You are on page 1of 11

97.

1485C

THE ASTRONOMICAL JOURNAL VOLUME 97, NUMBER 5 MAY 1989

GRAVITY FIELD OF THE SATURNIAN SYSTEM FROM PIONEER AND VOYAGER TRACKING DATA
1989AJ

J. K. Campbell and J. D. Anderson


Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California 91109
Received 30 September 1988; revised 7 January 1989

ABSTRACT

Analysis of the Doppler-tracking data and star-satellite imaging from the Voyager 1 and Voyager 2 spacecraft,
combined with Pioneer 11 Doppler tracking, has yielded improved values for the masses of Rhea, Titan, and
lapetus, and the mass and harmonic coefficients of Saturn. The present results for satellite masses and the mass
of Saturn are consistent with those previously published for Pioneer 11 (Null et al. 1981); the results for the
harmonic coefficients J2) and J6 are consistent with the Pioneer 11 results and with those recently published
by Nicholson and Porco ( 1988), which combined Pioneer 11 results with results from Voyager observations of
an eccentric ringlet (Porco et aL 1984). Our adopted solution was obtained from the combined dataset of
Voyager Doppler and satellite imaging data, and the Nicholson and Porco results, which gave an improved
solution for J6. In units of (km3 s-2), the satellite masses are (45 + 10) for Tethys, (154 + 4) for Rhea,
(8978.2 + 1) for Titan, and (106 + 10) for lapetus (all error estimates adopted in this paper are standard
errors and are our evaluation of real as opposed to formal errors). The value for Tethys mass is less accurate
but consistent with that determined from analysis of ground-based observations (Kozai 1976). Mass determi-
nations for Mimas, Enceladus, Dione, Hyperion, and Phoebe could not be obtained. A mass for the small
satellite Epimetheus could not be estimated reliably from Doppler tracking collected during the Pioneer 11 close
flyby (2680 km) because of high data noise attributable to interplanetary plasma along the propagation path.
The determined of the Saturn system is (37 940 630 + 200). The ratio of the mass of the Sun to the mass
of the Saturn system is 3497.898 + 0.018. The GL^ of Saturn is (37 931 272 + 200). Using
G = (6.6728 + 0.0016) X10-23 km3 s-2g_1, we obtain a mass for Saturn of (5.6845 + 0.0014) X1029 g. The
second, fourth, and sixth zonal harmonic coefficients of Saturn are J2 = ( 16298 + 10) X 10~6,
/4 = ( 915 + 40) X 10~6, and J6 = ( 103 + 50) X10-6 at an equatorial reference radius of 60 330 km, and for
an assumed eighth-harmonic value of /8 = lOX 10-6. All other gravitational harmonics were assumed to be
zero, except for C22 and S22, which were estimated. The Pioneer and Voyager data provide a possible detection
of a second-degree longitudinal harmonic with amplitude J22<2X 10~6. The differences in the equatorial
principal moments of inertia are limited to (B A)/^R2 <%Xl0~6. The data yielded a linear constraint
between J2, and J6 that is somewhat more accurate than the individual determination of J4 and J6. It is
J6 - 0.3635(/2 - 16298X10"6) - 0.8809(/4 + 915X106) = ( 103 + 16) X 10"6. The Saturn rotational pole
was fixed at the orientation determined by Simpson (1983) from knowledge of the Voyager trajectories and
occultation measurements of Saturns rings using the Voyager spacecraft. All solutions for masses and harmonic
coefficients were insensitive to small gravitational effects due to the Saturn ring mass, the tides raised on Saturn
by the larger satellites, and the second-order gravitational field of Titan.

I. INTRODUCTION tion, Effects of Small Components of the Gravitational


Field, Results and Discussions, and Conclusions. Much of
The flights of the Voyager spacecraft through the Satur- the general discussion in these sections closely parallels that
nian system in 1980 and 1981 provided an opportunity to given in Campbell and Synnott ( 1985 ) on the determination
improve the knowledge of the gravity field of the planet and of Jupiters gravity field from Pioneer and Voyager data.
satellites beyond that determined from the Pioneer 11 en-
counter (Null etal. 1981). Pioneer 11 flew roughly twice as II. TRAJECTORIES
close to Saturn as did the Voyagers, but its signal path passed
very nearby the Sun at encounter (29R0 ), introducing sig- The distance and times of closest approach of the Pioneer
nificant plasma noise into the S band signal. The Voyager 1 and Voyager spacecraft to Saturn and its satellites are given
tracking signal passed through several solar shock waves on in Table I, with the positions of the major satellites taken
the encounter day; which added noticeably to the Doppler from ephemerides derived from analysis of Voyager images
noise level; Voyager 2 had no problems with solar plasma, (Campbell et al. 1982). The positions of Epimetheus, Hype-
but the spacecraft was tracked in a lower-accuracy nonco- rion, and Phoebe were provided by R. Jacobson of JPL (pri-
herent mode for portions of its encounter period. However, vate communication ), also from analysis of Voyager images.
the Voyager spacecraft had much closer approaches to the The spacecraft paths through the Saturnian system are
major satellites than Pioneer 11, which resulted in improved shown in the plane of the hyperbolic trajectory in Fig. 1 for
gravity results overall. Pioneer 11, and in Figs. 2 and 3 for Voyager 1 and Voyager 2,
Detailed solution results obtained from the combined Pio- respectively. Figure 4 shows in meridional projection the
neer and Voyager datasets, along with the solution for each trajectories for the three spacecraft, inbound to and out-
mission, are given in Sec. VI. The paper is divided into the bound from Saturn. Pioneer 11 had the closest flyby to Sat-
following six sections: Trajectories, Data, Method of Solu- urn ( 81 00 km ) and to any satellite ( Epimetheus, 2680 km ),

1485 Astron. J. 97 (5), May 1989 0004-6256/89/051485-11S00.90 1989 Am. Astron. Soc. 1485

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1486 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1486

Table I. Close approaches to Saturn and satellites.


