Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 85, NO.

B7, PAGES 3939-3951, JULY 10, 1980

GeophysicalParametersof the Earth-Moon System


A. J. FERRARI, W. S. SINCLAIR, W. L. SJOGREN, J. G. WILLIAMS, AND C. F. YODER

Jet PropulsionLaboratory, CaliforniaInstituteof Technology,Pasadena,California91103

Doppler tracking data from Lunar Orbiter 4 have been combinedwith laser ranging data from lunar
retroreflectorsto yield a number of geophysicaland geodeticparametersfor the earth and moon. This
joint solutiongivesvaluesof (1) thelunarprincipalpolarmomentC/MR 2 = 0.3905+ 0.0023,(2) GME --
398600.461 +_0.026km3/s2,and(3) an earth/moonmassratioat 81.300587+ 0.000049.Alsodetermined
are the harmonicsof a completelunar gravity field throughdegreeand order 5, the obliquity of the lunar
pole, selenocentriccoordinatesof the lunar retroreflectors,geocentriccoordinatesof the McDonald Ob-
servatory,and the lunar secularacceleration.The lunar potential Love number is weakly determinedat
0.022 +_0.013, and a surprisinglylarge dissipationof rotational energy is inferred, though either solid
body tidal dissipationor liquid core mantle interactionscould be causes.

INTRODUCTION monic coefficientsC2oand C22have inhibited further improve-


ment in the accuracy of the lunar moment of inertia.
This investigationpresentsthe first simultaneousreduction
Most of the early determinationsof the gravity harmonics
of Doppler tracking data from a lunar satelliteand lunar laser
ranging data to determine geophysical parameters of the were obtained from direct fitting to lunar orbiter tracking
earth-moon systemsuch as the lunar moment of inertia and a data. In general,the data from theseeccentricorbits were ill-
low-degreelunar gravity field. Each of thesedata types alone suited for accurate determinationsof the second-degreehar-
providesincompleteinformation on the moment of inertia. monics,sincethe higher-degreetermsin the gravity field had
The gravitational harmonicsof the seconddegreegive two in- significanteffectsas the satellitepassedthrough typical peri-
dependentrelationshipsamong the three principal moments apsis altitudes of 50-100 km. The data acquired from the
of inertia. The laser ranging data measurethe two fractional Apollo satelliteswere even more acutely affectedby this con-
differencesin the lunar momentsof inertia through their ef- dition, since these orbiters were generally in low, nearly
fectson the physicallibrations (rotation vector) of the moon. circular orbits. Extraction of a second-degreeand second-or-
The union of these data types provides four measuresof the dersetof estimates
['romthesedatawouldrequirethedeter-
three principal momentsand is usedto determinea unique es- mination of a gravity model consistingof hundredsof terms to
timate. This investigationseeksto utilize the complementary fit the data to its intrinsic noiselevel. Recently, the long-term
strengthsof the Doppler and laser rangingdata to achievein- variationsin the orbital elementsof Explorer 35 and 49 have
been used to determine a limited set of zonal coefficients
ternally self-consistentestimatesfor the moments of inertia.
The lunar laser rangingdata have sensitivityto the second- [Blackshearand Gapcynski, 1977]. Error studies show that
and third-degree gravity field through variations in the lunar theseorbits too are sensitiveto short-periodvariations due to
rotation. These physicallibrationsare sizable,becauseone he- harmonicsof up to degree and order 10. A single arc of data
mispherealways facesthe earth. Though it has long been un- from the Lunar Orbiter 4 missionhas been usedpreviously to
derstoodthat the second-degreeharmonicsdrive the physical determine a third-degree and third-order gravity model [Sjo-
librations, it is a more recent discovery that several of the gren, 1971]. These data have ideal geometric characteristics
third-degree terms are of observable size [Eckhardt, 1972; for low-degree harmonic determination, since the orbit was
Kaula and Baxa, 1973].The rangingdata have excellentsensi- large(periapsisheight2700 km and apoapsisheight6200 km)
tivity to C3,•and S3,•and good sensitivityto C3o,while the ef- and relatively insensitiveto the higher-degreegravity struc-
fects of the remaining third-degree harmonics are weak or ture. However, for our purposes many uncoupled attitude
correlated with the effectsof other harmonics.Gravity fields, control maneuversmake the majority of these data useless.
based on lunar laser data, have been published by Williams Previous reductions of spacecraftDoppler data produced
[1977] and Calame [1976, 1977]. A combination of laser and only differencesof the moments of inertia. Previous reduc-
tions of lunar laser data have assumed a nominal subset of
very long baseline interferometrydata was usedby King et al.
second- and third-degree gravity harmonic coefficientswhile
[1976]. It seemsonly natural to try to add the strengthsof the
Doppler tracking and lunar laserdata typesto improve the es- solvingfor the remaining coefficients.
timatesof the lunar homogeneityparameterC/MR '• and the
low-degreelunar gravity field plus additional geophysicalpa- DATA DISTRIBUTION AND ACCURACY
rameters of the earth and moon.
Lunar laserranging data measurethe round trip travel time
PREVIOUS RESEARCH of a laser pulse from McDonald Observatoryto any one of
four retroreflectorarrays on the moon [Bender et at., 1973].
Previous determinations of the moment of inertia have been
The range data usedin this analysiswere taken in the time pe-
performed via a two-stepprocess:the Doppler and laser data riod betweenAugust 1969and 1977.Data taken through 1973
have been reduced separately,and estimatesfor the gravity have been published as normal points by Abbot et at. [1973],
harmonics and the fractional moment differences have been
Shetuset at. [1975], and Muthottandet at. [1975]. The quality
combinedin an a posteriorimanner to determine the moment of the rangedata varies,the newerdata being of progressively
of inertia. The uncertaintyin the estimatesfor the triaxial har- better quality. An instrumentalerror accompanieseach data
Copyright ¸ 1980 by the American GeophysicalUnion. point and was used for the relative weighting of the ranges.
3939
Paper number 80B0276.
0148-0227/80/080B-0276501.00
3940 FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM

ARC 7
seven of the eight arcs on the lunar farside spanned 90ø-
250øE longitude in noncontiguouscoveragewith one nearside
short arc. A schematic diagram of the data distribution is
shown in Figure 1. The Doppler data were acquired at a
samplerate of one point per minute. In order to make compu-
tations lessexpensive,only one data point in five was used in
the analysis. Since the gravity field model to be determined
was of a low degree and order, we felt that the deletion of
270 o .
these data would still permit a complete sample of the neces-
sary frequencies.The actual impact of this data selection
processwill be discussedin a later sectionof the paper. The
Doppler data were weighted at a 1-mm/s (0.015 Hz) standard
error.

CONSTRAINTS AND RELATIONSHIPS


EQUATOR
Of the two second-degreeharmonicsdetermined by the lu-
nar orbiter Doppler data and the two fractional moments of
inertia determined from the lunar laser data, only three pa-
•C6•
E•TH
rameters are independent. When the two data setsare com-
bined, this dependenceis satisfiedby including an equationof
Fig. 1. Doppler data distribution.The lines representthat portion of constraint. Since the mean rotation of the moon is locked onto
the orbit nearestperiapsis. its orbit about the earth, the additional equationsof constraint
are introduced to zero the time-averagedtorque on the moon
The 2212 rangeshave been weightedoverall to give an aver- (about the axestoward the mean earth and mean rotationdi-
age uncertainty of 1.4 m, which is an order of magnitude rections). For purposes of convenience, several additional
larger than the instrumental uncertainty. This weighting, constraintsare introducedwhile combiningthe normal equa-
which resultsin the assumedrms errorsbeing3.8 timeslarger tions, generatedby separatecomputer programs,for the two
than the postfit residuals,was used in this analysisto com- data types.
pensate for systematicerrors in the observationsand model. For a rigid moon the Doppler trackingestimatesfor the sec-
The Lunar Orbiter 4 Doppler data usedin the reductionare ond-degreegravityharmonicsC20and C22providetwo inde-
those used by Sjogren[197l] augmentedby sevenother data pendentrelationshipsamongthe three principalmomentsof
inertia:
arcs. The arcs vary in length from l0 hours to 2.5 days. We
have performeda comprehensivereview of the postflightatti-
tude control summaries for all the lunar orbiter missions. Of
C20
= •-• • C (1)
all the data taken, only thoseof Lunar Orbiter 4 are usable for
the determination of a low-degree gravity field. All other 1
phasesof these missionseither had periapsisheightsof ap- C22= 4MR2 (B- A) (2)
proximately 100 km, making theseorbitsvery sensitiveto the
very high degree structureof the lunar gravity field, or had A, B, and C are the principal momentsof inertia in increasing
frequent attitude maneuvers.The Lunar Orbiter 4 spacecraft order by size. M is the lunar mass,and R is the mean radius.
wasin a nearlypolar orbit (inclined85ø to the lunar equator) The laser ranging experiment measuresthe variations in the
having an orbital period of 12 hours.During the entire phase moon's orientation due to physicallibrations, and thesedata
of this missionthe periapsislatitude varied only several de- give estimatesof the fractional differencesin the moments:
greesabout its nominal value of 0ø. The spacecraftattitude
# = (c- = (B - (3)
control system was quiescent during the eight data blocks
listed in Table 1. The selenocentricperiapsislongitudesof Consequently,of the above four parametersyielded by the

