You are on page 1of 12

CHINESE JOURNAL OF PHYSICS

VOL. 50, NO. 5

October 2012

The Quasi-Elliptic Motion of the Moon Maurizio M. DEliseo


Osservatorio S. Elmo-Via A. Caccavello 22, 80129 Napoli, Italy (Received January 12, 2012) Among the many perturbing eects exerted by the Sun on what would otherwise be the Keplers ellipse of the Moon around the Earth, it is possible to set apart a particular class of terms of periodic and secular nature related to the size of the eccentricity and to the location of the perigee. These matters can be jointly treated by studying the behavior of a single mathematical entity, the eccentricity vector, a complex number that in polar form has the eccentricity as magnitude and the angle marking the position of perigee as argument. The comparison between the theoretical behavior of this vector and the observations has historically represented an important test of the inverse-square law of gravitation.
PACS numbers: 96.20.-n, 45.10.Hj, 96.12.De

I. INTRODUCTION

The solar perturbation makes the Moons motion relative to the Earth not amenable to a simple geometric and kinematic description, but not so much that the basic model of an elliptic orbit need be abandoned altogether. It turns out that, as a rst approximation, one may keep Keplers picture of the motion with the changes that are reected in two statements. First, the apsidal line (joining apogee and perigee) of the Keplerian ellipse rotates slowly in the same direction as the Moon itself making one complete revolution in about nine years, and to this motion is superimposed a semiannual oscillation. Second, the ellipse becomes pulsating, that is its shape is periodically more or less attened because the eccentricity oscillates around a mean value, with the same period and a phase shift of /2 with respect to that of the apsidal line. Both occurrences are expected from mathematical analysis and conrmed by observation (historically, this path has been followed in reverse order). Qualitatively, the direct rotation of the apsidal line can be inferred from a careful reasoning on the imbalance of the eects of the solar perturbation in some specic geometric congurations of the Sun and Moon in the course of time, but we omit the details for brevity [1]. With regard to the variation of the eccentricity, one can say that when the apsidal line is parallel with the direction toward the Sun (that is when the Moon is full at perigee and new at apogee, or vice versa) the attraction of the Sun tends to elongate the ellipse of the relative lunar motion and increase its eccentricity. If, on the other hand, the apsidal line is perpendicular to the

Electronic address: s.elmo@mail.com

http://PSROC.phys.ntu.edu.tw/cjp

720

c 2012 THE PHYSICAL SOCIETY OF THE REPUBLIC OF CHINA

VOL. 50

MAURIZIO M. DELISEO

721

line joining the positions of the full and new Moons, the solar attraction will tend again to widen the ellipse and diminish its eccentricity. In particular, for a terrestrial observer, the variation of the eccentricity is reected in the periodic perturbation known as the evection, a gain and a loss in the angular position of the Moon compared with that provided by a pure Kepler motion, with a period of about 32 days. All these eects are entangled with each other in the behavior of the eccentricity vector e, a complex number whose polar form has the eccentricity as its magnitude and the angle marking the position of perigee (and therefore of the apsidal line) as its argument. If there were no attraction of the Sun on the Moon this vector would be constant, implying the immutability of the ellipse, but actually the vector e is a dynamical variable that changes with time. A careful inspection of the function e(t) allows us to quickly isolate the expression of the position vector of the Moon that incorporates all the mentioned occurrences and make the orbit in this approximation as quasi-elliptic, meaning by this that part of the actual lunar motion described in terms of an elliptic orbit of which some parameters slowly change with time and are mathematically represented at the lowest order of their variability. In the past, the study of this matter has been instrumental in reconciling observation and theory in support of the inverse-square law of gravitation [2].