1989AJ

Pioneer 11 Voyager 1 Voyager 2


Object a -1 0
d(l(m) v(km s )^ tca^) l
ca ca
Saturn 81.0 32 184.0 25 E 161.0 24 E
Epimetheus 2.7 17 E-1.7 121.0 22 E+1.2 147.0 24 E+0.7
Mimas 104.0 17 E-0.1 88.0 17 E+2.0 310.0 34 E-0.8
Enceladus 222.0 30 E+2.0 202.0 27 E+2.1 87.0 14 E+0.4
Tethys 330.0 32 E+1.9 416.0 34 E-1.5 93.0 15 E+2.8
Dione 291.0 20 E-0.5 162.0 15 E+3.9 502.0 30 E-2.3
Rhea 346.0 9 E+6.0 74.0 18 E+6.6 645.0 27 E+3.1
Titan 363.0 10 E+25.5 6.5 17 E-18.1 666.0 16 E-17.8
Hyperion 676.0 9 E-28.0 880.0 20 E+17.0 471.0 12 E-26.0
Japetus 1,033.0 10 E82.5 2,477.0 15 E+43.5 909.0 11 E-74.0
Phoebe 13,716.0 11 E-76.5 15,054.0 18 E-19.3 2,075.0 12 E+238.0

a
d is distance at closest approach
b
v is velocity at closest approach
c
Pioneer 11 periapsis occurred on 1 September 1979 at 16h30m34SGMT, the Voyager 1 periapsis on 12 November 1980
ar23h45rn33S GMT, the Voyager 2 periapsis on 26 August 1981 at 03h24m05S GMT
^ E is time of Saturn closest approach

but high data noise induced by solar plasmas limited the data from Pioneer and Voyager obtained by the JPL/NASA
mass-determination accuracies for these bodies. Otherwise, Deep Space Network (DSN) and Foyger-based optical
the Voyagers had the closest satellite flybys, and provided the measurements obtained with the imaging system. The radio
best opportunity to determine their masses. data measured the range and range rate of the spacecraft as
III. DATA viewed from tracking stations on the surface of the Earth.
The optical data were primarily used to refine the satellite
Two fundamentally different types of data were processed orbits, and consisted of cross line-of-sight measurements of
together in the present analysis: Earth-based radio-tracking the satellites as viewed from the spacecraft. These two com-

Fig. 1. Pioneer 11 path through the Saturn satellite system, shown in the Fig. 2. Voyager 1 path through the Saturn satellite system, shown in the
plane of the spacecraft trajectory, beginning and ending 96 hr from Saturn plane of the spacecraft trajectory, beginning and ending 72 hr from Saturn
closest approach. The projected satellite orbits are shown along with the closest approach. The projected satellite orbits are shown along with the
position at the times (labeled) when Pioneer 11 was closest to them. Time positions at the times (labeled) when Voyager 1 was closest to them. Time
ticks mark the Pioneer 11 positions at 24 hr intervals. ticks mark the Voyager 1 positions at 12 hr intervals.

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1487 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1487

plementary data types were acquired essentially over the last


30 days of the Voyager approaches. High phase angles,
1989AJ

which produce partially illuminated satellites, decreased the


utility of optical measurements after the Fqyager flybys; thus
the post-encounter data consisted of Earth-based radio
tracking only.
Coherent two-way and noncoherent one-way Doppler-
tracking data were received at the DSN tracking stations
located in Goldstone, California; Madrid, Spain; and Can-
berra, Australia. For the present analysis, nearly continuous
data from 10 days before to 5 days after Saturn closest ap-
proach were processed for each mission, with the exception
of a 2 hr outage near the closest approaches due to occulta-
tion of the Pioneer 11 radio signal by Saturn, 4.5 hr outages
due to the Voyager 1 and Voyager! occultations, and outages
of up to 12 hr during the Voyager 2 encounter period, when
only noncoherent Doppler data were generated because of
operational constraints necessitated by the Voyager 2 radio
system.
All Pioneer and Voyager Doppler data were corrected for
the systematic effects of the Earths troposphere (Chao
1971) and ionosphere (Green et al. 1980), as described by
Campbell and Synnott (1985). For all Voyager data except a
Fig. 3. Voyager 2 path through the Satura satellite system, shown in the 12 hr span of Voyager 1 data near the Saturn encounter, the
plane of the spacecraft trajectory, beginning and ending 72 hr from Saturn Doppler data at 13 and 3.6 cm wavelengths (Sand Xbands,
closest approach. The projected satellite orbits are shown along with the respectively) were used to calculate corrections caused by
positions at the times (labeled) when Voyager2 was closest to them. Time dispersive interplanetary plasma. This was not possible for
ticks mark the Voyager 2 positions at 12 hr intervals. Pioneer dzidi, which consisted of Doppler tracking at 13 cm

Fig. 4. Paths of Pioneer 11, Voyager 1, and Voyager 2 shown in meridional projection, with the orbits of the satellites indicated by first letter of satellite name.
Paths are time tagged with hours from Saturn closest approach.

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1488 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1488

wavelength only. The standard error of the coherent strengthening the confidence in the solutions for Saturns
Doppler for Voyager 2 was better than 5 mHz for 60 s inte- gravity coefficients.
1989AJ

