Available online at www.sciencedirect.com ?m~^^ ELSEVIER ScienceDirect Icarus 191 (2007)702-711 ICARUS www.elsevier.com/locate/icarus Arecibo radar observations of Rhea, Dione, Tethys, and Enceladus G.J. Black^'*, D.B. Campbell^ L.M. Carter^ ^ Department of Astronomy, University of Virginia, PO Box 400325, Charlottesville, VA 22904, USA Department of Astronomy, Cornell University, Ithaca, NY 14853, USA ^ Smithsonian Institution, Center for Earth and Planetary Studies, Washington, DC 20013, USA Received 19 September 2006; revised 25 May 2007 Available online 25 July 2007 Abstract We have measured the bulk radar reflectance properties of the mid-size saturnian satellites Rhea, Dione, Tethys, and Enceladus with the Arecibo Observatory's 13 cm wavelength radar system during the 2004 through 2007 oppositions of the Saturn system. Comparing to the better studied icy Galilean satellites, we find that the total reflectivities of Rhea and Tethys are most similar to Ganymede while Dione is most similar to Callisto. Enceladus' reflectivity falls between those of Ganymede and Europa. The mean circular polarization ratios of the saturnian satellites range from ~0.8 to 1.2, and are on average lower than those of the icy Galilean satellites at this wavelength although still larger than expected for single reflections off the surface. The ratio for the trailing hemisphere of Enceladus may be the exception with a value ~0.56. The 13 cm wavelength radar albedos and polarization ratios may be systematically lower than similar results from the Cassini orbiter's RADAR instrument at 2.2 cm wavelength [Ostro, S.J., and 19 colleagues, 2006. Icarus 183, 479^90]. Overall, these reflectivities and polarization properties, together with the shapes of the echo spectra, suggest subsurface multiple scattering to be the dominant reflection mechanism although operating less efficiently than on the large icy moons of Jupiter. All these saturnian moons and icy jovian moons are atmosphere-less, low temperature water ice surfaces, and any differences in radar properties may be indicative of differences in composition or the effects of various processes that modify the regolith structure. The degree of variation in radar properties with wavelength on each satellite may constrain the thickness and efficiency of the scattering layer. ? 2007 Elsevier Inc. All rights reserved. Keywords: Saturn, satellites; Radar observations; Enceladus 1. Introduction Saturn's moons Rhea, Dione, Tethys, and Enceladus are a class of airless, icy bodies 500 to 1500 km in diameter that despite these relatively small sizes still show evidence of vary- ing degrees of geologic activity in the past (Plescia and Boyce, 1985) and, in the case of Enceladus, the present (Porco et al., 2006). The collective formation of these moons and subsequent evolution remains an outstanding problem. Their bulk densi- ties range from a high of 1.6 g/cm^ for Enceladus (Porco et al., 2006) to perhaps as low as 1.0 g/cm^ for Tethys (Dourneau and Baratchart, 1999), betraying compositions of mostly water ice but with formation histories that led to differing non-ice frac- tions and interior structure. Compositions are likely related to Corresponding author. Fax: -l-l 434 924 3104. E-mail address: gblack@virginia.edu (G.J. Black). 0019-1035/$ - see front matter ? 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.icarus.2007.06.009 the positions of their formation from Saturn, especially with re- gard to the fraction of ammonia (e.g., Mosqueira and Estrada, 2003) which is expected to be prevalent in these satellites due to the cool temperatures in the saturnian nebula (Lewis, 1972). The Cassini mission has recently performed close flybys of these moons, confirming that varying degrees of geologic activ- ity (Wagner et al., 2006) were sustained for an extended period of time on these small objects after their formation, and con- tinuing to the present on Enceladus (Porco et al., 2006). The surfaces of each are known to be fairly clean water ice (Clark et al., 1984) with high optical albedos, and with each exhibiting color and brightness variations indicating heterogeneous cov- erage of non-ice impurities and surface microstructure. Large scale variations take the form of leading versus trailing hemi- sphere asymmetries (Buratti and Veverka, 1984), which for these synchronously rotating bodies may be related to interac- tions with Saturn's magnetospheric plasma and/or the diffuse Arecibo observations of Saturn's satellites 703 E-ring in both of which these moons are embedded (Verbiscer et al., 2007), or focusing of micrometeorite impacts due to their orbital motion (cf. Buratti et al., 1990). Measurements of radar properties and any inter-satellite variations can address these issues of surface characteristics and formation histories. The radar reflectivity of a surface de- pends on its composition and structure, so we undertook an effort to measure the radar scattering properties of these ob- jects. Materials with high dielectric constants and loss tangents, such as the silicate surfaces of the terrestrial planets and as- teroids, will to first order yield reflections from the vacuum- surface interface and have radar albedos of <0.2 (cf. Ostro, 1993). For surfaces smooth on horizontal scales within roughly an order of magnitude of the incident wavelength, single re- flections dominate and the reflectivity is directly related to the bulk dielectric constant of the surface. In contrast, surfaces of water ice, if extremely clean and at temperatures around 100 K, can have essentially negligible absorption at radio wave- lengths (cf. Thompson and Squyres, 1990) and subsurface mul- tiple scattering mechanisms become important. Thus the radar can probe wavelength-scale