TABLE 1. Doppler Data Coverage


Woomera, Goldstone,
Australia United States Madrid, Spain
Arc No. of s.d., No. of s.d., No. of s.d., Data Span, Calendar
No. Points Hz Points Hz Points Hz days Date, 1967
I 180 0.0066 253 0.0069 140 0.0100 2.25 May 29
2 89 0.0054 48 0.0098 0.5 May 8
3 99 0.0069 53 0.0085 0.5 May 9
4 150 0.0060 91 0.0059 134 0.0086 1.375 May 10
5 175 0.0085 169 0.0053 210 0.0100 2.33 May 27
6 54 0.0115 49 0.0199 0.375 May 18
7 73 0.0054 28 0.0030 71 0.0099 0.625 May 31
8 119 0.0192 1.0 June 4

Total numberof pointsis 667 (Woorecta)+ 813 (Goldstone)+ 705 (Madrid) = 2185;0.01 Hz = 0.65
FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM 3941

two data types,only three are independent,since In addition to the above constraintsthere is also a scaling
factor which was applied to the partial derivativesof the lunar
C,•o= - 2C,•,(2fl- 3'+ fl3')/3'(1+ fl) (4) physicallibrations with respectto the seventhird-degreehar-
It is usefulto relateC/MR'• to the aboveparametersby using monics. Since the third-degree terms in the physical librations
are proportionalto the third-degreeharmonicsdivided by the
C/MR 2= 4C22/3' (5) moments of inertia, or equivalently, C,•o,it was necessaryto
while the mean moment I is related to the polar moment by allow for the • 1% variation of the final value of C,•ofrom that
used in originally constructingthesepartial derivatives.This
I C (3+ fl + 3'- f13')
• 0.999655
C scalingfactor only entersthroughthe lunar laserrangingdata.
---•= MR2
MR 3(1+fl) MR2 (6) Consideration of the torque equations for physical libra-
tions showsthat two linear combinationsof fl, 3', C3•,and C33
For an elasticmoon there are small changesin the physical enter the libration theory more strongly than the individual
librations which are significant for the lunar laser ranging parameters.These combinations
data [Yoder, 1979]. It is preferable to include the mean rota-
tional distortionsin the constantpartsof fi, 3',C2o,C22,and C/ R MR:
MR'•. Also includedin the definitionof the last three parame- •' = • + •aa• (30C33
- 11C3,) (12)
ters are the mean tidal distortions.For fl and 3' most of the
permanenttidal distortionis includedexceptfor that fraction
correspondingto the lunar elastic Love number k,•. Periodic 3"=3'+ •aa• (90C33
- C3,) (13)
and mean distortionsdue to k,• are accountedfor during the
may be deduced from Williams et al. [1973] and Kaula and
calculation of the physical librations. The constraints,modi-
fied accordingto the above definitionsfor an elasticbody, be-
Baxa [1973]; •' is determined by the obliquity of the lunar
come equatorto the ecliptic,while 3"can be obtainedby measuring
the annual term in longitude librations. These are the two
0 = C:o3'(1+ fl) + 2C::(2fl - 3'+ f13')- k:(fi - 3')E (7) largest libration terms. Consequently,•' and 3" are much
more accurately determined by the lunar laser data than •
0 = (C/MR:)3'- 4C:: + k:E (8)
and 3'.
whereE = 7.40x 10-6 - 0.985(ME/MM)(R/a)3,wherea is the The above constraints allow the two data sets to be com-
semimajoraxis of the lunar orbit. The lunar laserranging data bined with the errors in C/MR'• properly propagatedfrom
provide a weak measurementof the lunar k: in the joint solu- C:o, C::, fl, and 3'.
tion. There is no modeling of the nonstatictidal perturbations
STATISTICAL ANALYSIS
on the lunar orbiter or of the nonstatictidal displacementsof
the retroreflectors.The latter effect mostly mimics the signa- A statisticalerror analysiswas performed to study the ef-
ture due to the reflectorcoordinatesand getsabsorbedduring fects of various error sourceson the solutionparametersprior
the solutionas submeterdisplacementsin thesequantities. to the final processingof the data. This analysiswas used to
There are severalother major constraintswhich are usedin define the optimum set of solutionparameterswhich could be
the data reduction.The principal axesof minimum moment A determined from the eight Doppler arcs.The statisticalanaly-
and maximum moment C have mean directions which are sis assumes a maximum likelihood estimator with Gaussian
biasedslightlyfrom the mean earth and mean rotation direc- errors on the observations. In the common formulation of a
tions,respectively.Thesebiasesare small, both --,1.4 arc rain, maximum likelihood or weighted least squaresestimator the
and are causedby the torquesarising from the third-degree computed statisticsare based on the effects of random data
harmonics.See Eckhardt [1972], Kaula and Baxa [1973], Wil- noise only and do not reflect the effect of model errors in the
liams et al. [1973], and Eroshkin[1974] for a more extensive solution. Hence we investigatedthe contribution of model er-
discussionof thesebiasesand other effectson the physical li- rors to the solution.
brationsdue to the third-degreeharmonics.Becausethe com- The error analysesinvestigatethe sensitivityof the gravity
puter program which processedthe lunar laser data used a field and moments of inertia determination to data noise, un-
principalaxissystem,while the programfor the Doppler data modeled higher-degree gravity effects, and solar radiation
analysisused a mean earth/polar axis coordinate system,it pressuremodel errors. A previousstudy [Ananda et al., 1977]
was necessaryto connectthe harmoniccoefficientsin the two has investigatedthe effectsof the presentuncertaintiesin sta-
systems while performingthejoint solutions.The smallsizeof tion coordinates,troposphericmodeling, and timing biasesbe-
the biasesallows C:o, C::, and the third-degree harmonics to tween stations.They have a negligible impact on the moment
be set equal in the two coordinatesystems.To first order in of inertia accuracy.In our analysisthe errors(Table 2) for the
the small rotations the balancing of torques in the mean gravity field of Ferrari [1977] are used.A solar radiation pres-
earth/polar axis coordinatesyields sure model error of 30% is assumed.
A nominal set of statisticswas generated for the complete
C,•,= ]K(C3o- 10C3,) (9)
fifth-degree and fifth-order gravity solution, which would re-
S:• =0 (10) sult from the eight Doppler arcs,consideringthe contributing
effects of the uncertainties in the harmonics of degree 6-8.
S,•:= K(S3, - 30S33) (11)
The results indicated that the sectoral harmonic C,•,•is sensi-
while all three coefficientsare identically zero in the principal titive to unmodeledeffectsof the sectoralterms (C66,S66)and
axis system.The constantK is roughly R/4a. The more exact (C77, S77),sincethere was about a 40% increasein the com-
value of K -- 0.001108that is used in this analysisis adopted puted error when the uncertaintyin thesehigher-degreesec-
from a solution for the forced physical librations by D. H. torals was considered. An accurate estimate of this term is de-
Eckhardt (private communication,1974). sirable, since it is used along with C,•oand the fractional
3942 FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM

TABLE 2. Errors Used in StatisticalAnalysis of nominal model parameterswas usedfor each data type to
Unnormalized Normalized carry out thesecalculations.It was assumedthat the nominal
setof model parametersusedwas closeenoughto the final or
I rn CX 107 SX 107 CX 105 •-x 105
convergedset that the differential correctionprocessfor the
6 0 70. 0.19 least squareswas within the linear region of convergence.
6 I 33. 28. 0.42 0.36 Tests were performed using slightly different setsof nominal
6 2 6.1 6.7 0.49 0.54
conditionswhich showed this assumptionto be true.
6 3 0.79 0.76 0.38 0.36
6 4 0.25 0.27 0.63 0.64 Doppler partial derivativesand residualsweregeneratedfor
6 5 0.051 0.053 0.64 0.64 eachof the eight data arcspreviouslydiscussed [Moyer, 1971].
6 6 0.013 0.014 0.56 0.56 In the initial stagesof convergencethe solutionparametersfor
7 0 180. 0.47
each of the data arcsincludedonly the dynamical state (posi-
7 1 22. 22. 0.30 0.30
7 2 4.3 4.8 0.43 0.48 tion and velocity) of the spacecraftat each epoch. A nomi-
7 3 0.81 0.78 0.58 0.55 nal third-degreeand third-order lunar gravity field was used
7 4 0.081 0.080 0.38 0.38 as the reference model. The state only solution vector was
7 5 0.020 0.022 0.57 0.57
processed for eachdata arc until the leastsquaresalgorithm
7 6 0.0040 0.0044 0.64 0.64
7 7 0.0012 0.0012 0.67 0.67
experienced convergence.An exampleof the structureof the
8 0 97. 0.24 resultingDopplerresidualsafter the firststageof processing
is
8 1 30. 30. 0.43 0.43 shownin Figure 2a. The remainingstructureis attributedto
8 2 3.4 3.0 0.41 0.41 errors in the nominal gravity field model. Once convergence
8 4 0.52 0.52 0.52 0.52
of the stateonly vectorwas attainedfor each Doppler arc, an
8 7 0.065 0.065 0.50 0.50
8 5 0.0083 0.0083 0.46 0.46
additionalcalculationgeneratedan augmentedpartial deriva-
8 6 0.0016 0.0016 0.57 0.57 tive matrix which included partials for the complete set of
8 7 0.00029 0.00029 0.57 0.57 desired lunar gravity field harmonic coefficients,the lunar
8 8 0.00007 0.00007 0.54 0.54 gravitationalconstantGMM, and the dynamicalstatevectors.
Reference for coefficientsfrom Ferrari [1977]. Radiation pressureis For the lunar laser ranging data a nominal set of model pa-
30%, tracking station coordinatesare 10 m, troposphericerror is 10%, rameterswas assumed,and partial derivativeswere computed
and timing bias is 0.0001 s. for that model. The partial derivativesand the weighted data
residualsfor the ranging data were combined with those for
the Doppler data in the final stageof the iteration process.
moment differencesfl and y to determinethe momentof iner- The solutionalgorithm usedfor the leastsquaresestimatoral-
tia parameter C/MR 2. The perturbinginfluencesof the re- lows for the combination of the two normal equations from
maining sixth-, seventh-,and eighth-degreeterms were found the two different sets of nominal values for the harmonics.
to be small and negligible. This analysis indicated that a The vector of solution parameters, as was previously in-
larger gravity field model than the originally planned fifth de- dicated, containsa wide variety of terms,someof which affect
gree and order was necessaryto include the degradingeffects either the Doppler or the laser measurementalone and some
of the higher-degree harmonics. of which have a common influence. The solution vector con-
A secondset of statisticswas generated for the solution pa- tains a total of 133 terms. A complete list of these parameters
rameters of an augmentedgravity field model consistingof a is given in Table 3. Included in the solutionare 35 harmonics
full fifth-degree and fifth-order model and the sixth- and sev- of the lunar gravity field, the lunar gravitationalconstant,the
enth-degreesectoralterms.The resultsfrom this new solution earth plus moon gravitationalconstant,the dynamic statevec-
set indicated that the computederrorswere within 10%of the tor for each of the eight Doppler arcs,coordinatesfor the four
actual errors which result from consideringthe uncertainties lunar laser retroreflectorsand McDonald Observatory, differ-
in the remaining sixth- through eighth-degree unmodeled ential corrections to the lunar orbit, the coefficient of the
gravity terms.On the basisof theseresultsthis solutionsetap- Nordtvedt term (a relativity term), startingconditionsfor the
pearsto be optimum within the constraintsof the data used.In libration integrations, fractional differencesin the lunar mo-
addition, the results of the consider portion of the statistical ments of inertia, the earth rotation and orientation (direction
analysis give confidencethat the solution is almost totally of pole in space) parameters,lunar deformation constants
dominated by data noiseand that the effect due to unmodeled (Love number and dissipation),and the lunar polar moment
parametersis negligible.We believe that this set of statistical of inertia parameter. Table 3 also indicateswhich data type
studiesplacesrealisticconfidencelimits on the solutionssub- possessed the primary sensitivityto each of the solutionpa-
sequentlyobtained from the data reduction. rameters(D is Doppler, and L is laserranging).The valuesfor
the dynamical constantsused in this data reduction which
SOLUTION APPROACH
were assumedto be known (i.e., fixed in the model equations)
This section discusses the method of data reduction and the are listed in Table 4.
solutionparameters.The combineddata set (Doppler and lu- The weightedleast squaressolutionof the combinedsetsof
nar laser ranging) has parameterswhich are common to both normal equationsfor the Doppler and lunar laser ranging
data typesand hasconstraintsbetweensomeof the other solu- data is subjectto 12 equationsof constraint.These equations
tion parameters.Both the Doppler and the lunar laser ranging reducethe degreesof freedomof the solutionby 12. The con-
data are nonlinear functionsof the different parametersin the straints have been discussedearlier. They are equations (7)-
solution vector. The method of least squaresis usedto solve a (9) and 11,the scaledrelationbetweenthe seventhird-degree
linearizedsetof normal equationsto estimatethe solutionpa- harmonicsin the Doppler and laserdata sets,and the propor-
rameters.The data typeswere processedseparatelyto produce tionality of the scalingfactor to C2o.$2• is zeroed but is not
a set of partial derivativesand weighted data residuals.A set counted as a solution parameter.
FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM 3943

The sensitivity studiesshowed that the uncertaintiesin the the uncertainties will be correctly propagated back through
sixth- and seventh-degreesectoralharmonicssignificantlycor- the rest of the solutionparameters.Conditioningwas also ap-
rupt C:: and consequentlythe polar moment of inertia. At- plied to the fourth- and fifth-degreeterms.The a priori values
temptsat solvingfor theseadditional sectoralharmonicsgave werezero,and the a priori errorsare givenby o = 3.5 X 10-4/
valueswhich were uncertain and too large. Consequently,the /2.The conditioningis light, sincethe powerin the postfitfield
values were lightly conditioned [Ferrari, 1977;Jackson, 1979] is 8 (degreefour) and 4 (degreefive) timessmallerthan that in
with unnormalizeda priori information,taking0 + 2.3 x 10-9 the a priori uncertainties.The a priori conditioningdoeslittle
for the 6, 6 harmonicsand 0 + 1.3 x 10-'ø for the 7, 7 har- to decreasethe power in the fourth-degreefield, sincethe co-
monics. Whether one believes the final solution values or not, efficientsare typically severaltimes their postfit errors, but the

MADRID JGOLDS•01•
RI• I WOOMERA
RISE
.067

.O5O

ß033

.017

-. 017

-.033

-.050

-.067
1.75 5.50 5.25 7.00 8.75 10.5{) 12.25 14.00 15.75 17.5{) 19.25 21.00
•ay
27,
1967 TIME,
hrs
MAD
RIDRISE ]GOLD
STOI•RISE ]WOOMERA
RISE

ß017
0
-.017

-.o33
-.050
•. 75 24.50 26.25 28.O0 29.75 31.50 33.25 35.O0 36.75 38.50 40.25 42.O0 43.75
TIME, hrs
I MADRIDRISE
.O5O

ß033

.017

-. 017

-. 033

-.067
45.50 47.25 49.00 50.75 52..• M.• 56.00
TIME., hrs

MADRID [ GOLDSTONi[
RISE [ WOOMERA
RISE

0 1.75 3.50 5.25 7.00 8.75 10.50 12.25 14.00 15.75 17.50 19.25 21.00
May
27,1967 TIME,
hrs
MADRID
RISE I GOLDSTON•
RISE [ WOOMERA
RISE

< .017 -•--- ' - -

ca ' ß '. ,.,...