gration times, or equivalently the one-way range-rate error


was 0.3 mm s_ 1 at a 60 s sample rate. The typical noise level IV. METHOD OF SOLUTION
for noncoherent Doppler used in the Voyager analyses was
15 mHz, or 2 nuns1. The Voyager 1 coherent Doppler The gravity solutions were obtained by simultaneously fit-
noise was a factor of 3 larger than for Voyager!; the Pioneer ting the radio range and Doppler measurements, and the
11 Doppler noise, produced by the integrated effect of solar optical position measurements in the Jet Propulsion Labora-
plasma along the signal paths which passed nearby the Sun, tory Orbit Determination Program (ODP) (Moyer 1971).
was a factor of 21 greater than the Voyager! Doppler noise. The zero-degree gravity parameter is the product of the
The Voyager 1S band Doppler data noise increased by a universal gravitational constant G and the mass ^ for each
factor of 3 over a 12 hr timespan beginning approximately 12 body. The ODP is formulated to determine the total Saturn
hr before the Saturn flyby. For this time period, attempts to system mass G^ s, and the planet mass is inferred by sub-
calibrate the coherent two-way Doppler signal at 13 cm tracting the sum of the masses of the satellites from . In
wavelength for any effects due to interplanetary plasma were general, parameters describing the positions of the space-
unsuccessful. The success of this technique is dependent on craft and the satellites Tethys, Rhea, and Titan with respect
the stationarity of the plasma over the round-trip travel time to the barycenter of the Saturnian system were estimated
of the radio signal through the plasma, which could be sever- along with G^s and the GL^s for the satellites, gravita-
al hours for a near-Earth plasma. E. Kursinski and R. Woo tional harmonics of Saturn, the locations of the Earth-track-
of JPL (private communication) have shown that an inter- ing stations, and the locations of the Saturn barycenter and
planetary plasma shock wave moving across the signal path the Earth-Moon barycenter with respect to the solar system
could account for the observed Doppler noise. Armstrong barycenter. The same set of ephemerides and dynamical con-
and Woo (1981) have studied similar effects observed in stants was used in the numerical integration for each space-
Voyager 1 Doppler-tracking records collected in 1979 dur- crafts trajectory and partial derivatives, and these served as
ing the interplanetary flight to Saturn. Gazis et al. (1985) the nominal values for the combined solutions discussed be-
reported observations of 35 interplanetary shocks by the Pio- low. To minimize the effects of systematic nongravitational
neer 11, Voyager 1, and Voyager 2 spacecraft during a 150 errors on the determined parameters, spacecraft velocity in-
day period in 1980 (ending before the Voyager 1 flyby of crements resulting from trajectory correction and from atti-
Saturn in November), when Pioneer 11 and Voyager ! were tude control were also estimated, as discussed by Campbell
both within 1 AU of Voyager 1. For the 12 hr timespan at and Synnott (1985).
Saturn, the Voyager 1 Doppler data at 3.6 cm were used in The solar barycentric coordinates of the Saturn bary-
the gravity solution, to minimize the noise effects. center and the Earth-Moon barycenter, and the masses of
Null et al. ( 1981 ) described a solution technique that es- the Sun and the remaining planetary systems, were obtained
sentially tried to account for the sequentially correlated ran- from JPL Development Ephemeris 118 ( see, e.g., Newhall et
dom nature of the plasma effect on the radio signal, by esti- al. 1983). A priori uncertainties for these coordinates were
mating an exponentially correlated Doppler bias parameter taken from the covariance matrix associated with DEI 18.
for each of a number of subintervals of the total tracking arc. The a priori Saturn-centered coordinates of the major sat-
In the present analysis, however, we have chosen to combine ellites were obtained (as described by Null et al. (1981))
the Pioneer Doppler data with the higher-accuracy Voyager from a computer implementation of the theories of Struve
data, using data weights for the Pioneer Doppler that were (1933) using numerical coefficients determined by Kozai
computed from the rms of Doppler residuals over subinter- (1957,1976); the theory and coefficients for Hyperion are
vals of the tracking arc. from Woltjer (1928) and those for lapetus from Sinclair
The only noncoherent one-way data included in the gravi- ( 1974). Partial derivatives with respect to constants of the
ty analysis spanned approximately 2 hr just after the Voyager theory were developed by G. Null of JPL (private communi-
1 Titan flyby and approximately 6 hr just after the Voyager! cation) to allow improvements as new position data became
Tethys flyby, during which times there were no coherent available. An error covariance on these theoretical constants
two-way data. Parameters describing the bias and drift of the was provided by J. Lieske of JPL (private communication)
on-board oscillator which served as the frequency reference and used as a priori information in the present analysis. The
for the one-way signal were included in the solutions for orbits of the larger satellites for which masses were not deter-
Titan and Tethys masses. The two oscillators had quite dif- mined were fixed according to ephemerides determined by
ferent behaviors as a consequence of flying through Saturns R. Jacobsen from fits to Voyager optical data. The orbits for
radiation environment. The Voyager 1 oscillator had been the small satellites were provided by S. Synnott of JPL (pri-
radiation hardened as a consequence of its flyby in 1979 vate communication), also from fits to Voyager optical data.
through the intense Jovian radiation environment; its crystal The rotational pole of Saturn could not be estimated accu-
structure, and thus its reference frequency, was not signifi- rately from the combined dataset, and was fixed at the direc-
cantly altered by subsequent radiation exposure. The tion determined by Simpson and Tyler ( 1983). Their pole
Voyager 2 oscillator was subjected to a lower radiation dose determination was made using stellar occultations observed
at Jupiter in 1979 because of its greater flyby distance, and by instruments aboard the Voyager spacecraft, and by a ra-
thus was less radiation hardened, and more vulnerable to the dio occultation of Voyager 1 observed on Earth. Both sets of
Saturn radiation field. observations depended on accurate knowledge of the space-
In addition to Doppler tracking, Earth-spacecraft range craft orbit (Campbell et al. 1982).
measurements accurate to better than 0.1 km were made A priori information constraining the first two even zonal
during the Voyager missions, providing Earth-Satum range harmonics J2 and /4, obtained by Null et al. (1981) from
fixes and an improved ephemeris of Saturn, thereby satellite orbital precession rates, was used to obtain the indi-

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1489 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1489

vidual solutions for each spacecraft but had no influence on Table II. Combined Pioneer and Voyager gravity results.
the combined Pioneer and Voyager results.
1989AJ