II. THE ECCENTRICITY VECTOR

We dene rst the eccentricity vector for an Earth-Moon system thought of as being free from any outside inuence. We x a geocentric coordinate system. The position of the Moon in the orbital plane that we mathematize as the complex plane, is given by the vector r . Then r2 = |r |2 = rr , r being the complex conjugate of r , while the real and the imaginary parts of a complex quantity, such as r are Re (r ) = (r + r )/2 and Im (r ) = i(r r )/2, respectively. In polar coordinates, r () = r() exp(i), where the polar angle is the true longitude measured from a suitably chosen oriented axis. The function r (t) = r [(t)] will be known as soon as we nd the time dependence of , and its time derivative is ) ( r r. = r + i (1) r Let E be the mass of the Earth, M that of the Moon, G the gravitational constant. These two bodies for all practical purposes can be supposed to be punctiform. At a given time, let the Earth be located at the origin of the coordinates, and let r be the geocentric vector of the Moon. The force per unit mass by which the Moon is attracted toward the Earth and the force per unit mass by which the Earth is attracted toward the Moon are, respectively, GE r , r3 GM r . r3 (2)

In order that the Earth stays at the origin even at later times, it should be put to rest with respect to the Moon. This can be formally achieved by submitting the Moon at any time

722

THE QUASI-ELLIPTIC MOTION . . .

VOL. 50

F
= r r , r3

GM r r3

r GE r3    +

3  

+M ) r G(Er 3     +

to a force per unit mass equal and opposite to that by which it attracts the Earth. On this assumption one can write the equation of the geocentric motion of the Moon as G(E + M ). (3)

We multiply Eq. (3) by r , and take the imaginary part [3]. Since Im ( r r ) = d ), Im (rr dt Im (rr ) = 0, r3 (4)

we get, upon integration, the angular momentum integral of the two-body motion: ) = . Im (rr With the help of Eq. (1), we immediately deduce rst the polar expression of : =2 ) = r2 = Im (rr dAr = const., dt (6) (5)

so that is twice the constant areal velocity, Ar being the area swept out by r , and furthermore the formal operation d d = d, = dt d r2 d from which it follows, in particular, that dr d i iei r = 2 e = 2 = i 3 . dt r r d r r Using this identity, Eq. (3) may be written in the form ( ) d i r + 0= r , dt r which, upon integration, becomes the Hermann-Laplace integral [4]: i r , e= r r (10) (8) (7)

(9)

VOL. 50

MAURIZIO M. DELISEO

723

where e e exp(i ) is a complex constant, the eccentricity vector. Now we multiply both sides of Eq. (10) by r and take the real part. We nd, with simple steps, Re (rr ) ) Re (er ) = Re (irr , r Re (e r ) = ) r, Im (rr 2 . (11)

(12)

r [1 + e cos( )] =

(13)

Solving for r and multiplying by exp(i), we get the polar expression of the lunar vector: r = rei = ei 2 . [1 + e cos( )] (14)

Comparing this equation with the representation of an ellipse in polar coordinates, we deduce that the eccentricity vector e = e exp(i ) has the eccentricity as magnitude and the angle marking the position of perigee (where r = rmin ) as argument; besides, we obtain an expression of in terms of two orbital elements: 2 = a(1 e2 ) = a(1 e2 ). (15)

By denoting with T the orbital period and with n the real number such that T = 2/n, we obtain, from Eq. (6), T 2/n, (16) dt = 2Ar , 0 2a2 1 e2 , by the well-known formula for the area of the ellipse, from which it follows that = na2 1 e2 ,

(17)

that conrms and claries the meaning of relationship (15). Notice that = na2 to order 3 e, and further, by Eqs. (15), (17), that n = /a , so that n is Keplers mean motion of the Moon. Since e is already present in e, actually contains and controls also another constant of motion, the semi-major axis a. We can nd very simply the function r (t), the solution of Eq. (3), to order e [5]. From Eqs. (14), (15), (17), we have r = rei a[1 Re (e ei )] ei 1 1 = aei ae ae ei2 , 2 2

(18)

724

THE QUASI-ELLIPTIC MOTION . . .