structures not only on, but below the surface at depths which depend (perhaps sensitively) on non-ice contaminants. The canonical examples of this prop- erty are the icy Galilean satellites for which there is no evi- dence of specular surface reflections at wavelengths from 3.5 up to 70 cm, and their entire echoes are apparently the result of multiple scattering (Campbell et al., 1978; Ostro et al., 1992; Black et al., 2001a). They are strongly backscattering and pref- erentially preserve the polarization sense under circularly polar- ized illumination, a signature of multiple scattering and unlike single reflections which reverse the sense. Various models have been explored to quantitatively explain the high albedos and polarization ratios. All rely on subsur- face scattering and range from those that consider the effects of buried structures such as craters (Baron et al., 2003; Gurrola, 1995; Eshleman, 1986), randomly oriented facets (Goldstein and Green, 1980), or refractive lenses (Hagfors et al., 1997), to models of distributions of non-specific but efficient scatter- ers (Black et al., 2001b) that produce a coherent enhancement in the backscatter direction (cf. Hapke, 1990; Peters, 1992). We present new 13 cm wavelength radar observations of mid-size saturnian satellites and only a few qualitative interpretations, leaving quantitative modeling for future efforts. 2. Observations We observed these saturnian satellites with the Arecibo Ob- servatory's 13 cm (2.4 GHz) radar system during four epochs from 2004 through early 2007 at times near opposition in order to minimize the Earth-to-target distances. A log of the observa- tions is given in Table 1. The round-trip light-travel time to the Saturn system at opposition is ~2.25 h, permitting ~30 min of observing time per day due to the Arecibo telescope's restricted elevation range. Equipment problems resulted in some observ- ing sessions being shorter than this maximum tracking time. In 2006 only half the transmitted power relative to prior years was available since one of the two transmitter klystrons was out for maintenance. In 2004, short radar observations of the icy Galilean satel- lites whose radar characteristics are well known were made to assist in calibrating the telescope gain, which is 73.3 db av- eraged over the zenith angles of 12.5?-19.7? covered by the observations, consistent with routinely measured standard as- tronomical calibrators. For the Arecibo telescope, both gain and system temperature are functions of zenith position. The system temperature was measured nightly and varied between 25 and 32 K. Depending on its location in the beam, the ther- mal flux from Saturn can increase the base system temperature by about 10%, as expected from its decimeter brightness tem- perature of ~200 K (de Pater and Dickel, 1991). Reflection of the radar signal from the ring system is also received, but is spread over ~650 kHz and blends into the spectral back- ground removed during reduction. With the reflectivity of the rings around 30^0% (Nicholson et al., 2005), they contribute at most 0.5 K to the system temperatures. On each date a circularly polarized monochromatic signal was transmitted and the echo was recorded in both circular po- larizations, labeled as SC for the same circular sense as the transmitted signal and OC for the opposite circular sense. Spec- tra of the recorded signals were then made to isolate the echo in frequency. The rotation of a satellite induces a varying Doppler shift in the reflected signal across its disk. For these targets this effect spreads their echos into bandwidths of order several 100s of Hz at the radar's frequency. Table 2 gives the physical parameters of each satellite and the resulting average Doppler bandwidth of each, which varies by ~ ?3% year-to-year due to the changing latitude of the sub-Earth track. Only this average bandwidth was used in the final analysis. For each target, the spectra, si, from all dates are weighted and summed to produce a final echo spectrum, S, as shown in Fig. 1 according to I^w/?; E' where the weight for each spectrum is given by PgtgrV^ (1) (2) ' sys for transmitted power P, antenna gains gr and gr at trans- mit and receive positions respectively, integration time T, and system temperature Tgys- Disk integrated scattering parameters reduced from these spectra are given in Table 3. The total echo power in each spectrum was converted to a physical cross sec- tion via the system parameters and known target distance, then normalized by the target's projected area to convert into a spe- cific cross section, hereafter referred to as the radar albedo. Unless noted, the quoted uncertainties used here represent one standard deviation of the noise statistics and not possible systematic errors, which could be as much as 25% if errors in the transmitted power, system temperature, and telescope gain are all of order 10%. Such systematic errors should affect all results equally and the circular polarization ratio, which is de- fined to be the ratio of the power in the SC channel to that 704 G.J. Black et al. /Icarus 191 (2007) 702-711 Table 1 Observation log Target Date W. longitude W. latitude Beam separation'' Obs. time Power (UT) (deg) (deg) (arcmin) (min) (kW) Enceladus 2004 Jan 17 92 -28 0.39 30 740 2005 Jan 23 260 -25 _ 30 801 2005 Jan 24 161 -25 0.45 27 795 2005 Jan 31 196 -25 - 29 771 2005 Feb 01 98 -25 - 29 802 2005 Feb 03 262 -25 0.17 25 779 2006 Feb 08 265 -21 - 30 439 2006 Feb 09 167 -21 - 29 444 2007 Jan 24 68 -15 - 22 649 2007 Jan 26 232 -15 - 25 649 2007 Jan 27 134 -15 - 25 642 2007 Jan 28 36 -15 0.33 29 741 2007 Feb 04 71 -15 0.58 26 655 2007 Feb 05 333 -15 0.58 26 756 2007 Feb 07 135 -15 - 26 800 2007 Feb 08 39 -15 - 26 877 Tethys 2005 