The combined solutions were obtained by first scaling the Solution Formal Adopted
individual information arrays of the least-squares process so parameter a Value uncertainty b uncertainty0
that effectively the assumed measurement noise for each
missions dataset approximately equaled the post-fit root- GMS 37940630 26 200
mean-square (rms) of the residuals for the dataset. These GM
Tethys 39 4 10
arrays were then combined by the usual methods of least GM
Rhea 154 2 4
squares, discussed, for example, in Lawson and Hansen Titan 8978.2 0.3 1
(1974). lapetus 107 3 10
/2 ^ 16298 5 10
V. EFFECTS OF SMALL COMPONENTS OF THE J
A -920 26 50
GRAVITATIONAL FIELD
6 x 10
6
J 75 41 80
The sensitivities of the parameter solutions to the effects of C
22 0.7 0.2 1
five small components of the gravitational field were calcu- S
22 -0.2 0.3 1
lated and in this section we summarize the result of these
calculations. The five components are: the tides raised on *GM units are km3sec*2
Saturn by the satellites, the masses of the Saturnian rings, the ^ One-sigma value from simultaneous ft to all data
masses of the small satellites, the assumption for the magni- c
Reflects observed sensitivity. See Sec. VI.
tude of /8, and the second-degree gravity harmonics for Ti-
tan.
Using the potential function (Kaula 1968), as described
by Campbell and Synnott ( 1985 ), and the tidal coefficient of All the combined solutions in the tables were obtained by the
degree 2 (Love Number) from Gavrilov and Zharkov method outlined in Sec. IV. In addition to the parameters
( 1977), the largest tidal effect was computed for Titan, for shown in Tables II-IV, constants describing the motions of
which the only signature of detectable magnitude (greater the spacecraft, the Earth and Saturn systems, the satellites of
than 1 mHz) in any of the Doppler datasets appears in the Saturn, and the Earth receiving stations were also estimated,
Pioneer Doppler, with maximum amplitude of 3 mHz. and discussed in Sec. IV. Table VI summarizes Saturn gravi-
Tyler et al. (1983) estimated the total mass of the ring ty solutions obtained from spacecraft data, and contains our
system from the Voyager 1 radio occultation of the Saturn adopted solution.
rings as 3x 10-8 Saturn masses, with a factor of 4-8 as a The formal errors given in Tables II-IV, based on the
probable error in the mass estimate. Assuming an upper assumption of zero mean, Gaussian measurement errors, are
bound on total ring mass of 3x107 Saturn masses, the expected to be somewhat optimistic because they are calcu-
maximum effect on the Voyager 1 Doppler is less than 3 lated assuming that no significant systematic effects have
mHz, on the Voyager! Doppler less than 20 mHz, and on the been neglected in the data analysis. But surely there are un-
Pioneer Doppler less than 150 mHz. The effect of this ring- modeled systematic effects on the Doppler data from the
mass assumption on the combined Pioneer and Voyager attitude-control system and from intervening plasma. In ad-
gravity solution was found to be negligible. dition, small systematic errors probably exist in the imaging
Several small satellites in the region extending out to the data due to several possible distortion mechanisms in the
orbit of Tethys were found in Voyager images ( Synnott et ah Voyager vidicon camera, which affect determination of the
1981) . All closest-approach centers of satellite images.
distances to The
the adopted
Pioneer and error bars given in
Voyager spacecraft were greater than 100 000 km, and with Table II represent a more realistic assessment of the true
size estimates available from the Voyager data (Stone et al.
1982) and reasonable values of density, Doppler perturba- Table III. Pioneer 11 gravity results.
tions of less than 1 mHz were found.
Slattery ( 1977) estimated a value of /8 based on interior
models of /8 = lOx 10-6, and this value has been as- Solution Value and
parameter a formal uncertainty0
sumed in the present analysis. The effect of ignoring /8 and
higher-order terms in the Saturn gravity expansion is less
than 1 mHz for all Doppler sets. GM s 37940635
Following Hubbard and Anderson (1978), one can com- GM
Tethys 39
pute upper limits for Titans second-degree gravitational GA
*Rhea 154
harmonics J2 and ^22- For a uniform density distribution, GA/
Titan 8978.2
-6 6
/2 < 49 X10 , and C22 < 14X 10~ , for which the perturba- GM lapetus 136123
tion to the Voyager 1 Doppler data collected near the Titan 163021 17
flyby is less than 4 mHz. The effect of assuming these values -9121 32
of J2 and ^22 for Titan in the combined gravity solution is ; 106 75
negligible. u 0.7
22
S
22 -0.2
VI. RESULTS AND DISCUSSION
Estimates for planet and satellite masses, and gravity har- a
monic coefficients, as obtained from the complete set of Pio- GM units are km3 sec-2
neer and Voyager tracking data, are given in Table II, and ^Uncertainty shown is formal one-sigma value. Where no uncertainty
from the individual mission datasets in Tables III and IV. is given, solution value is from combined solution (Table II).

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1490 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1490

Table IV. Voyager gravity results.3 sets are the primary sources of information from which each
parameter estimate is derived. For example, the Doppler
1989AJ

Solution Voyager 1 Voyager 2 Voyager 1 and data from Voyager have the greater sensitivity to the Saturn
parameter b data only data only Voyager 2 data zonal harmonics, but the Voyager 1 data are less sensitive in a
formal statistical sense than the Pioneer 11 Doppler to a
GM s 37940702 62 37940520 40 37940635 28 change in the mass of lapetus. On the other hand, none of the
GM Tethys 39 40 6 45 4 individual mission datasets is sensitive at a useful level to the
GM Rhea 156 2 154 153 2 planet zonal harmonic J6, in part because of flyby geometry,
GM Titan 8978.2 0.3 8978.2 8978.2 0.3 in part because of data outages, but the combined mission set
GM y.lapetus 131 48 109 4 106 4 does have modest sensitivity to J6.
16297 6 16288 12 16296 4 Because of these different sensitivities, only a subset of all
-932 61 -969 34 -924 21 the possible parameters are estimated in the individual mis-
x 106 75 75 75 sion solutions. For each mission, those parameters were not
0.4 1 -0.3 0.6 1.1 0.3
0.3 1.6 0.7 0.6 0.5 0.4 estimated that showed far larger formal uncertainties or esti-
mate excursions in a full parameter-estimation run than
a
were found from one or more of the other datasets. In a case
Uncertainty shown is formal one-sigma value. Where no uncertainty in which a constant was not estimated, its value was fixed by
is given, solution value shown is from combined solution (Table II). other datasets. Because these ignored parameters were
b GM units are km3 sec2 weakly correlated with the parameters actually solved for,
this procedure produced insignificant effects on the values of
the estimated parameters shown in Tables III and IV for the
estimate uncertainties. These error bars either span the individual missions. Table V summarizes the results of sev-
range of solutions obtained from different missions, or the eral sensitivity tests of the adopted solutions, from the indi-
range of solutions obtained from sensitivity experiments vidual missions and from the combined mission solution, to
with one dataset. These sensitivity tests are described below. alternative modeling assumptions and variations in data
The results in Tables III and IV help identify which data- treatment.

Table V. Summary of solution sensitivity testing.

