavior, which will show whether a segregated or laterally-homogeneous surface exists there.
Improved groundbased measurements of the thermal emission of Europa, Ganymede, and Callisto as a function of orbital phase, solar phase, and wavelength would be complementary to the IRIS data and provide an additional constraint on comprehensive thermal models. Published values, such as those tabulated by Morrison (1977), are barely of adequate quality to provide good constraints, especially when comparing fluxes in the 10- and 20\dmic\ regions.
Photometry of Voyager images might provide an independent check on the conclusion, derived from the IRIS data, that Callisto has a rougher surface than Ganymede (Chapter 6). By using the Hapke (1984) equations, particle albedo, surface texture, and larger-scale surface topography can be determined independently from a complete enough photometric data set, and this has been done for Ganymede by Helfenstein (1986). A similar exercise for Callisto might allow quantitative comparison of its surface morphology with Ganymede's in a way that could be compared to the thermal emission data. However, because of possible effects such as the lateral conduction of heat, the scale of the topography that affects the thermal emission may be different from that that affects the scattering of sunlight, and the two data sets may not be directly comparable.
The present study of the thermal emission from the Galilean satellites has been hampered by the lack of comparable disk-resolved data from other airless bodies. The thermal emission from the Earth's Moon is still poorly described or understood, and a more complete survey, especially of the beaming of thermal emission and the wavelength dependence of the emission, is badly needed. I am not aware of any published data on the emission from the moon in the 20\dmic atmospheric window, for instance, that would show whether the Moon's emission spectrum steepens towards the terminator as does Callisto's. A complete survey may have to wait for the revival of interest in Lunar studies, which have been dormant since the early 70's.