,.,.., , ß , ,'ß•... ,.,•,,,.
,,, .ß
"•-.017
- - -' '' '"' '
g -.o33 , , , '
22.75 24.50 26.25 28.O0 29.75 3l. 50 •. 25 35.O0 36.75 38.50 40.25 42.O0 43.75
TIME, hr$
{MADRIDRISE
ß033

ß017

0

-.017 ---- --

-.033 45.50 47.25 49.00 50.75 52.5{) M.• •.00

TIME,
hrs b
Fig. 2. Doppler residualsfrom arc 5. (a) Residualsbasedon a fixed third-degreeand third-ordergravity field. (b) Residu-
als based on the full solution of this article.
3944 FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM

TABLE 3. Solution Parameters linear combinations of parameters to be more accurately


Sensitiv- known than their individual values.This is particularly true of
ity to No. of fi and ¾,C2oand C2•,someof the third-degreeharmonics,and
Data Param- the principal axis coordinatesof the laser retroreflectors.The
Type eter eight statevectorsfor the Lunar Orbiter 4 shortarcs,the orbit
Spacecraftstatesfor eightarcs(threeposition D 48 of the moon, an adjustment of orientation between the ecliptic
and three velocity components) and equator, and the initial conditionsfor the libration in-
GMs• D 1 tegration are not given. The uncertaintiesgiven for solution
GME + GM•4 L 1
Harmonic coefficientsfor Doppler arcs D 35 parametersare formal errorsthat resultfrom the Doppler and
Lunar orbit elements L 6 range errorsspecifiedin the sectionon data plus the combina-
Nordtvedt effect L 1 tion and solution strategy given in the previous section. For
Secular acceleration of the moon L 1
both of these data types the data errors were chosen to be
Physicallibration initial conditions L 6
larger than the postfit root mean square(rms) residualsin an
fi, V L 2
Harmonic coefficients for lunar laser L 7 attempt to compensate for systematicerrors. This is a com-
Lunar Love number L 1 mon practice when working with large data sets,becausesys-
Lunar dissipation L 1 tematic errors in either the observationsor the software(in the
Coordinates of four reflectors L 12
form of small unmodeledeffects)may not propagateinto the
Coordinatesof McDonald Observatory L 3
solution like random errors. Specifically, the random errors
Earth orientationparameters L 2
Earth rotation corrections L 4 corrupt the solutionlike the inversesquareroot of the number
ScalingbetweenDoppler and laserharmonic DL 1 of observations,while a systematicerror may not be improved
coefficients
by increasingthe number of observations(but usually is). Our
C/MR 2 DL 1
Total 133
choice of data errors has been made as a simple attempt to
Constraints 12 give realisticerror estimatesfor the results.
Independentparameters 121 Table 5 givesthe harmonic coefficientsof the lunar gravity
field.Alsogivenarethenormalized
coefficients
(O•m,•q•m)
and
D is Doppler, and L is laser.
the a priori errors used to condition the solution. The gravity
field from theseparametersis plotted in Figure 3 as accelera-
tions at a 100-kin altitude reference surface. The lo uncer-
conditioning on the fifth-degree field does tend to keep it tainty is plotted in Figure 4. The far side has large positive
smaller. Solutions were attempted with and without condi- anomaliescorrespondingto the highland terrain [Sjogrenand
tioning, and the former were felt to have more realistic Wollenhaupt,1973], whereasthe front side is milder, showing
amounts of power in the fifth- and higher-degreeterms. a relatively small depressionin OceanusProcellarum.
The two setsof normal equationsgeneratedfor the Doppler The second-degree harmonicsfrom Table 5 plus the fi and
and lunar laser data were combined and solved simultane-
¾in Table 6 yield an estimatefor the lunar polar moment of
ously to estimate a set of differential correctionsto the 133- inertiaparameterof C/MR • = 0.3905+_0.0023.This value for
componentsolutionvector. The third- and fourth-degreehar- C/MR 2 compareswell with recent independentestimates.
monics,fi, ¾,the lunar Love number, and the dissipationfrom Table 7 lists estimates for C/MR • derived from a number of
this preliminary solutionwere usedin a new integrationof the reported values of C2ocombined with the values of fi and ¾
physicallibrations.The lunar laserdata were reprocessed with from this solution. It should be appreciated that previously
these new physical librations, the solution values from the publishedestimatesof the moment of inertia have usedinde-
preliminary solutionwere used as nominal valuesin the con- pendently derived values of (C20,C22)and (,B,¾),though the
straint equations,the normal equationsfrom the Doppler and latter parameters have been accurateenough in recent years
laser data were combined again, and the final solution was
generated.This iteration of the physical librations was done
becauseEckhardt [1972] has shown that some of the fourth- TABLE 4. PhysicalConstants(AssumedKnown)
degree harmonics are capable of giving libration terms like
those of C32 and S32 as well as small coordinate shifts. From Constant Value

the lunar laser data alone the third- and fourth-degreehar- Speedof light c 299 792.458km/s
monics are inseparable,and the corruption of C32and S32by Lunar radius R 1738 km
the fourth-degreeharmonicsis only at the 1-2% level, so par- AU 149,597,871 km
tial derivativeswere never generatedbeyondthird degree.Ite- GeocentricCoordinatesof DSN Tracking
ration of the physicallibrationsallowed the small effectsof the Stations
fourth-degreeharmonicsdetermined by the Doppler data to Radius Radius
be introduced into the laser data, and the final values of C32 Sta- x cos x sin
and S32shiftedby approximatelytheir errorsas a resultof the tion Latitude, Latitude, Longitude,
iteration. The size of libration signaturesdecreasesabout 2 or- Location No. km km deg
ders of magnitude per harmonic degree. Goldstone, 12 5212.0517 3665.629 243.1945377
California
RESULTS
Madrid, 61 4862.6082 4114.879 355.7509964
This sectiongives the resultsfrom the joint solution with a Spain
Woomera, 41 5450.2031 -3302.189 136.887511
priori conditioningon the harmonics.The valuesfrom the so- Australia
lution will usually be given with more digits than would seem
justified by the errors.This is done becausecorrelationscause DSN is Deep SpaceNetwork.
FERRARI ET AL.' GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM 3945

TABLE 5. Harmonic Coefficients

Normalized Unnormalized

C and S C and S
A Priori, A Priori,
I rn Cx 104 Sx 104 C, ox 104 •,,OX 104 ox 104 Cx 104 Sx 104 C, oX 104 S, oX 104 oX 104
2 0 -0.90404 0.0054 -2.0215 0.012
2 1 -0.0007854 0 0.000027 -0.001014 0 0.000035
2 2 0.34553 0.000268 0.0020 0.00010 0.22304 0.000173 0.0013 0.000062
3 0 -0.045832 0.0068 -0.12126 0.018
3 1 0.2843 0.051945 0.018 0.023 0.3071 0.056107 0.019 0.025
3 2 0.143118 0.04939 0.0014 0.0011 0.048884 0.01687 0.00049 0.00037
3 3 0.1030 -0.023977 0.012 0.012 0.01436 -0.0033435 0.0017 0.0017
4 0 0.0005 0.037 0.22 0.0015 0.11 0.68
4 1 -0.0757 0.0311 0.017 0.027 0.22 -0.0718 0.0295 0.016 0.026 0.21
4 2 -0.0666 -0.1290 0.028 0.037 0.22 -0.01440 -0.02884 0.0062 0.0083 0.050
4 3 -0.0142 -0.1320 0.022 0.017 0.22 -0.00085 -0.00789 0.0013 0.0010 0.014
4 4 -0.0733 0.0267 0.019 0.022 0.22 -0.001549 0.000564 0.00041 0.00046 0.0045
5 0 -0.134 0.060 0.14 -0.446 0.20 0.48
5 I -0.0381 0.0786 0.082 0.084 0.14 -0.0326 0.0673 0.070 0.072 0.12
5 2 0.0961 -0.0322 0.051 0.053 0.14 0.01556 -0.00522 0.0083 0.0086 0.023
5 3 -0.0448 0.0385 0.075 0.082 0.14 -0.00148 0.00127 0.0025 0.0027 0.0048
5 4 0.0768 0.0586 0.044 0.035 0.14 0.000598 0.000456 0.00034 0.00027 0.0012
5 5 0.0495 0.0557 0.045 0.053 0.14 0.000122 0.000137 0.00011 0.00013 0.00029
6 6 0.0007 0.152 0.060 0.069 0.10 0.16 x l0 -6 0.353 x 10-4 0.14 x 10-4 0.16 x 10-4 0.23 x 10-4
7 7 0.014 -0.014 0.065 0.065 0.07 0.25 x 10-6 -0.25 x 10-6 0.12 x 10-5 0.12 x 10-5 0.13 x 10-5