Mission Resultant perturbations a


Perturbation solution
induced affected GMS G^Tethys ^^Rhea ^^Titan ^"^1 apetus ^xlO^

Saturn ephemeris fixed Pioneer 11 +20


at Voyager 1 determination
Alternate algorithm for Voyager 1 +0.45
centerfinding of Voyager 1
Titan images
Omit stochastic velocity Voyager 1 -22 +1.3 -31 -20
model for Voyager
spacecraft Voyager 2 +20 +3 -9 +11 +42

Combined +40 +3 -1 +0.2


Pioneer,
Voyager
Modifications to Voyager 2
non-coherent doppler data set:
i) include all 6 hrs, weighted Voyager 2 -90 +13 -8 +15 +49
at rms of post-fit residuals
ii) Same as i), except omit Voyager 2 -90 +2 +11 +42
first 2 hrs
in) Omit all data Voyager 2 +2 -8

a
Relative to combined solution given in Table H. The absence of a perturbation result indicates
sensidvity at an insignificant level reladve to the formal sigma for that parameter.
^GM units are km3 sec-2

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1491 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1491

The mass of lapetus estimated from the Pioneer 11 data of a constant bias, were included in the final Voyager 2
was found to be sensitive to Saturn ephemeris variations Doppler dataset; these data were downweighted by a factor
1989AJ

which induce differences in the projected Saturn-centered of 10 relative to the rms of post-fit residuals.
velocity vector of the spacecraft and hence in the observed Another quantification of possible systematic effects sum-
Doppler. Comparisons of partial derivatives indicate that marized in Table V was obtained from the sensitivity testing
the ephemeris errors can mimic or closely resemble satellite of the individual and combined Voyager results to the model
mass or harmonic-coefficient signatures on the Doppler. As for nongravitational acceleration ( NGA ). An upper limit to
indicated in Table V, if the Saturn ephemeris is fixed at the the sensitivity to NGA was obtained by removing complete-
Voyager 1 solution, the mass of lapetus estimated from Pio- ly the estimation of the stochastic velocity component, used
neer 11 Doppler data differs by 20 km3 s-2 from the Pioneer to model the small, frequent velocity increments induced by
11 mass estimate obtained if the ephemeris coordinates are the Voyager attitude-control system.
fixed at the combined Pioneer and Voyager values. The results of Table IV were obtained by weighting the
Similarly, the mass of Saturn estimated along with the coherent Doppler and optical data according to the rms of
Saturn ephemeris coordinates from the Voyager 1 data the respective post-fit residuals. In general, for both Voyager
differs by 182 km3 s-2 from the Foyager 2 estimate. The high solutions, varying the optical data weighting relative to the
value of GJifs in column 2 of Table IV may indicate that the Doppler weighting did not significantly affect the gravity
Voyager 1 Doppler or range data has a small systematic sig- solutions. In another sensitivity test, an alternate method
nature left over after all the known systematic effects are used by J. Riedel of JPL (private communication) was used
accounted for, possibly due to the effect of the interplanetary to compute the center of Titan images used in the Voyager 1
plasma shock waves hypothesized earlier. Any unmodeled optical dataset. This resulted in changes to center locations
signature may be absorbed in the determination of the ephe- on the order of several times the residual rms. When these
meris or mass parameters. Both the Pioneer 11 and Voyager 1 data were substituted for the nominal Titan imaging data,
gravity-parameter solutions were found to be insensitive to the mass of Titan increased by 0.45 km3 s-2.
the high data noise at encounter. When the large number of The plots of Figs. 5 and 6 are of post-fit Doppler residuals
Doppler data with residuals of 3<7 or larger were removed from the near-encounter region of each mission. In general,
from each dataset, no significant differences were observed Doppler counts were sampled at 60 s intervals for the time
in any of the gravity-parameter solutions for the individual period near the planet flyby. The less dense samples shown in
missions. For this sensitivity test, the data weightings were the plots were computed at 300 s intervals. The epoch of
fixed at the higher residual rms values. closest approach is near the center time point of each plot.
For comparison to our results, we note that E. M. Stan- For consistency in plotting, Voyager 1 3.6 cm (Xband) re-
dish of JPL (private communication) has obtained a value siduals for the 12 hr period beginning 12 hr before periapsis
of 37 939 520 + 1700 km3 s-2 for the mass of the Saturn were scaled downward to S band by the ratio 11/3. The rms
system from reduction of photographic plates and Earth- of 60s S' band residuals is approximately 116 mHz for Pio-
Viking lander range measurements. His value is in reasona- neer 11, or equivalently 7.6 nuns 1i 57 mHz or 1 mms-1 for
ble agreement with that of Table II. Voyager 1 3.6 cm residuals, and 5 mHz or 0.3 mms-1 for
Both sets of Voyager Doppler included noncoherent coherent Voyager 2 13 cm Doppler residuals, and 14 mHz or
Doppler that, as discussed earlier, was shifted in frequency 2 mms-1 for Voyager 1 and Voyager 2 noncoherent 13 cm
as a consequence of flying through the Saturn radiation envi- residuals. Small, short-period (P<few hours) systematic
ronment. The Voyager 1 Doppler data included approxi- trends still remain in all post-fit Doppler residuals. These
mately 1.5 hr of noncoherent Doppler acquired near the Ti- trends are likely due in part to errors in the correction for
tan closest approach during preparation for occultation Earth ionosphere, but in the case of Pioneer 11 and Voyager 1
entry, and for a period after occultation exit. These data are principally due to the integrated effect of solar plasma
exhibited a slight bias and drift in reference frequency due to along the Pioneer 11 signal path which passed nearby the
oscillator instability, which was removed by solving for bias Sun, and to the perturbation to the Doppler signal induced
and drift parameters. These data were downweighted by a by the plasma shock wave hypothesized for the Voyager 1
factor of 10 relative to the post-fit rms of the data residuals, encounter period. The coherent Doppler residuals are least
and did not contribute significantly to the strength of the noisy for Voyager 2; however, the noncoherent Doppler re-
Titan mass determination, nor to the remaining gravity-pa- siduals exhibit trends remaining from the attempts to model
rameter solutions. the effects of the Saturn radiation dosage on the behavior of
The preliminary Voyager 2 Doppler dataset included ap- the on-board oscillator.
proximately 6 hr of noncoherent Doppler acquired just after A typical set of residuals for the last few weeks of optical
the flybys of Saturn and Tethys. As discussed earlier, these data is shown in Fig. 7 for Voyager 2. The data are fit to 0.2
data contained a significant signature attributable to a shift pixels rms (1 pixel = 105 radian; see Smith etal. ( 1977) for
in the oscillator reference frequency. The majority of the a description of the Voyager imaging system). The slight
signature could be removed by hypothesizing a correlation trends at the very end of the data arc reflect the fact that the
between Saturn radiation dosage level and oscillator re- distances at which these frames were taken were only 3 X 106
sponse, and by including an oscillator bias and drift model. km, and therefore a 0.2 pixel residual amounts to only 6 km,
As indicated in Table V, there was significant sensitivity of at which level there may be systematic effects in the satellite
the Voyager 2 estimates obtained for the masses of Saturn, and spacecraft motions that are difficult to model. The re-
Tethys, and lapetus, and also for J2 and to the noncoher- siduals in pixels appear to be biased downward by 0.1 pixel,
ent Doppler content and assumed weighting. A noticeable and this may indicate that there is a real bias in centerfinding
signature remained in the data residuals after the fit to the of about this level, which would not be at all unexpected
entire 6 hr of noncoherent Doppler. Therefore, only the last because of difficulties in dealing with overexposed satellite
4 hr of these data, when the signature before the fit consisted images.