VOL. 50

2 = na n + 2en cos( ). r2 r2 Integrating the last expression we get

(19)

nt + + 2e sin(nt + ) [ ] = nt + i e ei(nt+) eei(nt+) ,

(20) (21)

where the integration constant , called the mean longitude at epoch, is the element that species the position along the orbit at the initial time, and is therefore an arbitrary constant (the fourth, after a, e, ) of the elliptic motion. The sine term in the right-hand side of Eq. (20), named the equation of center, represents the gain or loss in longitude due to the eccentricity compared with the uniform circular motion described by the secular term nt. By the way, from this observed inequality one deduces the eccentricity of the lunar orbit, which is e 0.0549. From Eq. (21) it is easily seen, by developing in series the complex exponentials in , that in Eq. (18) one can assume ei e + ei(nt+) + e ei2(nt+) , ei2 ei2(nt+) , so that we nd 3 1 r = aeil ae + ae ei2l , (24) 2 2 where we have introduced the mean longitude of the Moon l nt + . The lunar vector r is thus expressed to order e as the composition of a circular motion and of two terms containing the eccentricity vector, one constant and the other periodic of period one-half compared to that of the circular motion and with a phase-shift determined by the position of perigee. The check of this solution is immediate, because we have ( ) = an2 eil + 2e ei2l , r r n2 a 3 r = r3 r3 2 n [1 + 3e cos(l )] r [ ] 3 3 = n2 1 + e eil + eeil r 2 2 ( ) an2 eil + 2e ei2l . (22) (23)

The knowledge of the unperturbed position vector r (t) of Eq. (24) is all that one needs to nd the main perturbations of the eccentricity vector due to the Sun. In particular, it allows the approximate symbolic computation of the secular motion of the lunar perigee for that part (by far the greatest) dependent only on m = n /n 0.0748013, the ratio of the Suns sidereal mean motion to the Moons sidereal mean motion (n = 0.9856090 /day, n = 13.1763631 /day, respectively).

VOL. 50

MAURIZIO M. DELISEO

725

III. THE TIME VARIATION OF THE ECCENTRICITY VECTOR

Assume, in full generality, that the elliptic motion of the Moon around the Earth is perturbed by any cause whatsoever, so that the equation of motion (3) takes the form =R+P r r + P, r3 (25)

where, for brevity, R stands for the inverse-square centripetal acceleration and P is a perturbing acceleration. In this situation the eccentricity vector is no longer a constant of motion. In order to nd the time variation of e, we start from Eq. (10) written in the form r i )r , e = Im (rr r and consider its time derivative i dr = [Im ( + Im (rr ) e r r )r r ] . dt r = R + P will produce the following expression The substitution, in Eq. (27), of r i dr = [Im (rr )R + Im (P r )r + Im (rr )P ] e . dt r (28) (27) (26)

, r , r , that are Equation (28) must be solved perturbatively, because the functions r , r present in its right-hand side refer to the actual, and still unknown, orbit of the Moon. So, in order to write the rst approximation of Eq. (28), sucient for our purposes, we feed the right-hand side of the equation with the values derived from the two-body solution. That we can stop at the rst approximation is benecial, because we must not attempt to nd the equations of perturbation related to the (former) constants a and , which are ,r , in order to carry necessary for the computation of the next values of the quartet r , r , r out the second approximation to achieve a greater precision. With this understanding, using Eq. (8) to cancel the term containing R, we obtain the perturbation equation for e: = e d (eei ) = ( e/e + i )e dt i + P ] , = [Im (P r )r

(29)

from which we immediately deduce the corresponding equations for e and : /e) , e = Re (e e /e) . = Im (e (30) (31)

726

THE QUASI-ELLIPTIC MOTION . . .