to justify this procedure.It can be seenthat the result of this to be effectivelyfixed and the error in C/MR • to be domi-
paper is in excellent agreementwith the result of Blackshear nated by the 0.5% error of the second-degreecoefficients.The
and Gapcynski[1977], probably the only other independent value of/• in Table 6, 631.69 +_0.13 x 10-6, differsfrom the
study of comparable accuracy.These authors follow a differ- value of King et al., [1976],631.27 +_0.03 x 10-6, by 3 times
ent approach from that used here. They use the long-period our uncertainty.It has recently been discovered(R. W. King,
and secularperturbationson the orbits of Explorer 35 and 49 private communication,1979) that there was an inconsistency
over long spansof time without attemptingto accountfor the in the Apollo lunar surfaceexperimentpackagedata process-
influence of harmonics other than C22and the zonals C2o-C6o. ing which gave an artifically small error while restrictingthe
Gaœcynski et al. [1975] used a subsetof the data of Blackshear sizeof the changein/• from the nominal lunar rangingexperi-
and Gapcynski[1977], and similarly, Sinclairet al.'s [1977] de- ment 2 value [Williams, 1977].The valuesof ff and ¾'are also
termination was a preliminary result of the study in this pa- given in Table 6. The value of the lunar obliquity to the eclip-
per. It is clear from the scatterin the table that the quoted er- tic dependsmainly on/•' with small correctionsfor harmonics.
rors on someof the earlier valuesof C2owere unrealistically This derivedvalue dependson D. H. Eckhardt's(private com-
small. The geophysicalimplicationsof the moment of inertia munication, 1978) theory with a small correctionapplied for
parameter and the spectralpower in the gravity field will be changesin the lunar laser determinedorbital inclination (X.
discussedin a subsequentsection. X. Newhall et al., private communication,1978) comparedto
Table 6 givesfour parameterswhich enter into the libration the lunar theory (T. C. Van Flandern, private communication,
model and somederived quantities.The lunar laser data de- 1978).
termine the linear combinationsof/• and ¾with C3l and C33 The two remaining measurementsin Table 6 are the lunar
given in (12) and (13) much more accuratelythan/• and ¾ potential Love number k2 and the parameter k:T, which
alone.Thus it is the improvederrorson C3l and C33,which are measureslunar rotational energydissipation.The •60% error
controlledby the lunar orbiter data, that permit the errors on of the Love number allows detection to be claimed, but a re-
/• and ¾ to be diminished over those based on the laser data duction would be necessaryfor meaningful constraints on
alone. As can be seenfrom (7) and (8), the small relative er- models of the lunar interior. Within the stated error our value
rorsof 2 x 10-4 in/• and 4 x 10-4 in ¾causethe ratio C:o/C:: is consistentwith valuesderived by Chengand Toks6z [1978].

Fig. 3. Gravity field at 100-km altitude. Units are in milligals.


3946 FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM

! 80 270 0 90 180
LONGITUDE

Fig. 4. Gravity field uncertaintyat 100-kmaltitude.Units are in milligals.

The latter are derived from two interior modelswhich depend consequentlyMs/(ME + M•,). A definedvalue of the speedof
on fitting the lunar seismicdata. Values of 0.0293 and 0.0335 light and an accuratevalue for the length of the astronomical
are given for their models A and B, with the difference a unit (Table 4) permit the latter to be convertedinto GME +
rough measureof the error. Though there is a substantialun- GM•. The derived values of GME and the earth-moon mass
certainty in k,, the product k,•T is better determined.The non- ratio follow directly. Earlier lunar laser resultsare given by
zero value of this parameter implies that rotational (physical King et al. [1976] and Williams [1977]. These resultshave rea-
libration) energy dissipationis taking place in the moon. The sonableconsistency,in comparisonto the errors,as do the re-
dominant observable effect is a shift in the direction of the sultsof Espositoand Ng [1976] and Esposito[1978],which are
pole away from the Cassini alignment. Unfortunately, the basedon the independenttechnique of interplanetaryspace-
physicalcauseof thisterm hastwo possibleexplanations. The craft tracking. Also concordantis the recent value [Lerch et
first involves tidal dissipationin the solid body of the moon, al., 1978] obtained from satellite laser ranging.
which would give a lunar Q. In this case,T is the time delay in In Table 9 are given the selenocentriccoordinatesof the lu-
the lunar solid body tides, equivalent to a phaseshift, which nar retroreflectorarrays. For convenience,coordinatesin both
influencesthe lunar physicallibration through the time-vary- the principal axis and mean earth/polar axis systemsare
ing componentof the moment of inertia tensor.Then k,/Q = given. The principal axis system is shifted from the mean
0.00165 + 0.00023. With the Love number in the table, Q = 14 earth/polar axes by the libration parameters%i• -- 80.025",
with a spreadof 5-25. Using the k,• from model A of Cheng P•bias = -84.956" and P2bias -- 0.305"due to the gravity field
and Toks6z gives Q -- 18. The Q apply to tidal deformations (D. H. Eckhardt, private communication, 1978), elasticity,
at a period of 1 month. The secondexplanation is rotational and rotational energy dissipation.
dissipationat an interfacebetweena liquid core and a mantle. In Table 10 are given further resultsfrom the solution. The
If it is assumedthat all of k,•T arisesfrom a viscousinteraction valueof ri•vg,the averagelunar secularacceleration in longi-
of the core-mantle interface, with the core density taken as 8 tude, is identical with the average of severalsolutionsover the
g/cm3, then a 500-km-radiuscorewould requirea kinematic same time interval given and discussedby Williams et al.
viscosityof at least 1.5 St, and a 360-km core a viscosityof at [1978].The lunar secularaccelerationis modeledas thoughit
least 20 St, while a 300-km core would need a minimum of arisesfrom tideson the earth alone, but the surprisinglylarge
• 100 St. A minimum-sized core of 230-km results from a vis- value of k2T implies a small contribution from tides on the
cosityof 3000 St. It seemsthat the unexpectedlylarge value of moon. The value in Table 10 should be considered a total
k,•T requiressome combinationof a surprisinglysmall solid value. If the lunar dissipationis all in the solid body, then the
body Q or a surprisinglylarge coreviscosityor radius.The re-
sultsof this paper are improvementson the valuesreported by
Yoder et al. [1978]. TABLE 7. C/MR 2 Solutions FromPrevious Valuesof C2o
The product of the gravitational constantand the massesof Standard
the earth and moon and the mass ratios of earth, moon, and Reference Value* Deviation
sun are given in Table 8. The lunar orbiter data gives GMM,
Akim [1966] 0.3981 0.042
while the lunar laser data give the size of the lunar orbit and Lorell and Sjogren[1968] 0.3916 0.003
Michael et al. [1969] 0.4002 NA
Lorell [1970] 0.3781 0.002
Liu and Laing [1971] 0.3857 0.004
TABLE 6. Moment of Inertia and Libration Parameters
Sjogren[1971] 0.3958 0.006
Parameter Value Michael and Blackshear[1972] 0.3939 NA
Bryantand Williamson[1974] 0.3889 0.008t
fi 631.687+ 0.132x 10-6 Gapcynskiet al. [1975] 0.3918 0.003
3, 228.022+ 0.100x 10-6 Sinclair et al. [1977] 0.3922 0.003
/•' 629.980+_0.016x 10-6 Ananda [1977] 0.4076 0.008
3" 228.593+ 0.051x 10-6 Ferrari [1977] 0.3954 0.007
,4/MR 2 0.3903 + 0.0023 Blackshearand Gapcynski[1977] 0.3910 0.002
B/MR 2 0.3904 + 0.0023 This article 0.3905 0.0023
C/MR 2 0.3905+ 0.0023
NA means not available.
I/MR 2 0.3904 + 0.0023
Obliquity to ecliptic 5553.64+ 0.2" *Assumingfi and 3' of this paper.
k2 0.022 + 0.013 •Realistic error estimate(W. C. Bryant and R. G. Williamson, per-
k2T 0.0072 + 0.0010 days sonal communication,1979) as opposedto formal error of 0.0004 in
the original paper.
FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM 3947

TABLE 8. Masses and Ratios TABLE 10. Tidal Acceleration and Nordtvedt Effect

Parameter Value Parameter Value

GMat 4902.7993+ 0.0029km3/s2 Tidal Acceleration


GME + GMat 403,503.260+ 0.026 km3/s2 k2• 0.0122 + 0.0016
GME 398,600.461+ 0.026km3/s: tiavg -23.8 + 3.l "/(century)
2
Mtr/M•t 81.300 587 m 0.000 049 Nordtvedt Coefficient
Ms/(Me + M•t) 328,900.540 m 0.021 Coefficient of cos D -0.06 + 0.11 m
,/ -0.006 _+0.012
Ms is the massof the sun.