American Astronomical Society Provided by the NASA Astrophysics Data System


1492 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1492

N
n:
600-
h) PI0NEER 11 s
SUN -EARTH -S P AC ECR AF
< (SEP) ANGLE = 7 deg
O
CO
LU

OU

b PERIAPSIS

i i i i i i r
00:00 10:00 20:00 06:00 16:00
09/01 09/01 09/02
EARTH RECEIVE GMT, HRS ON 1, 2 SEPT 1979

COI 600-
o VOYAGER 1
r-H
SEP ANGLE = 45 deg
I
<
ZD
O
CO
LU Fig. 5. Postfit Doppler residuals from
O' the solution of Table II. The Pioneer 11
or residual rms is a factor of 21 greater
LU
I than the rms of Voyager 2 residuals,
Q_ due to the low value of the SEP angle
Q_
O
O during the Pioneer 11 Saturn flyby.

TITAN C/A PERIAPSIS I RHEA C/A


<
C -600- i i r "ii r
00:00 10:00 20:00 06:00 16:00
11/12 11/13
EARTH RECEIVE GMT, HRS ON 12, 13 NOV 1980

600-
VOYAGER 2
SEP ANGLE = 35 deg
<
=>
o
CO
LU
Od
C
CL
CL
O
O

< -600- PERIAPSIS TETHYS C/A


CO i i i i i ii i iilliiiiiir
08:00 18:00 04:00 14:00 00:00
08/25 08/26 08/27
EARTH RECEIVE GMT, HRS ON 25, 26 AUG 1981

Nicholson and Porco ( 1988 ) estimated J2, */4, and J6 from ics of Saturn through J6. Therefore, we decided to determine
a combination of the Pioneer 11 results from Null et al. the gravity parameters with the Nicholson and Porco results
(1981) and from Voyager observations of an eccentric ring- included as a priori information with the Voyager
let at 1.2908R s (Porco et al. 1984). Under the assumption Doppler and star-satellite imaging data. The ringlet
that the eccentricity is forced by Titan, the Voyager ringlet constraint from Eq. (39) of Nicholson and Porco
imaging data place a tight constraint on even zonal harmon- (1988) is (/2- 16297XlO"6) - 1.507(/4 + 910Xl0-6)

American Astronomical Society Provided by the NASA Astrophysics Data System


U
00LO
1493 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1493
CT)O'
the ringlet constraint being in the solution. Even with the
h) addition of the ringlet constraint, there is a high correlation
CO< (0.943) between and J6. Hence a linear relationship
00
CO between the zonal harmonics can be found with higher accu-
racy than either J4 or J6 alone. That relationship is
J6-0.3635(/2 - 16298X10"6) -0.8809(74 + 915X10"6)
= (103+ 16)XlO~6. The expected contribution of J2 to
this relationship is a small fraction of the standard error, so
we essentially have found a relationship between J4 and J6
that must be satisfied for any viable interior model of Saturn.
11/12 11/13 Table VI summarizes four gravity-parameter solutions
EARTH RECEIVE GMT, HRS ON 12, 13 NOV 1980 obtained from these four datasets: (1) the Pioneer 11
Doppler tracking and satellite secular rate data as reported
by Null /a/. ( 1981 ) ; ( 2 ) Doppler and star-satellite imaging
data from the two Voyagers, combined with dataset 1 (this
paper); (3) the eccentric ringlet observations from the two
Voyagers combined with dataset 1 (Nicholson and Porco
198 8 ) ; ( 4 ) the complete dataset; consisting of the Pioneer 11
and Voyager Doppler and star-satellite imaging, combined
with the Nicholson and Porco constraints on J2, /4, and J6
(this paper). Our adopted solution was obtained using data-
set 4.
08/25 08/26 08/27 Our solution for Cl2 and S22 is relatively good; in fact, the
EARTH RECEIVE GMT, HRS ON 25, 26 AUG 1981 formal solution would indicate a possible detection of a sec-
ond degree and order sectorial harmonic with amplitude
Fig. 6. Postfit Doppler residuals for Voyagers 1 and 2 from the solutions of
Table IV. In general, the on-board frequency reference for the noncoher- J22= ( 1-0 + 0.5) X 10-6 and a phase in cartographic longi-
ent Doppler data was considerably less stable than for the coherent tude of 0 + 25. However, in the presence of the various
Doppler. systematic errors discussed earlier, we are not willing to cer-
tify an accuracy in the gravity field of less than 10-6. In-
stead, we place an upper limit on J22 of 2x 10-6. For pur-
+ 1.586(/6 107X 10~6) = ( -2.8 + 12) XlO6, for as- poses of comparison, the value of J22 resulting from the tide
sumed values oft/g = 10x 10-6, /10 = Jl2 = 0. With the raised on Saturn by Titan is on the order of 10-8. Similarly, if
addition of the constraint, only J6 was improved. Except for there were a single object in the outer B ring at a radius of
Tethys, changes to the mass values and uncertainties were 1.81S, the radius associated with Saturn electromagnetic
insignificant; we adopted mass solutions that resulted from discharges (Warwick et al. 1982), our bound on J22 would
place an upper bound on its mass. From formulas given by
Hubbard and Anderson ( 1978 ), the value of J22 produced by
a satellite of mass in units of Saturns mass at a distance r is
J22 (3/4)a(i s/t*)3/?2. The response-4coefficient a for Sat-
urn is on the order of 0.2, so /^ < 10 . A relatively large
mass is required to raise a significant tide on Saturn, even at a
distance of 1.8 li? s. Such a mass would produce large unob-
served dynamical effects in the rings and probably would be
unstable to tidal disruption. It most certainly would be de-
tectable in the observed orbital precession rates of Mimas
( Kozai 1976; Null et al. 1981). In short, there is no physical
source external to Saturn for the relatively large value of J22
indicated by our formal solution. If there is a sectorial har-
CALENDAR DATE 1981 monic J22 of order 2X10-6, it must be excited by Saturn
itself.
With regard to satellite masses, our recommended values
for Tethys, Rhea, and lapetus differ, by 0.9cr, 1.2(7,
and 2.4cr, respectively, from the preliminary values re-
ported in Tyler etal. (1982). Our masses for Rhea and lape-
tus differ by 0.06<r and 0.5cr from those reported by Null
et al. ( 1981 ) from an analysis of the Pioneer 11 data alone.
The significant change in the mass of lapetus from the pre-
liminary value is particularly noteworthy. The Pioneer 11
analysis yielded a mass of ( 19.3 + 7.3) X1023 g, while the
preliminary mass from Voyager 2 was ( 18.8 + 1.2) X 1023 g.
Our combined analysis (Table VI) yields a smaller mass of
CALENDAR DATE 1981 (15.9 + 1.5) X1023 g.
Fig. 7. Voyager 2 optical data residuals from the Voyager 2 solution of Table Our value of (6.7 + 1.5) X 1023 g for the mass of Tethys
IV. differs by 0.32(7 from the value of (6.22 + 0.13 ) X1023 g de-