Jan 23 299 -23 0.21 30 801 2005 Jan 24 130 -23 - 27 795 2005 Jan 25 320 -23 - 18 782 2005 Feb 02 41 -23 - 17 790 2005 Feb 03 232 -23 - 25 779 2006 Feb 08 221 -19 0.24 30 439 2006 Feb 09 51 -19 0.59 29 444 2006 Feb 11 72 -19 0.29 29 442 2007 Jan 24 6 -15 0.57 22 649 2007 Jan 28 47 -15 0.72 29 741 2007 Feb 07 149 -15 0.07 26 800 2007 Feb 08 339 -15 0.73 26 877 Dione 2004 Jan 17 36 -26 _ 29 740 2005 Jan 24 125 -23 0.22 27 795 2005 Jan 31 324 -23 0.75 29 771 2005 Feb 01 95 -23 0.40 29 802 2006 Feb 09 186 -19 0.29 29 444 2006 Feb 11 88 -19 - 29 442 2007 Jan 24 7 -13 0.50 22 649 2007 Jan 26 270 -13 0.54 25 649 2007 Jan 27 41 -13 0.36 25 642 2007 Jan 28 172 -13 0.75 29 741 2007 Feb 07 44 -14 0.40 26 800 2007 Feb 08 176 -14 0.51 26 877 Rhea 2004 Jan 16 318 -26 _ 29 720 2004 Jan 17 37 -26 0.27 22 740 2004 Jan 18 117 -26 - 29 767 2005 Jan 25 307 -23 0.68 18 782 2005 Feb 03 303 -23 0.81 25 779 2006 Feb 08 256 -19 0.75 30 439 2006 Feb 11 134 -19 0.33 29 442 2007 Jan 24 58 -13 0.61 22 649 2007 Jan 26 218 -13 0.37 25 649 2007 Jan 28 17 -13 0.15 29 741 2007 Feb 08 171 -13 0.48 26 877 No entry indicates the primary target on that date and thus centered in the beam. Titan was the primary target on 2007 Jan 28, Feb 4, Feb 5. in the OC channel, should be largely immune to systematic errors in the gain and transmitter power in particular. Consis- tent 13 cm wavelength radar measurements of inner Solar Sys- tem targets and the icy Galilean satellites over more than three decades have demonstrated the stability of the system calibra- tion within the statistical uncertainties of those measurements. We further note the system stability shown by the results of multi-epoch observations of the stronger Saturn system targets Titan (Campbell et al., 2003) and the ring system (Nicholson et al., 2005). Thus systematic errors and absolute calibration issues should be significant only when comparing Arecibo ob- servations with those of other systems at other wavelengths. Arecibo observations of Saturn's satellites 705 Table 2 Satellite parameters Target Diameter Rotation Bandwidth (km) (h) (Hz) Rhea 1529 108.43 368 Dione 1123 65.69 451 Tethys 1066 45.31 615 Enceladus 504 32.88 400 The bandwidth is the frequency spread of the echo due to the target's rotation and is given by ? = 4;rrZ)cos?/A,P for diameter D, rotation period P, wave- length A, (= 13 cm), and subradar latitude S. These bandwidths are averages over the different latitudes in Table 1. Diameters are from Thomas et al. (2006). Rhea such as from the Cassini RADAR at 2.2 cm wavelength as dis- cussed later. Since the angular separation between satellites if often less than the width of the Arecibo beam at this wavelength, on many dates more than one satellite could be observed simulta- neously. At those times one satellite was selected as the primary target and centered in the beam with the secondary target(s) falling at some angular distance off-center as indicated in Ta- ble 1. The beam shape was assumed to be Gaussian with an average half power beam width of 1.95 arcmin (Heiles et al., 2000) and included by modifying the gains in the weighting function. Secondary targets would also drift in frequency due to their motion relative to the primary by as much as ~ 10 Hz/s, which was corrected by frequency drifting the raw sampled data 50 Hz res 270 Dione 0 1 -1 0 1 -1 0 1 kHz kHz kHz 100 Hz res 180 90 270 -2-1 0 1 2-2-1 0 1 kHz kHz Tethys 2-2-1 0 1 2 kHz 100 Hz res 270? -2-1012 -2 10 12 -2 10 12 kHz kHz kHz Enceladus 100 Hz res 270? Fig. 1. Final summed 13 cm wavelength echo spectra are shown for each target. Panels show the power received in the opposite circular sense as transmitted (OC), same circular sense (SC), and their weighted sum (TC). Short bars on the abscissas indicate the average expected bandwidth of each target due to the rotationally induced Doppler shifts of the signal (see Table 2). The frequency resolution used in each row is indicated; a smaller resolution is used for Rhea given its higher signal-to-noise ratio (SNR). The ordinale scales are standard deviations of the noise. The inward pointing lines in the last panel indicate the central west longitudes of each observation on that target, with the length of the line proportion to the weight function for each. 706 G.J. Black et al. /Icarus 191 (2007) 702-711 Table 3 Satellites' 13 cm wavelength radar properties Target Radar albedo IJ,C (SC/OC) SNR Beq/B Rhea OC 0.61 ? 0.03 sc 0.71 ?0.04 TC 1.31 ?0.05 Dione OC 0.41 ? 0.07 SC 0.32 ? 0.07 TC 0.74 ?0.10 Tethys OC 0.66 ? 0.09 SC 0.79 ? 0.09 TC 1.45 ?0.13 Enceladus (total) OC 1.07 ?0.22 SC 0.86 ? 0.20 TC 1.94 ?0.31 Enceladus (lead) OC 1.16 ?0.29 SC 1.39 ?0.29 TC 2.55 ? 0.42 Enceladus (trail) OC 1.00 ? 0.28 SC s:o.54^ TC 0.78 ? 0.28 1.17 ?0.09 0.81 ?0.21 1.22 ?0.21 0.83 ? 0.25 1.28 ?0.41 ?0.56^ 46 49 17 20 0.90 ? 0.03 0.97 ? 0.02 0.93 ? 0.09 0.73 ?0.18 0.97 ? 0.03 0.84 ? 0.08 0.78 ?0.17 0.58 ? 0.22 0.79 ? 0.20 0.83 ?0.15 0.52 ? 0.28 Error estimates are based on one standard deviation of the statistical uncertainties. ^ Three standard deviation upper limit. during processing prior to forming the power spectra. The ab- solute frequency difference between targets could be as large as ~100 kHz. Initially a standard frequency hopping technique (cf. Ostro et al., 1992) was used in 2004 where the transmitter frequency alternates at 10 s intervals between four frequen- cies separated by 10 kHz. This scheme is employed to provide signal-free background samples at the three frequencies not cur- rently used during any given hop period. This was found to be an extra complication when extracting multiple targets and so subsequent sessions used a single transmit frequency. The background at the target's frequency was then obtained by in- terpolating the smoothly varying bandpass across the echo's narrow bandwidth. 