k2T value of Table 6 contributes+0.6"/(century)• to ri, but if scale signaturesof several milliseconds,which the tabulated
the dissipationis at a core-mantleinterface,then the contribu- expressiondoes not remove. This illustrateshow systematic
tion to the secularaccelerationis approximately10%as large. errors can invalidate the formal errors for some parameters.
In principle, dissipationin the earth and moon could be dis- Recently, a 1979 BIH systemhas been defined which would
tinguishedthroughdifferentseculareffectson the eccentricity, remove most of the annual signaturethat we find.
but a longer data spanwill be required to achievethis. The quoted error in longitudeis internal to the lunar laser
The coefficient of the Nordtvedt term is virtually null ranging systemand includesrandom differencesin UTO, but
within the errors, a result which continues to confirm the peak excursionsin the UTO differencescan reach twice this
equivalenceprinciple in relativity. The error is reducedover value. Longitude differencesbetween multiple sitescould be
the valuesreportedby Williamset al. [1976]and Shapiroet al. derived with much smaller errors so long as the data spans
[1976], who give more thorough discussionsof the implica- overlap. Discussionsof the UTO differenceshave been made
tions. by $tolz et al. [1976], Harris and Williams [1977], King et al.
The geocentriccoordinatesof the intersectionof axesof the [1978], and Calame and Guinot[1979]. The geographicpole in
2.7-m telescopeat McDonald Observatoryare givenin Table the LLR systemshould be aligned with the 1903.0 Conven-
11. Both cylindricalcoordinates(rs,•, z) and sphericalcoordi- tional International Origin pole, since the BIH polar motion
nates (r, h, q•)are tabulated. It shouldbe appreciatedthat in valueshave been adopted.
all geodeticsystemsthe zero point in longitudedependson as-
SOLUTION RESIDUALS
sumptionsor definitions.The mutual adjustmentof the zero
points of longitudesystemsis outsideof the scopeof this pa- Residuals for the longestDoppler arc are shown in Figure
per. For the caseof lunar laserrangingin this studythe most 2b. These residualsare very nearly at the inherent noiseof the
important definitionsare that Newcomb's general precession data. The Doppler residualshave a standarddeviatior•of ap-
in longitude is used,the ephemerisof the earth about the sun proximately0.01 Hz (Table 1). It is satisfyingthat the Doppler
is based on DE 90 (E. M. Standish,private communication, residualsappear very nearly random in nature. In performing
1975), and the nominal values of UTI and polar motion are this analysis,only one out of every five Doppler points was
the Bureau International de l'Heure (BIH) 5-day smoothed usedin an effort to minimize the computercost.That decision
resultsfrom circular D (1968 system).The longitude is mea- resultedin a degradationin the formal uncertaintieso! the so-
sured east. There is a further complicationdue to systematic lution by a factor of 2.2. If every Doppler datum had been
signaturesbetweenthe BIH and lunar laser ranging (LLR) used, an improvement in the real accuracyof the solution pa-
valuesfor UTO at McDonald, which is UT1 plus the projec- rameters would have been expected, but the amount of im-
tion of polar motion into the longitude at that site. As a con- provementwould likely be lessthan the above formal factor.
sequence,severalsystematicterms are solvedfor and the re- Figure 5 showsthe one-way rangeresidualsfor the lunar la-
sult of this solutionin millisecondsof time (1 ms • 0.4 m) is ser data from the joint fit. The rms range residual is 39 cm or
given in Table 11, where L' is the mean longitudeof the sun 2.62 ns. If the laser data are fit independently of the Doppler
(annual period) and [2 is the lunar node (18.6-year period). data, then the rms residual only drops to 38 cm. This small
The coefficientsof the linear and 18.6-yearterms shouldnot change demonstratesthat there is no obvious inconsistency
be consideredsignificantdespitethe quoted errors, because betweenthe Doppler and laser data for the parameterswhich
they are not stablewith changingdata span.Put another way, overlap, the second-and third-degreeharmonics.It is evident
the UTO differences(LLR - BIH) showsystematiclong time from the figure that the earliest data are least accurate, and

TABLE 9. Reflector Coordinates TABLE 11. ObservatoryCoordinatesand UTO Corrections


East Value
Sys- Longitude, Latitude,
McDonald Coordinates
Reflector tem Radius, m deg deg
5,492,414.40 + 0.23 m
Apollo 11 PA 1,735,479.32 23.4508575 0.6948663 255.9780334ø + 0.0000053ø (0.51 m)
Apollo 11 ME 1,735,479.32 23.4729736 0.6732504 3235,696.83 + 0.41 m
Apollo 14 PA 1,736,341.46 -17.5004071 -3.6218407 6374,664.69 + 0.29 m
Apollo 14 ME 1,736,341.46 -17.4786234 -3.6443727 30.5032333+ 0.0000034ø (0.37 m)
Apollo 15 PA 1,735,482.48 3.6070107 26.1566115
Apollo 15 ME 1,735,482.48 3.6284697 26.1330647 SystematicTerms UTO(LLR) - UTO(BIH) in Milliseconds
Lunakhod 2 PA 1,734,643.44 30.9056351 25.8521561 Term in L' (- 1.56_+0.21) cosL'
Lunakhod 2 ME 1,734,643.44 30.9219591 25.8319508 Term in L' (-0.95 _+0.23) sin L'
Term in t (+0.40 +_0.12) (t(yr) - 1969.5)
PA is principal axis coordinate system, and ME is mean earth Term in [2 (-3.43 +_0.98) sin [2
coordinate system.
3948 FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM

o oo this has been accountedfor in the weighting. One low-weight


o
o
o **oO.•
point in the first year is off scaleat +4 m. The prominentver-
o
tical banding is due to an absenceof data near new moon.
•.•.o
o o
oo Each band is a separatesynodicmonth (29.53-daysinterval).
oo

o oø goøø oo The noisein the presentdata spanis similar to that previously


o /'
analyzed from a shorter span [Harris and Williams, 1977];
---30%of the rms can be attributed to the noise in the adopted
nominal values of the earth's rotation. Months with larger er-
o o%• rors in the rotation will have longer bands in the figure. The
øo
postfitrmserror of the laserdata is still 2.7 timesthe assigned
instrumental noise. Not all of this can be accountedfor by er-
o
rors in earth rotation, and the difference is not well under-
stood.For the purposeof this studythe mostimportant errors
would be related to the physicallibration model. An examina-
tion of detailed plots of the residualsdoesnot show obvious
physicallibration signatures,
that is, separationof the residu-
als of different reflectorsranged at similar times. It is the na-
.• o •oa•OOo ture of the time scalesassociatedwith the physicallibrations
that errorscan build rapidly with an increasingdata span. Fu-
Oo•o
ture data will test whether the presentmodel is sufficient.
DISCUSSION

In addition to the joint solution, separate solutions were


,.,., o made with only the lunar orbiter or lunar laser data. Though
o

most of the third-degreeharmonicsagreedreasonablyin com-


o oc?•-,=. parisonto the errors,whichcanbe quitedifferentfor the sepa-
o,,.,•o •.,,,.,,bc
'
o o•1•' rate solutions,there is one exceptionwhich was disturbingly
large. For C30the lunar laser data alone yield -9.5 + 2.2 x
o 10-6, while the Dopplerdata alonegive -19.5 +_3.7 x 10-6.
o,.,•
The joint value is -12.1 + 1.8 x 10-6. The laser data alone
o oo••, ø give a value of 0.0056 +_0.0012 daysfor k2T, a value which
o o o•,o•oO differsby 1« times its error from 0.0072 + 0.0010 days of the
joint solution. ,
. ,:,.•,.oo o•. ø
oo o o,•bo ecl., Another joint solution was performed without a priori er-
,• o.,øcl=eo
rors on the harmonics. When a priori conditioning was ap-
plied to all higher degrees,the direct effect appearedto be a
o ,, oø o• o •o'•",• o '•o general decreasein the power in the fifth degreeand a sizable
reduction in the magnitude of the sixth- and seventh-degree
sectoralharmonics. Of course,there is always an indirect ef-
fect on the low-degreeharmonicswheneverthe higher-degree
harmonics change value. In the unconstrained solution, C50
appearedsuspiciouslylarge. Though the magnitudeof this co-
efficientis reducedwith a priori conditioning,it still seemsun-
expectedlylarge. Since the geometryof C50resemblesthat of
C30(odd zonals),there may be a commoncausefor the pecul-
iarities noted for these coefficients.