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1494 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1494

Table VI. Summary of Saturn gravity-parameter solutions.


1989AJ

Data Set 3
Pioneer 11, Data Set 4
Voyager Ringlet Pioneer 11,
Data Set 1 Data Set 2 Observations Voyagers 1,2
Pioneer 11 Pioneer 11, Nicholson Voyager Ringlet
Solution^ Null, et al Voyagers 1,2 and Porco Observations
Parameter (1981) CThis Paper) (1988) (This Paper)

GM, 37938544 2386 37940630 200 37940630 1200


GM Tethys 39 10 45+ 10
GM Rhea 152 * 34 154 4 15414
GM Titan 9058 114 8978.2 1 8978.21 1
GM lapetus 129 49 107 10 1061 10
16299 18 16298 10 16297 18 162981 10
-917 37 -920 50 -910 61 -9151 40
106 75 80 107 50 1031 50
-22 0.7 1 0.71 1
?
22 -0.2 1 -0.21 1

a
All results adjusted to Saturn reference equatorial radius of 60330 km
^GM units are km3sec-2

termined by ground-based observations of the Mimas- by compression of material within the satellite. It is more
Tethys orbital resonance (Kozai 1957). Although we usual- difficult to explain the high density of Dione. Compression
ly recommend using a mass determined from flyby data effects in lapetus, with its low mean density, are not expected
rather than one dependent on long-term perturbation effects to exceed 3%, well within the uncertainty of 10% in the
of celestial mechanics, we make an exception for Tethys. As density determination. Although all the satellite densities of
stated earlier, much of the Doppler tracking during the Table VII are consistent with a solar ice/rock ratio, the sig-
Voyager 2 close approach to Tethys was in a noncoherent nificantly higher density of Dione suggests that it may be a
mode, and hence was dependent on the stability of the on- member of a distinct class of satellites, as discussed by Tyler
board crystal oscillator. Our analysis of systematic errors in et al. (1982). The mean uncompressed density of all the
the noncoherent Doppler data sets the 1er error at +22% satellites is significantly lower than the mean for the five
for the mass of Tethys, and hence the same percentage error principal satellites of Uranus at 1.48 0.06 g cm-3 (Ander-
for the density. A density error of 22% is marginally dis- son et al. 1987), and indicates a fundamental compositional
criminatory among competing compositional models. On difference between the satellites of the two planets. Perhaps
the other hand, the determination from the Mimas-Tethys the Uranian satellites are enhanced by elemental carbon, as
orbital resonance is accurate to 2%, a far more satisfactory suggested by Anderson et al. ( 1987).
level of error. Consequently, we adopt masses determined by
spacecraft tracking for only Rhea, Titan, and lapetus. Val-
ues of for Mimas, Enceladus, Tethys, and Dione from
Kozai ( 1957) are 2.50 0.06, 4.9 2.4, 41.53 0.83, and Table VII. Mass, size, volume, and density of the intermediate-size satel-
70.2 2.2 km3 s-2, respectively. lites of Saturn.
The mean densities for all six satellites of intermediate size
are given in Table VII, where we have used the most recent Mean Mean
Radius Volume Density
Voyager radii determinations for Mimas (Dermott and (km) 10'23cm3)
(x 10 (g cm-S )
Thomas 1988), Enceladus (Davies and Katayama 1983a),
Tethys and Dione (Davies and Katayama 1983b), Rhea Mimas 0.375 0.009 198.8 0.6 0.329 0.003 1.14 0.03
(Davies and Katayama 1983c), and lapetus (Davies and Enceladus 0.73 0.36 251 5. 0.662 0.040 1.12 0.55
Katayama 1984). After increasing the volume of Tethys by Tethys 6.22 0.13 524 5. 6.03 0.17 1.03 0.04
1.0%, Dione by 1.4%, Rhea by 7.5%, and lapetus by 2.6% Dione 10.52 0.33 559 5. 7.32 0.20 1.44 0.06
to allow for compressional effects, we bbtain a total volume Rhea 23.1 0.6 764 4. 18.68 0.29 1.24 0.04
for all six satellites of (50.48 0.65) X1023 cm3 and a total lapetus 15.9 1.5 718 8. 15.50 0.52 1.02 0.10
mass of (56.85 1.69) X 1023 g, which yields a mean un-
compressed density for the system of (1.126 + 0.036) Masses for Mimas, Enceladus, Tethys, and Dione are from Kozai (1957).
Radii are from Dermott and Thomas (1988) for Mimas, and from
gem-3. The relatively high density of Rhea can be explained Davies and Katayama (1983a, b, c, 1984) for the other satellites.