3. Discussion The final spectra shown in Fig. 1 demonstrate the detection quality of each targeted moon. Generally their radar albedos are high with roughly comparable power returned in both circular polarizations, as listed in Table 3. Such properties strongly in- dicate a subsurface multiple scattering mechanism, also known as volume scattering, but perhaps operating to somewhat dif- ferent extents on each moon. Single scattering from the sur- face would return the majority of power in the OC channel. Single scattering dominates echoes from inner Solar System targets; for example, the Moon at 13 cm wavelength has a total albedo of ~0.07 and a polarization ratio of only ~0.1 (Evans and Hagfors, 1966), and what little power is returned in the SC polarization sense is due to wavelength scale surface and near-surface roughness. Some asteroids reflect comparable power in both polarizations with little specular reflection, how- ever their total albedos remain low (e.g., Magri et al., 1999; Benner et al., 1997), a combination which is usually interpreted to result from considerable surface and near surface wavelength scale roughness. To return significant power in the SC channel would require a more efficient multiple scattering mechanism. The shape of the echo spectrum gives an indication of the scattering mechanism. If single surface scattering dominates, the echo from a spherical body will be strongest at the sub- radar point where the incidence angle is near vertical producing a strong specular reflection in the center of the echo spectrum. Multiple scattering mechanisms can broaden the spectrum by returning significant power at higher incident angles. This dis- tribution of power within the echo spectrum can be expressed as an equivalent bandwidth, defined in the sense used in Ostro et al. (1992), 5, ?eq ? A/ Eyf (3) where y? are the spectrum amplitudes and A/ is the frequency resolution. The broad widths (see Table 3) of the echoes further indicate a multiple scattering mechanism. The mid-size satellites of Saturn are generically similar to the icy Galilean satellites of Jupiter in that all have low- temperature, water ice dominated surfaces exposed to the vac- uum. Thus one might expect the regolith structure of these sat- urnian moons to be similar to that of the icy Galilean satellites given the dominance of water ice, not too dissimilar cratering rates (Zahnle et al., 2003), and ancient surfaces. The latter point may not hold for Enceladus however in light of recent Cassini observations of current surface activity near its south pole (e.g., Porco et al., 2006), but most of these moons' surfaces have es- timated ages of order 2 to 3 Gyr (Plescia and Boyce, 1985). The extremely low absorption of centimeter wavelength radia- Arecibo observations of Saturn's satellites 101 0.0 0.5 1.0 1.5 Geometric Optical Albedo Fig. 2. Comparison of icy satellite 13 cm wavelength radar albedos versus their visual wavelength geometric albedos. For both systems the rough correlation between optical and radar albedos suggests a commonality between the primary darkening agent in both wavelength regimes. Radar error bars represent one standard deviation of the statistical uncertainties, optical error bars represent the range of variation with longitude. Jupiter satellite radar albedos are from Ostro et al. (1992); lapetus radar albedo from Black et al. (2004). Optical albedos for the Saturn system are from Verbiscer et al. (2007); for the Jupiter system from Domingue and Verbiscer (1997). tion in clean water ice will permit penetration of the radiation to much greater depths than on non-ice targets. Regolith devel- opment will cause density and hence refractive discontinuities, creating a population of scattering centers within these low-loss layers. When observed with a radar, these characteristics of the icy Galilean satellites lead to a condition where they become very effective backscatterers. At 13 cm wavelength Europa's total radar albedo is ~2.6 and its circular polarization ratio 1.5 (Ostro et al., 1992), compared to non-icy, inner Solar Sys- tem targets which have radar albedos and polarizations ratios an order of magnitude lower. Furthermore, for the icy Galilean satellites the lack of any specular echoes is also consistent with a volume scattering mechanism. If the saturnian satellites have similar regolith properties then their radar albedos and polariza- tion ratios should also be large. Non-Galilean satellite scattering properties would indicate a different surface structure and/or composition, having implications for the current and past con- ditions in the Saturn system in either case. The total 13 cm wavelength radar albedos are plotted ver- sus optical geometric albedos in Fig. 2, along with the cor- responding values for the icy Galilean satellites. For Europa, Ganymede, and Callisto, there is a good correlation between these two quantities, implying a certain amount of homogeneity between the optical surface and the depths probed by the radar, i.e., several tens of wavelengths. The inter-satellite trend fol- lows the overall icy purity inferred for these moons from their optical albedos and depths of infrared water bands, with Europa having the lowest fraction of non-ice contaminants, Callisto the highest, and Ganymede intermediate (cf. review by Clark et al., 1986). The Saturn satellites have optical albedos similar to the jovian satellites as well as similar water ice band depths in the infrared (Clark et al., 1984), but follow a different trendline in Fig. 2. Thus the general correlation between optical and radar albedos for both satellite systems can be explained with a vary- ing fraction of an optical and