GEOPHYSICAL IMPLICATIONS

The estimatesfor the lunar gravity field harmonicsgiven in


Table 5 were convertedto degreevariances(spectralpower)
as follows:

The valuesof the spectralpower componentsand their formal


errors are listed in Table 12. It should be recalled that the
fifth-, sixth-,and seventh-degree
harmoniccomponentsare af-
fected by the conditioningapplied to the least squaressolu-
tion. The standard deviation per harmonic coefficient,o/, for
I each degreewas computed(seeFigure 6) from the degreevar-
ianceso/•= V/(2l + 1).
If the lunar topographyharmonics[Bills and Ferrari, 1977b]
FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM 3949

TABLE 12. Lunar Potential SpectralComponents crustaldensitypoof approximately


2.85g cm-3 [BillsandFer-
Degree, rari, 1977a]. If it is also assumedthat the moon possesses a
I V1x 10•2 Error x 10•2 ol x 106 Error x 106 small core which is 300 km in radius (•: -- 0.173), as a single
2 9366.8 0.3 43.28 0.26
seismicevent suggests,we can examine the impact of the
3 1197. 12. 13.1 1.3 mean density and moment of inertia estimatesas constraints
4 515. 62. 7.6 2.6 on our simple model. It shouldbe understoodthat the choice
5 542. 431. 7.0 6.3 of these model parameters for the lunar crust and core serve
6 231. 84. 10.8 6.5 as a set of nominal conditions which we combine with the in-
7 4. 85. 1.4 6.5
ternal structureconstraints(14) and (15). The assumedcore
and crustparametersare in no way unique, and even a simple
three-layer lunar density model contains five parameters
are converted into their gravitational equivalent by using a which cannot be determined from the two constraint equa-
surfacemass density approximation,the standard deviations tions alone.
can be comparedto the derived gravity field results(see Fig- The solution of the mean density and mean moment of in-
ure 6). The resultsfor Olshowthat the relative amplitudesof ertia equation for the above model parametersresultsin a
the two spectra are different, which would imply that the mantlehavinga densityof p• = 3.355gcm-3 and a corehav-
moon possesses lateral densityvariationsat depth [Bills and ing a densityof p: = 11.88gcm-3. Thesedensityvaluescorre-
Ferrari, 1977a]. spondto a moon having a mantle composedof basalticmate-
The estimates for the lunar mean density • and the mean rial and a very densecore. It is the departureof the estimate
momentof inertia parameterI/MR: and their respectiveun- of I/MR: from 0.40, the value for a radially homogeneous
certaintylevelsplace two constraintson the internal structures planet, which allows our model to have this internal density
of the moon. Both of these constraints are derived from the
structure. If our initial assumptionshad allowed for a larger
values in this solution. The radial structure of the moon is de-
coreradius(•: > 0.173),the resultingcoredensitycontrastAp:
termined from the seismictravel time taken from the Apollo would have been smaller. For example, a core with a density
passiveseismicexperiment[Toks6z et al., 1974]. The mean of 8 gcm -3 can be accommodated
with a coreradiusof 364
densityand mean moment of inertia parametersact as overall km (•: -- 0.210)andp• -- 3.354gcm-3.
boundary conditionsfor the lunar model. We will now investigatethe effect of the error in the esti-
If we consider the moon to be composedof N concentric mate for I/MR: on the derived value for the core density.
layers,each of a differentcomposition,we then have the fol- Since the relative error for the mean density is a factor of 10
lowing two equationsrelating the internal structureto the esti- more preciselyknown than the relative error for the mean
matesof t5and I/MR: [Billsand Ferrari, 1977a]: moment parameter,we will assumethat the error in the den-
N
sityis negligiblein our calculations.If a linearizedvariationin
/• = po+ • Apk•3 (14) (14) and (15) is taken, the changein the coredensity,Ap:, is
k=l

I N
1GO
_ I I I I I I
• I5 •--Po+ •APk•k
5 (15) ½

where Pois the densityof the outermostlayer, Apkis the den-


sitycontrastbetweenthe outerlayer and the next layer, and •k
= Rk/R is the ratio of the radius of the •h layer to the mean EQUIVALENT GRAVITY
lunar radius. _

The resultsobtaMed M this analysisfor est•ates of GM,


C/MR •, •, and T, whenusedM conjunctionwith est•ates for
the univesalconstantG = 667.32• 0.31 x 10-• •3 g-• s-•
[Heyl and Chrzanowski,1942]and the mean lunar radiusR =
1737.53ß 0.03 • [Bills and Ferrari, 1977b],yield the follow-
Mg resultsfor the mean density:
_-
• = 3.3437• 0.0016gcm-3
Note that the formal error M the mean density is domMated
/ OBSERVED
by the error M G, whereasthe focal error M the mean mo- f•' GRAVITY
ment of Mertia (Table 6) is domMated by the error est•ate 1-- /GRAV
ITY
for C:o propagated through its •plicit constraMt equations _
/ ERRORS
/
(6), (7), and (8). /
0.5-
The nomMal valuesfor • and I/MR: can be usedto con- /
-/
stmct a s•ple model for the lunar Mterior. SeismicresultsM-
dicate that the moon has a crest, a multilayered mantle, and
possiblya small core [Nakamura et al., 1974; Toksb'zet al.,
1974].For the purposesof this analysiswe will assumea lunar
Mterior model consistMgof a uniform crest, mantle, and core. .1
2 3 4 5 6 7 8 9
Recent analysesof the global propertiesof the lunar gravity,
HARMONIC DEGREE
topography, and seismicdata Mdicate that the mean crustal
thicknessis approx•ately 70 • (• = 0.96) for an assumed Fig. 6. Standarddeviationper harmoniccoefficientversusdegree.
3950 FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM

related to the error in I/MR 2 by REFERENCES

Abbot, R. I., P. J. Shelus,J. D. Mulholland, and E. C. Silverberg,La-


ser observations of the moon: Identification and construction of
•(A•ø2)
•- 2•23[-•
--(•I/MR2)]
•-• (16) normal points for 1969-1971, Astron. J., 78, 784-793, 1973.
Akim, E. L., Determination of the moon's gravitational field from the
Using the current estimatefor the mean moment of inertia pa- motion of Luna 10 (in Russian), Dokl. Akad. Nauk SSSR, 170,
rameter and its error (0.3904 _ 0.0023), we computean uncer- 799-802, 1966.
tainty of 3.9 g cm-3 in the densityof a lunar corewhichis 300 Ananda, M., Lunar gravity: A point massmodel, J. Geophys.Res., 82,
3049-3065, 1977.
km in radius. This computation assumesthat the full impact Ananda, M.P., A. J. Ferrari, and W. L. Sjogren,An improvedlunar
of the error in I/MR • is appliedto the coredensitydetermina- moment of inertia: A proposedstrategy,Moon, 7, 101-120, 1977.
tion. If similar calculations are performed on a model with a Bender,P. L., D. G. Currie, R. H. Dicke, D. H. Eckhardt,J. E. Fall•r,
larger core radius, for example, one having a core 500 km in W. M. Kaula, J. D. Mulholland, H. H. Plotkin, S. K. Poultney, E.
C. Silverberg, D. T. Wilkinson, J. G. Williams, and C. O. Alley,
radius, the same set of errors and model parameterswould re-
The lunar laser ranging experiment, Science,182, 229-238, 1973.
sult in a core densityuncertaintyof 1 g cm-3. These calcu- Bills, B. G., Topographicvariance spectrafor the moon, Mars, Venus
lations demonstratethat if the moon does indeed possessa and the earth: An equilibrium model, in Abstractsof Lunar and
small core, •: < 0.173, and if the mean moment of inertia pa- Planetary ScienceIX, pp. 88-90, Lunar and Planetary Institute,
Houston, Tex., 1978.
rameter is to provide compositionalconstraints,its precision
Bills, B. G., and A. J. Ferrari, A lunar density model consistentwith
should be improved by an order of magnitude over the cur- topographic,gravitational,librational, and seismicdata, J. Geophys.
rent results. Res., 82, 1306-1314, 1977a.
Bills, B. G., and A. J. Ferrari, A harmonic analysisof lunar topogra-
phy, Icarus, 31, 244-259, 1977b.
CONCLUSIONS
Blackshear,W. T., and J.P. Gapcynski, An improved value of the lu-
nar moment of inertia, J. Geophys.Res., 82, 1699-1701, 1977.
This analysis has presented results of a direct solution for
Bryant, W. C., and R. G. Williamson, Lunar gravity anaJysisresults
the lunar moment of inertia. The results obtained are consis-
from Explorer 49, paper presented at Mechanics and Control of
tent with previous determinations made from different data Flight Conference,Amer. Inst. of Aeronaut. and Astronaut., Ana-
sets,thereby adding new confidenceto our knowledgeof the heim, Calif., 1974.
nominal value of C/MR 2 = 0.3905 q- 0.0023. These results in- Calame, O., Free librationsof the moon determinedby an analysisof
laser range measurements,Moon, 15, 343-352, 1976.
dicate that this error level is controlled by the lack of usable Calame, O., Free librations of the moon from lunar laser ranging,in
Doppler data. The shortageof maneuver-freearcsof Doppler ScientificApplicationsof Lunar Laser Ranging,edited by J. D. Mul-
tracking from satellite orbits which are far enough from the holland, pp. 53-63, D. Reidel, Hingham, Mass., 1977.
moon to attenuate the effectsof the high-degreeand high-or- Calame, O., and B. Guinot, Earth rotation by lunar distances
der coefficients makes the reduction of this current error level (EROLD), Annual Report for 1978, pp. D27-D48, Bur. Int. de
l'Heure, Paris, 1979.
a difficult problem to overcome.Processingof all the existing Cheng, C. H., and M. N. Toksfiz, Tidal stressesin the moon, J.
Doppler data in the eight data arcsused in this solution would Geophys.Res., 83, 845-853, 1978.
reduce the current formal error, since only one point in five Eckhardt,D. H., Physicallibrationsdue to the third and fourth degree
was used.Further improvementin the C/MR 2 error level can harmonicsof the lunar gravity potential, Moon, 6, 127-134, 1972.
Eroshkin, G. I., Influence of the dynamical figure of the moon on its
only be achievedwith the addition of relatively long Doppler rotational-translationalmotion, in The Stability of the Solar System
tracking arcsfrom other satellites,suchas Explorer 49. and of Small Stellar Systems,edited by Y. Kozai, pp. 201-207, D.
The current error level in the mean moment of inertia Reidel, Hingham, Mass., 1974.
places only a weak compositional constraint on lunar core Esposito,P. B., Evaluation of geocentricgravitationalconstant(ab-
models which are of radius 300 km or smaller. An improve- stract), Eos Trans.AGU, 59(4), 260, 1978.
Esposito,P. B., and A. T. Y. Ng, Geocentric gravitational constant
ment of approximately an order of magnitude in the uncer- determined from spacecraftradiometric data, Phys. Earth Planet.
tainty of C/MR 2 is requiredif it is to be usedas a composi- Interiors, 12, 238-289, 1976.
tional constraint for lunar core models with a radius of 300 Ferrari, A. J., Lunar gravity: A harmonic analysis,J. Geophys.Res.,
km. 82, 3065-3084, 1977.