American Astronomical Society Provided by the NASA Astrophysics Data System


97.1485C

1495 J. K. CAMPBELL AND J. D. ANDERSON: SATURNIAN SYSTEM 1495

VIL CONCLUSIONS ments from the Cassini spacecraft flybys of the principal sat-
This paper essentially confirms the results reported by ellites are expected to further reduce the errors on those
1989AJ

Null et al ( 1981 ) and by Nicholson and Porco ( 1988) for satellite masses. A good flyby determination of the mass of
the Saturn zonal harmonic coefficients 72, J4, and J6. The Dione is particularly important.
results for C22 and S22, though consistent with zero, may
alternatively indicate an excitation of sectorial harmonics by
the planet. Mass determinations for Rhea, Titan, lapetus, We wish to thank R. A. Jacobson, T. P. Krisher, E. R.
and Saturn are improved considerably as a consequence of Kursinski, and S. P. Synnott for helpful discussions through-
the Voyager missions. We note that future large reductions out the work. We are grateful for an informal review of the
in the uncertainty of J6 will probably not result from analysis manuscript by G. W. Null and E. M. Standish, and of Sec. VI
of radio-tracking data (because there is no mission planned by A. J. R. Prentice, and for comments by an anonymous
with a low Saturn orbit), but may come from a proposed referee, all of which resulted in improvements to the text.
Cassini mission which could observe optically the preces- This analysis represents one phase of research carried out
sions of the lines of nodes or apsides of the smaller inner at JPL under NASAs Geosciences Program. We acknowl-
satellites, or provide increased resolution for observing the edge our support from the Voyager Project Office at JPL,
kinematics of eccentric ringlets in the Saturn system, as de- and one of us (J. D. A.) acknowledges the support of the
scribed by Nicholson and Porco (1988). Doppler measure- Pioneer Project Office at NAS A/Ames Research Center.

REFERENCES
Anderson, J. D., Campbell, J. K., Jacobson, R. A., Sweetnam, D. N., Tay- Kozai, Y. (1957). Ann. Tokyo Obs. Ser. 25, 73.
lor, A. H., Prentice, A. J. R., and Tyler, G. L. ( 1987). J. Geophys. Res. Kozai, Y. (1976). Publ. Astron. Soc. Jpn. 28, 675.
92, 14877. Lawson, C. L., and Hansen, R. J. (1974). Solving Least Squares Problems
Armstrong, J. W., and Woo, R. ( 1981). Astron. Astrophys. 103, 415. (Prentice-Hall, Englewood Cliffs).
Campbell, J. K., and Synnott, S. P. ( 1985). Astron J. 90, 364. Moyer, T. D. (1971). Mathematical Formulation of the Double-Precision
Campbell, J. K., Synnott, S. P., and Bierman, G. J. ( 1985). Kalman Filter- Orbit Determination Program (DPODP), TR 32-1527 (Jet Propulsion
ing: Theory and Applications (IEEE, New York). Laboratory, Pasadena).
Campbell, J. K., Jacobson, R. A., Riedel, J. E., Synnott, S. P., and Taylor, Nicholson, P. D., and Porco, C. C. ( 1988). J. Geophys. Res. 93, 209.
A. H. ( 1982). Voyager 1 and Voyager 2 Saturn Encounter Orbit Deter- Null, G. W., Lau, E. L., Biller, E. D., and Anderson, J. D. ( 1981 ). Astron.
mination, paper 82-0419, presented at AI AA 20th Aerospace Sciences J. 86, 456.
Meeting, January 1982. Porco, C., Nicholson, P. D. Borderles, N., Danielson, G. E., Goldreich, P.,
Chao, C. C. (1971). TR 32-1526 (Jet Propulsion Laboratory, Pasadena). Holberg, J. B., and Lane, A. L. ( 1984). Icarus 60, 1.
Davies, E. M., and Katayama, F. Y. ( 1983a). Icarus 53, 332. Simpson, R. A., and Tyler, G. L. (1983). Astron. J. 88, 1531.
Davies, E. M., and Katayama, F. Y. (1983b). J. Geophys. Res. 88, 8729. Slattery, W. L. (1977). Icarus 32, 58.
Davies, E. M., and Katayama, F. Y. ( 1983c). Icarus 56, 603. Smith, B. A., etal. (1982). Science 215, 504.
Davies, E. M., and Katayama, F. Y. (1984). Icarus 59, 199. Stone, E. C., and Miner, E. D. (1982). Science 215, 299.
Dermott S. F., and Thomas, P. C. ( 1988). Icarus 73, 25. Synnott, S. P., Peters, C. F., Smith, B. A., and Morabito, D. A. (1981).
Gavrilov, S. V., and Zharkov, V. N. ( 1977). Icarus 32, 443. Science 212, 191.
Gazis, P. R., Lazarus, A. J., and Hester, K. (1985). J. Geophys. Res. 90, Tyler, G. L., Eshleman, V. R., Anderson, J. D., Levy, G. S., Lindal, G. F.,
9454. Wood, G. E., and Croft, T. A. ( 1982). Science 215, 553.
Green, D. W., Lam, V. W., and Royden, H. N. (1980). Effects of the Tyler, G. L., Marouf, E. A., Simpson, R. A., Zebker, H. A., and Eshleman,
Charged Particle Environment on Voyager Navigation at Jupiter and Sat- V. R. (1983). Icarus 54, 160.
urn, AI AA 80-1650 (American Institute of Aeronautics and Astronau- Warwick, J. W., Evans, D. R., Romig, J. H., Alexander, J. K., Desch, M.
tics, New York). D., Kaiser, M. L., Aubier, M., Leblanc, Y., Lecacheux, A., and Pederson,
Hubbard, W. B., and Anderson, J. D. ( 1978). Icarus 33, 336. B. M. (1982). Science 215, 582.

American Astronomical Society Provided by the NASA Astrophysics Data System

You might also like