radar darkening compound, such as meteoritic material. The shift in the trend for the saturnian satellites suggests their surfaces have an additional radar ab- sorber that reduces contributions from volume scattering but without affecting the optical albedos, thereby depressing the radar reflectivity for a given optical reflectivity. Alternatively, a thin mantling of clean, optically bright ice could increase the optical reflectivity without affecting the radar reflectivity. A coating of small, fresh grains may easily explain Enceladus' properties, but it may be more difficult to cover the other satel- lites at the optical wavelength scale by such small grains with- out giving them all Enceladus-like optical properties. Several classes of compounds can affect the radar scatter- ing by increasing the absorption in a water ice layer. Silicates, organics, or ammonia are plausibly present in the surfaces of these objects. While these new data themselves do not strongly distinguish between possible absorbers, ammonia compounds are likely candidates based on availability, formation models, and comparison to other data obtained at other wavelengths. Ammonia is expected to be more prevalent in the satellites of Saturn than in those of the Jupiter system due to the cooler tem- peratures in the saturnian nebula (Lewis, 1972), and has been invoked to explain the origin of Titan's nitrogen atmosphere via destruction of its originally accreted ammonia (Atreya et al., 1978; Owen, 2000). In addition, the presence of large quantities of ammonia in the moons' water ice mantles would signifi- cantly lower the melting points and maintain liquid or ductile interiors at lower temperatures than is possible in a completely water ice body. Lowering of the melting point in these objects is one means of explaining the evidence of more active geology than would otherwise be expected on these small, cold objects (Squyres et al., 1983). Ammonia in the near-surface should af- fect the radar reflectivity since it can increase the absorption of centimeter-wavelength radiation considerably (Lorenz and Shandera, 2001) and be less optically active than silicates or organics. To date there has been no robust detection of any sig- nificant quantity of ammonia on these objects. Infrared spectra are dominated by strong water features, with other compounds detected or constrained to be present only in trace amounts. For Enceladus, Emery et al. (2005) report a possible detection of ammonia at the level of 0.5% by weight, while Cassini mea- surements place an upper limit of 2% on its global abundance (Brown et al., 2006). Even the plumes from Enceladus' south polar terrain are at least 90% H2O with only trace quantities of ammonia (Waite et al., 2006). Fig. 3 plots the multi-wavelength radar properties of the four icy saturnian satellites discussed here together with those of the trailing hemisphere of lapetus (Black et al., 2004; Ostro et al., 2006) and the icy Galilean satellites (Ostro et al., 1992; Black et al., 2001a) for comparison. Any comparisons drawn from Fig. 3 depend on the absolute calibration of the different 708 G.J. Black et al. /Icarus 191 (2007) 702-711 O ?D 03 < c5 ?D (0 DC B o o ?D ?o (t? DC "ct? -I?? o Encelad JS Tethys Dione Rhea lapetus(T) 3.0 ' 2.5 2.0 1.5 ?t?? -^i? ^ -^i- ? 1.0 0.5 0.0 .-e-. + ; .& 0.0 0.5 1.0 1.5 Europa Ganymede Callisto 3.0 2.5 2.0 1.5 1.0 0.5 0.0 ? A 2.2 cm: Ostro et al. (2006) o 3.5 cm: Ostro et al. (1992) D 13 cm upper panel: this work except lapetus from Black et al. (2004) D 13 cm lower panel: Ostro et al. (1992) O 70 cm: Black et al. (2001a) 0.0 0.5 1.0 1.5 Fig. 3. Plots of total radar albedo versus circular polarization ratio for each target. Panels with data for lapetus' trailing (icy) hemisphere and for the icy Galilean satellites are included for comparison. Each point includes data from all available longitudes except lapetus. The abscissa of each panel is the same as the first panel's. Saturn satellite data at 13 cm wavelength are plotted with the quadrature sum of statistical uncertainties and a 25% systematic uncertainty. At 2.2 cm wavelength the error bars further include assumptions in the ranges of polarization ratios (see text and Ostro et al., 2006). Error bars for the Galilean satellites are the larger of the statistical uncertainties or the range of any variation with longitude. instruments, and so for these new data the radar albedo error bars in that figure represent the quadrature sum of the statistical uncertainties and a 25% systematic error Furthermore, while we have directly measured the radar properties in both circular polarizations at 13 cm wavelength, the values at 2.2 cm wave- length from Cassini were obtained in a single linear polarization only. For those measurements, Ostro et al. (2006) assumed po- larization behavior similar to that of the icy Galilean satellites in order to convert the single linear polarization into a total radar albedo. Barring other information, this appears to be justified given the already high albedos observed in that one linear polar- ization alone. Finally, the total radar albedo would then be the same whether measured in linear or circular polarization. For the Cassini points in Fig. 3 we have incorporated the same as- sumptions for linear and circular polarizations from Ostro et al. (2006) into the error bars on the Cassini measurements, namely that the linear polarization ratio is in the range [0.0, 0.7] and the circular polarization ratio is in the range [1.1, 1.6]. That latter range is inconsistent with these 13 cm observations, but taking the circular polarization ratios at 2.2 cm wavelength to be equal to those 13 cm wavelength, as observed for the Galilean satel- lites, does not align the radar albedos