Tables 5, 6, and 8-11 summarizethe final valuesof the pa- Gapcynski,J.P., W. T. Blackshear,
R. H. Tolson,and H. R. Comp-
ton, A determinationof the lunar moment of inertia, Geophys.Res.
rameters from the joint solution. In addition to the moments Lett., 2, 353-356, 1975.
of inertia of the moon there are estimates of the lunar GM, Harris, A. W., and J. G. Williams, Earth rotationstudyusinglunar la-
gravitationalharmonics,Love number,rotationaldissipation, ser ranging data, in ScientificApplicationsof Lunar Laser Ranging,
obliquity, and retroreflector coordinates.For the earth there edited by J. D. Mulholland, pp. 178-190, D. Reidel, Hingham,
Mass., 1977.
are values of GM, the tidal dissipation,and the McDonald Heyl, P. R., and P. Chrzanowski, A new determination of the constant
Observatorycoordinates.Since the accuratevalue of GME = of gravitation,J. Res.Nat. Bur. Stand., 29, 1-31, 1942.
398600.461q- 0.026 km3/s2 resultsfrom subtractingthe more Jackson,D. D., The use of a priori data to resolve non-uniquenessin
accurate value of GMM - 4902.7993 ___ 0.0029 km3/s2 from the linear inversion, Geophys.J. Roy. Astron. Soc., 57, 137-157, 1979.
total value, the accuracy of future lunar laser determinations Kaula, W. M., and P. A. Baxa, The physical librations of the moon,
including higher harmonic effects,Moon, 8, 287-307, 1973.
of GML.can improve nearly 1 order of magnitudebefore en- King, R. W., C. C. CounselmanIII, and I. I. Shapiro, Lunar dynam-
countering the limit set by the above determination of GMM. ics and selenodesy:Results from analysisof VLBI and laser data, J.
Geophys.Res., 81, 6251-6256, 1976.
Acknowledgments.We wish to thank R. N. Wimberly for his con- King, R. W., C. C. CounselmanIII, and I. I. Shapiro, Universal time:
tribution to the lunar orbiter data analysis and P. L. Bender for his Results from lunar laser ranging, J. Geophys.Res., 83, 3377-3381,
early encouragementof a joint analysisof the two data types.This pa- 1978.
per presentsthe resultsof one phase of researchcarried out at the Jet Lerch, F. J., R. E. Laubscher, S. M. Klosko, D. E. Smith, R. Kolen-
Propulsion Laboratory, California Institute of Technology, under kiewicz, B. H. Putney, J. G. Marsh, and J. E. Brownd, Determina-
contract NAS 7-100, sponsored by the National Aeronautics and tion of the geocentricgravitational constantfrom laser ranging on
Space Administration. near-earth satellites,Geophys.Res. Lett., 5, 1031-1034, 1978.
FERRARI ET AL.: GEOPHYSICAL PARAMETERS OF THE EARTH-MOON SYSTEM 3951

Liu, A. S., and P. A. Laing, Lunar gravity analysisfrom long-term ef- Stolz, A., P. L. Bender, J. E. Failer, E. C. Silverberg, J. D. Mulhol-
fects, Science,173, 1017-1020, 1971. land, P. J. Shelus, J. G. Williams, W. E. Carter, D. G. Currie, and
Lorell, J., Lunar orbiter gravity analysis,Moon, 1, 190-231, 1970. W. M. Kaula, Earth rotation measuredby lunar laser ranging, Sci-
Lorell, J., and W. L. Sjogren, Lunar gravity: Preliminary estimates ence, 193, 997-999, 1976.
from lunar orbiter, Science, 159, 625-628, 1968. Toksfiz, M. N., A.M. Dainty, S.C. Solomon,and K. R. Anderson,
Michael, W. H., Jr., and W. T. Blackshear, Recent results on the Structure of the moon, Rev. Geophys.Space Phys., 12, 539-567,
mass, gravitational field and moments of inertia of the moon, 1974.
Moon, 3, 388-402, 1972. Williams, J. G., Presentscientificachievementsfrom lunar laser rang-
Michael, W. H., Jr., W. T. Blackshear,and J.P. Gapcynski, Results ing, in ScientificApplicationsof Lunar Laser Ranging,edited by J.
on the massand gravitationalfield of the moon as determinedfrom D. Mulholland, pp. 37-50, D. Reidel, Hingham, Mass., 1977.
the dynamicsof lunar satellites,in Dynamicsof Lunar Satellites,pp. Williams, J. G., M. A. Slade, D. H. Eckhardt, and W. M. Kaula, Lu-
42-56, Springer,New York, 1969. nar physicallibrationslaserranging,Moon, 8, 469-483, 1973.
Moyer, T. D., Mathematicalformulationof the doubleprecisionorbit Williams, J. G., R. H. Dicke, P. L. Bender, C. O. Alley, W. E. Carter,
determination program, Rep. TR 37-1527, Jet Propul. Lab., Pasa- D. G. Currie, D. H. Eckhardt, J. E. Failer, W. M. Kaula, J. D. Mul-
dena, Calif., 1971. holland, H. H. Plotkin, S. K. Poultney, P. J. Shelus, E. C. Silver-
Mulholland, J. D., P. J. Shelus, and E. C. Silverberg, Laser observa- berg,W. S. Sinclair,M. A. Slade,and D. T. Wilkinson,New Test of
tions of the moon: Normal points for 1973, Astron. J., 80, 1087- the equivalenceprinciplefrom lunar laserranging,Phys.Rev. Lett.,
1093, 1975. 36, 551-554, 1976.
Nakamura, Y., G. Latham, D. Lammlein, M. Ewing, F. Duennebier, Williams, J. G., W. S. Sinclair, and C. F. Yoder, Tidal accelerationof
and J. Dorman, Deep lunar interior inferred from recent seismic the moon, Geophys.Res. Lett., 5, 943-946, 1978.
data, Geophys.Res. Lett., 1, 137-140, 1974. Yoder, C. F., Effectsof the spin-spininteractionand the inelastictidal
Shelus, P. J., J. D. Mulholland, and E. C. Silverberg, Laser observa- deformation on the lunar physical librations, in Natural and Arti-
tions of the moon: Normal points for 1972, Astron. J., 80, 154-161, ficial SatelliteMotion, editedby P. Nacozyand S. Ferrez-Mello, pp.
1975. 211-221, University of Texas Press,Houston, 1979.
Sinclair, W. S., A. J. Ferrari, W. L. Sjogren, and J. G. Williams, De- Yoder, C. F., W. S. Sinclair, and J. G. Williams, The effects of dis-
termination of lunar gravitational harmonic coefficientsfrom com- sipationin the moonon the lunar physicallibrations,Lunar Science
bined orbiter Doppler and laser ranging data, Bull. Amer. Astron. IX, p. 1292, Lunar and Planetary Institute, Austin, Tex., 1978.
Soc., 9, 437, 1977.
Sjogren,W. L., Lunar gravity estimate:Independentconfirmation,J.
Geophys.Res., 76, 7021-7026, 1971. (Received August 3, 1979;
Sjogren,W. L., and W. R. Wollenhaupt, Lunar shapevia the Apollo revised February 21, 1980;
laser altimeter, Science, 179, 275-278, 1973. acceptedMarch 7, 1980.)

You might also like