VOL. 50

IV. THE SOLAR PERTURBING ACCELERATION

Now we deal with the expression of the perturbing acceleration of the Sun in the actual motion of the Moon around the Earth. In order to simplify the calculations as much as possible, we suppress all the nonessential details and emphasize the core of the problem. We assume that the lunar motion takes place in the ecliptic (the plane in which the Earth revolves around the Sun) and that the relative motion of the Sun around the Earth is circular and uniform, so neglecting the inclination of the Moons orbit to the ecliptic and the eccentricity of the orbit of the Earth. This is theoretically possible because within our approximations the motion of the Moon projected on the ecliptic is decoupled from the motion perpendicular to it. Let S be the mass of the Sun and r its geocentric vector. As the Sun attracts both the Moon and the Earth, it is clear that in the geocentric system one must take into account the fact that on the Moon, other than the Earths attraction, will act only the dierence between the Suns accelerative attraction on the Moon and the Suns attraction on the Earth: GS (r r ) GS r 3 , | r r |3 r so that the geocentric equation of motion of the Moon is ) ( r r r r r = 3 + r 3 3 + P, 3 r |r r | r r (32)

(33)

where = GS , the standard gravitational parameter of the Sun, while in the right side are now present the inverse-square and the perturbing accelerations. Ultimately, P represents the tidal acceleration of the Moon caused by the Sun. Note that we have implicitly set S + E + M = S , that is we have assumed that the masses of Earth and Moon are negligible with respect to that of the Sun. The Sun is about 330,000 times more massive than the Earth, while its distance is about 400 times the distance of the Moon. It follows that the maximum tidal force is about 70 times less than the inverse-square force. Since the motion of the Sun relative to the Earth is assumed to be circular and uniform, we can write r = r exp[i(n t + )] where r = const. The constant n = /r3 is the sidereal mean motion of the Sun, while the phase is its longitude at t = 0. If we leave out the second and higher powers of r/r 1/400 and denote with l n t + the solar mean longitude, the perturbing acceleration P becomes ( ) r P = n2 (r r ) 1 + 3 cos(l ) + . . . n2 r r 2 2 r n r + 3 n r cos(l ) r ) 3 2 il ( i(l) 2 i(l ) = n r + n e r e +e 2 3 1 (34) = n2 r + n2 r ei2l , 2 2

VOL. 50

MAURIZIO M. DELISEO

727

and Eq. (33) takes the form = r r 1 3 + n2 r + n2 r ei2l . 3 r 2 2 (35)

It is worth noting that on the right-hand side the ratio r/r has disappeared, and with it any reference to the nite distance of the Sun (the solar parallax), and we are left with only two solar terms, the rst of which is a repulsive force between the Earth and the Moon that depends linearly on the Earth-Moon distance and therefore is of central type, representing an average solar perturbation.

V. THE MAIN PERTURBATIONS OF THE ECCENTRICITY VECTOR

It is useful to write Eq. (35) in the form =R+P = r r 1 3 + m2 n2 r + m2 n2 r ei2l , 3 r 2 2


Pr Pr

(36)

where we have indicated that the solar perturbation P is the sum two parts, Pr , Pr , of which the rst is a subtractive (i.e., directed away from the central body) radial acceleration. It easy to verify that P is a perturbation of the two-body motion. As r is proportional to a, the lunar vector r in the rst term on the right of Eq. (36) has the prefactor /a3 = n2 , and therefore the ratio between P and the inverse-square acceleration has m2 0.005595 as a factor, thus justifying a perturbative approach to the problem. To go on, we explicitly put in the right side of Eq. (29), ) ( 1 i2l 3 il , (37) r =a e e+ e e 2 2 ( ) = an ieil + ie ei2l , r and = na2 , since we work to rst-order in e, so that = Im (Pr r )r Im (P r )r 3 2 3 3 , = m n a Im (r 2 ei2l )r 2 ( P = m n a
2 3 3

(38)

(39)

) 1 3 i2l r+ r e , 2 2

(40)

where in Eq. (39) we have taken advantage of the fact that Im (Pr r ) = 0. Since the products in the right members of Eqs. (39), (40), have the same symbolic coecients, from now on,

728

THE QUASI-ELLIPTIC MOTION . . .