at those two wavelengths in Fig. 3. There also does not seem to be any single systematic correction factor as would result from an absolute calibration difference that better aligns the two radar dataseis for all sat- urnian satellites. The new measurements presented here overall show lower total albedos and lower polarization ratios than those of the icy Galilean satellites, and, keeping in mind the assumptions in the previous paragraph, may show a greater disparity between wavelengths for the saturnian targets. The icy Galilean satellites show nearly identical properties at both 13 and 4 cm wavelength scales, with significant reduction in radar albedo evident only at the much longer wavelength of 70 cm. The saturnian satellites observed here, as well as lapetus, instead may show real differ- ences between the shorter two wavelengths. While the apparent drops in saturnian satellites radar albe- dos between 2.2 and 13 cm wavelengths somewhat mirrors the drops between 13 and 70 cm wavelengths on the icy Galilean satellites, the polarization ratios may also drop for the former set. The icy Galilean satellites' albedo drop from 13 to 70 cm wavelength has been interpreted as a reduction in the number of scatterers at the larger wavelength (Black et al., 2001b). The drop between wavelengths observed on lapetus was suggested by Ostro et al. (2006) to result from a relatively shallow scatter- ing layer. These new data may further support their suggestion that the effective scattering layers on these moons are deter- mined by an increasing amount of absorber with depth, such that relatively cleaner layers are seen by the 2.2 cm wavelength, while the 13 cm wavelength senses lower, more absorbing lay- ers. Any drop in polarization ratio from 2.2 to 13 cm wavelength would likewise be due to increasing absorption reducing the efficiency of volume scattering as longer scattering paths that preserve the incident polarization sense are the paths that suffer relatively more attenuation. The result of any added absorption Arecibo observations of Saturn's satellites 709 at 13 cm wavelength is thus to reduce both the radar albedo and the polarization ratio. Conversely, the polarization ratios of the icy Galilean satellites may remain high at all wavelengths because their albedo drop with wavelength is due instead to a lack of scatterers available at the largest scales (Black et al., 2001b), effectively diminishing the number of scattering paths without preferentially selecting against longer ones. 3.1. Rhea These results confirm the preliminary analysis of a subset of these Rhea data by Black and Campbell (2004). In terms of radar properties, Rhea is a close analogue to Ganymede at both the 13 cm wavelength and the 2.2 cm Cassini wavelength. With a closer optical albedo match to Europa, this suggests that the radar senses a Ganymede-like subsurface but with an op- tically bright coating thin on the radar wavelength scale. Also like Ganymede, there is little difference between Rhea's proper- ties at both radar wavelengths, implying that the surface appears similar on both wavelength scales. Such a homogeneity may be consistent with Rhea having the oldest and thus most well- mixed regolith as inferred from cratering statistics (Plescia and Boyce, 1985). We have also split the data set into leading (0? < west lon- gitude < 180?) and trailing (180? < west longitude < 360?) hemispheres. The Cassini RADAR observed only the leading hemisphere of Rhea (Ostro et al., 2006). The ratios of the hemi- spherically separate 13 cm wavelength radar albedos versus the similar ratio of hemispheric optical albedos are plotted in Fig. 4. The leading side is significantly more re?ective than the trailing side at both optical (Verbiscer et al., 2007) and 13 cm radar wavelengths, indicating that perhaps the same material or process darkens the surface in both wavelength regimes. 3.2. Tethys Tethys has 13 cm wavelength radar properties very similar to Rhea and Ganymede, but possibly different values than at the Cassini 2.2 cm wavelength. A decrease in albedo and polar- ization ratio may signal a decrease in volume scattering via an increased absorption seen at depth by the longer wavelength. If real, this difference may constrain a gradient in composition or structure with depth. The breakdown of the Tethys dataset into leading and trailing hemispheres is also shown in Fig. 4. No significant asymmetry in radar albedo is seen. The Cassini RADAR observed only the trailing hemisphere of Tethys (Ostro et al., 2006). 3.3. Dione Dione's albedos at both radar wavelengths are lower than the corresponding values for Tethys and with possibly a larger wavelength dependence. One way to achieve this is for Dione's surface layers to contain a higher fraction of radar absorbing material, lowering the overall radar albedo and decreasing the effectiveness of any volume scattering. A decrease in the rela- tive contribution of volume scattering over single surface scat- o "a (D n (0 to tr 3 : 5, 0.0 0.5 1.0 1.5 (LTT) Optical Albedo 2.0 Fig. 4. Comparison between the ratios of leading to trailing hemisphere albedos at 13 cm radar wavelength (this work) versus at optical wavelengths (Verbiscer et al., 2007). The radar albedo points are shown with statistical uncertainties as systematic errors should factor out in the ratio. Rhea and Enceladus appear to have significantly higher radar albedos on their leading sides, in phase with the optical asymmetry for Rhea and out of phase for Enceladus. Dione and lapetus show no significant hemispheric difference in 13 cm wavelength radar albedo despite large optical asymmetries. tering would have the effect of lowering the polarization ratio as seems to be the case at 13 cm wavelength. At 2.2 cm wave- length Dione's radar albedo is similar to Rhea's, but if Dione's albedo drops by 13 cm wavelength it may suggest a steeper gradient in composition or structure with depth than Rhea, caus- ing a quicker decrease in volume scattering efficiency at 13 cm wavelength on Dione. Like Tethys, there does not appear to be a significant dif- ference in radar properties between leading and trailing hemi- spheres within the statistical uncertainties, despite its quite large optical asymmetry (Verbiscer et al., 2007). Our observations are much more concentrated on Dione's leading side and very sparse on it's trailing side however. Only the leading hemi- sphere of Dione has been observed by the Cassini RADAR (Ostro et al., 2006). 3.4. Enceladus Enceladus' total radar albedo is the highest of these moons, although its 13 cm wavelength polarization ratio is relatively low. The effective backscattering needed for a high albedo can- not rely on single scattering, and hence the required multiple scattering should also result in a similarly very high polariza- tion ratio. The radar albedos at both radar wavelengths appear essentially identical, although the polarization ratios may not be. If real, these results may be more difficult to explain than the wavelength variations seen on the other satellites. Unfortu- nately the Enceladus detection does not have the high signal-to- noise ratio one would like to robustly draw further conclusions. As shown in Figs. 4 and 5, splitting the Enceladus data into leading and trailing hemispheres indicates that most of the re- ceived radar signal apparently comes from the leading hemi- 710 G.J. Black et al. /Icarus 191 (2007) 702-711 Enceladus -total -10 1 -10 1 kHz kHz Enceladus - leading -1 0 1 kHz -10 1 -10 1 kHz kHz Enceladus -trailing -1 0 1 kHz . 1 oc. 1 . \Ji SC TC I.I . . I.I . 180 90-( ^270 -1 0 1 kHz -1 0 1 kHz -1 0 1 kHz Fig. 5. Enceladus 13 cm wavelength echo spectra split into leading and trailing hemisphere data sets. The top panel reproduces Fig. 1 and contains data from all longitudes. See Fig. 1 for description of axes and clock diagrams. Most of the Enceladus echo appears to originate from the leading hemisphere. There is no detection of the trailing hemisphere in the SC polarization, suggesting a very low polarization ratio for an icy surface. The overall sensitivity on the leading side is only about 15% better than on the trailing side due mainly to the larger number of observing sessions on the former. sphere, indicating a substantial radar asymmetry. These hemi- spherically separate results are also included in Table 3. Com- pared to the slight optical hemispheric asymmetry (Verbiscer et al., 2007), the radar asymmetry supports the model of Ence- ladus having a thin mantling of fresh, clean ice particles that is transparent to the radar which instead senses deeper, older, and perhaps a more original surface. The lack of a detection of the trailing hemisphere in the SC polarization channel sug- gests a surprisingly low polarization ratio for an icy surface. As a result of the geometry, latitudes southward of ~70? S were in continuous view during our observations, including most of the south polar terrain (Porco et al., 2006), which cannot then be used to explain a hemispheric difference seen here. If other processes besides impact gardening have been re- cently modifying Enceladus' surface, its structure may be dra- matically different than the older, mature regolith of the other satellites. A combination of high radar albedo and low polariza- tion ratio may indicate unique surface properties. On the Earth, the melt zones of the Greenland ice sheet have this combination of radar properties, apparently due to subsurface ice structures that result from seasonal melting and refreezing cycles (Rignot et al., 1993; Jezek et al., 1994), as do some high altitude glaciers in the Andes and Tibet (Haldemann and Muhleman, 1999). Al- though liquid water content controls the radar reflectivity of Earth's ice surfaces, those studies generically suggest that the radar scattering from Enceladus may be controlled by structures not found on the other icy satellites. Acknowledgments G.J.B. acknowledges support from the NASA Planetary Astronomy Program and D.B.C. acknowledges support from the NASA Planetary Geology and Geophysics Program. The Arecibo Observatory is part of the National Astronomy and Ionosphere Center, which is operated by Cornell University under a cooperative agreement with the National Science Foun- dation. References Atreya, S.K., Donahue, T.M., Kuhn, W.R., 1978. Evolution of a nitrogen at- mosphere on Titan. Science 201, 611-613. Baron, J.E., Tyler, G.L., Simpson, R.A., 2003. Three-dimensional numerical modeling of refraction and reflection scattering from icy Galilean satellites. Icarus 164, 404-417. Benner, L.A.M., and 12 colleagues, 1997. Radar detection of near-Earth As- teroids 2062 Aten, 2101 Adonis, 3103 Eger, 4544 Xanthus, and 1992 QN. Icarus 130,296-312. Black, G., Campbell, D., 2004. Rhea's surface: Ice properties measured by radar. Bull. Am. Astron. Soc. 36. Abstract 1123. Arecibo observations of Saturn's satellites 711 Black, GJ., Campbell, D.B., Nicholson, P.D., 2001b. Icy Galilean satellites: Modeling radar reflectivities as a coherent backscatter effect. Icarus 151, 167-180. Black, G.I., Campbell, D.B., Ostro, S.I., 2001a. Icy Galilean satellites: 70 cm radar results from Arecibo. Icarus 151, 160-166. Black, G.I., Campbell, D.B., Carter, L.M., Ostro, S.I., 2004. Radar detection of lapetus. Science 304, 553. Brown, R.H., and 24 colleagues, 2006. Composition and physical properties of Enceladus' surface. Science 311, 1425-1428. Buratti, B., Veverka, I., 1984. Voyager photometry of Rhea, Dione, Tethys, Enceladus and Mimas. Icarus 58, 254-264. Buratti, B.I., Mosher, I.A., lohnson, T.V., 1990. Albedo and color maps of the satumian satellites. Icarus 87, 339-357. Campbell, D.B., Chandler, I.E, Ostro, S.I., Pettengill, G.H., Shapiro, I.I., 1978. Galilean satellites?1976 radar results. Icarus 34, 254-267. Campbell, D.B., Black, G.I., Carter, L.M., Ostro, S.I., 2003. Radar evidence for liquid surfaces on Titan. Science 302, 431-434. Clark, R.N., Eanale, E.P., Gaffey, M.I., 1986. Surface composition of natural satellites. In: lAU CoUoq. 77: Some Background about Satellites, pp. 437- 491. Clark, R.N., Steele, A., Brown, R.H., Owensby, P.D., 1984. Saturn's satellites? Near-infrared spectrophotometry (0.65-2.5 microns) of the leading and trailing sides and compositional implications. Icarus 58, 265-281. de Pater, I., Dickel, I.R., 1991. Multi-frequency radio observations of Saturn at ring inclination angles between 5 and 26 degrees. Icarus 94, 474-492. Domingue, D., Verbiscer, A., 1997. Re-analysis of the solar phase curves of the icy Galilean satellites. Icarus 128, 49-74. Doumeau, G., Baratchart, S., 1999. Masses and densities of the four inner major Saturn's satellites. Astron. Astrophys. 350, 680-684. Emery, I.P, Burr, D.M., Cruikshank, D.P, Brown, R.H., Dalton, I.B., 2005. Near-infrared (0.8-4.0 pm) spectroscopy of Mimas, Enceladus, Tethys, and Rhea. Astron. Astrophys. 435, 353-362. Eshleman, V.R., 1986. Radar glory from buried craters on icy moons. Sci- ence 234, 587-590. Evans, I.V., Hagfors, T, 1966. Study of echoes from the Moon at 23 centimeters wavelengfli. I. Geophys. Res. 71, 4871^889. Goldstein, R.M., Green, R.R., 1980. Ganymede: Radar surface characteristics. Science 207, 179-180. Gurrola, E.M., 1985. Interpretation of radar data from the icy Galilean satellites and Triton. Thesis, Stanford University. Hagfors, T., Dahlstrom, I., Gold, T., Hamran, S.-E., Hansen, R., 1997. Refrac- tion scattering in the anomalous reflections from icy surfaces. Icarus 130, 313-322. Haldemann, A.EC, Muhleman, D.O., 1999. Circular-polarization radar prop- erties of high-altitude ice: Western Kunlun Shan and central Andes. I. Geo- phys. Res. 104, 24075-24094. Hapke, B., 1990. Coherent backscatter and the radar characteristics of outer planet satellites. Icarus 88, 407-417. Heiles, C, Perillat, P, Nolan, M., Lorimer, D., Bhat, R., Ghosh, T., Howell, E., Lewis, M., O'Neil, K., Salter, C, Stanimirovic, S., 2000. Main beam and first sidelobe parameters for Arecibo's receiver systems. Technical Report ATOMS 2000-4. NAIC, Arecibo, PR. lezek, K.C., Gogineni, P., Shanableh, M., 1994. Radar measurements of melt zones on the Greenland Ice Sheet. Geophys. Res. Lett. 21, 33-36. Lewis, I.S., 1972. Low temperature condensation from the solar nebula. Icarus 16, 241-252. Lorenz, R.D., Shandera, S.E., 2001. Physical properties of ammonia-rich ice: Application to Titan. Geophys. Res. Lett. 28 (2), 215-218. Magri, C, Ostro, S.I., Rosema, K.D., Thomas, M.L., Mitchell, D.L., Camp- bell, D.B., Chandler, I.E., Shapiro, I.I., Giorgini, I.D., Yeomans, D.K., 1999. Main belt asteroids: Results of Arecibo and Goldstone radar observations of 37 objects during 1980-1995. Icarus 140, 379-407. Mosqueira, I., Estrada, PR., 2003. Eormation of the regular satellites of giant planets in an extended gaseous nebula. I. Subnebula model and accretion of satellites. Icarus 163, 198-231. Nicholson, PD., Erench, R.G., Campbell, D.B., Margot, I.-L., Nolan, M.C., Black, G.I., Salo, H.I., 2005. Radar imaging of Saturn's rings. Icarus 177, 32-62. Ostro, S.I., 1993. Planetary radar astronomy. Rev. Mod. Phys. 65, 1235-1280. Ostro, S.I., Campbell, D.B., Simpson, R.A., Hudson, R.S., Chandler, I.E., Rosema, K.D., Shapiro, I.I., Standish, E.M., Winkler, R., Yeomans, D.K., 1992. Europa, Ganymede, and Callisto?New radar results from Arecibo and Goldstone. I. Geophys. Res. 97, 18227-18244. Ostro, S.I., and 19 colleagues, 2006. Cassini Radar observations of Enceladus, Tethys, Dione, Rhea, lapetus, Hyperion, and Phoebe. Icarus 183, 479-490. Owen, T.C., 2000. On the origin of Titan's atmosphere. Planet. Space Sei. 48, 747-752. Peters, K., 1992. The coherent backscatter effect: A vector formulation account- ing for polarization and absorption effects and small or large scatterers. Phys. Rev. B 46, 801-812. Plescia, I.B., Boyce, I.M., 1985. Impact cratering history of the saturnian satel- lites. I. Geophys. Res. 90, 2029-2037. Porco, C.C., and 24 colleagues, 2006. Cassini observes the active south pole of Enceladus. Science 311, 1393-1401. Rignot, E.I., Ostro, S.I., van Zyl, i.J., lezek, K.C., 1993. Unusual radar echoes from the Greenland ice sheet. Science 261, 1710-1713. Squyres, S.W., Reynolds, R.T., Cassen, P.M., 1983. The evolution of Enceladus. Icarus 53, 319-331. Thomas, P.C., Veverka, I., Helfenstein, P., Porco, C, Burns, I., Denk, T., Turtle, E., lacobson, R.A., the ISS Science Team, 2006. Shapes of the satumian icy satellites. Lunar Planet. Sei. XXXVII, 1639-1640. Abstract. Thompson, W.R., Squyres, S.W., 1990. Titan and other icy satellites: Dielec- tric properties of constituent materials and implications for radar sounding. Icarus 86, 336-354. Verbiscer, A., Erench, R., Showalter, M., Helfenstein, P., 2007. Enceladus: Cos- mic graffiti artist caught in the act. Science 315, 815. Wagner, R., Roatsch, T, Giese, B., Wolf, U., Neukum, G., 2006. Geologic map- ping of the saturnian satellites based on Cassini ISS images: Objectives, methods, and results. In: European Planetary Science Congress, p. 594. Waite, I.H., and 13 colleagues, 2006. Cassini Ion and Neutral Mass Spectrom- eter: Enceladus plume composition and structure. Science 311, 1419-1422. Zahnle, K., Schenk, P., Levison, H., Dones, L., 2003. Cratering rates in the outer Solar System. Icarus 163, 263-289.