VOL. 50

, , P to be multiplied to facilitate the calculations, we shall write the expressions of r , r together in a normalized form, without the presence of a, n, m, which will appear once and for all together as prefactor in the perturbation equation. This way Eq. (29) will assume the form (with = n2 a3 ) = e m2 n3 a 3 i + P ] [Im (Pr r )r n2 a 3 + P ] . = m2 ni [Im (Pr r )r

(41)

Once one knows the right-hand side of Eq. (41), which is the simplest dierential equation of the normal type, the solution of the equation in the rst approximation is given by the indenite integral + P ] dt, (42) e(t) = m2 ni [Im (Pr r )r where, in the integrand, e, e are treated as constants. Let us look rst at the possible form of the expressions constituting the function e(t). The term by term integration of the periodic terms resulting from the products indicated in Eq. (42) leads evidently to complex exponentials of the form Cm2 n i(hl +kl) e , (43) Cm2 ni ei(hl +kl) dt = hn + kn where h, k are integers not both zero, and where the constant C is real or complex. In addition, non-periodic terms will also be produced, which are evidently those that cause secular eects. Since we want to conne our study only to some specic aspects of the lunar motion, the greatest part of the terms (some tens) that constitute the complete expression are not of our concern here, so they must be discarded by an opportune selective action of e to signicantly reduce their number. A rst criterion is that they contain the rst power of e, e . This is required by the structure of the perturbation equations (30), (31), and is consistent with the rst-order approximation of our work. Among the constant terms, only one has this feature. With regard to the choice of the periodic terms, it is worth noting that letting k = 0 in Eq. (43) we obtain a term modulated by the position of the Sun alone for which n C C Cm2 ni eihl dt = m2 eihl = meihl , (44) h n h and so the integration lowers in the coecient the power from m2 to m, enhancing its greatness. So we choose precisely this term that is by far the largest among those periodic. In this way we get the dierential equation governing the time behavior of the eccentricity vector: 3 15 (t) = m2 nie + m2 nie ei2l . e 4 4 (45)

VOL. 50

MAURIZIO M. DELISEO

729

This equation can be solved by an iterative perturbation procedure. As said before, we begin considering e = e exp(i ) (and so e ) in the right side as constant, with the value of e opportunely xed and that of as referred at the instant of time assumed as initial. This is emphasized by the particular notation we have used, by distinguishing between e(t) and e. Thus integrating Eq. (45) we obtain 3 15 e(t) = m2 niet + me ei2l + e, 4 8 (46)

3 15 e (t) = m2 nie t + me ei2l + e . (47) 4 8 The non-secular part of these expressions shows that the eccentricity vector is subject to a periodic variation with a frequency of n times per interval of time. Inserting this solution back in the right-hand side of Eq. (45) and taking into account the terms up to order m3 , we obtain ( ) 15 2 15 15 3 2 i2l (t) = m nie + m ni mee ei2l + m2 nie ei2l , (48) e 4 4 8 4 (t), and so Eq. (45) must be replaced by which modies the previous evaluation of e ( ) 3 225 15 (t) = e + m m2 nie + m2 nie ei2l . 4 32 4

(49)

It is worth noting that the coecient of the periodic term is unchanged. Now we can use Eqs. (30), (31), to obtain the functions e(t) and (t). We have (t)/e] dt e(t) = Re [ee ( ) 15 2 = m ne Re iei(2l 2) dt 4 15 = me cos(2l 2 ) + e, (50) 8 where the tilde over in the result reminds us of having to take into account the linear growth of the argument of perigee with time, according to the outcome of the integration of performed below. We nd that the lunar eccentricity oscillates periodically within the limits ( ) ( ) 15 1 e 1 m e 1 , (51) 8 7 about its mean value, which can be consistently interpreted as that empirically deduced from the equation of center. Further we have ( ) ( ) 3 2 225 3 15 (t)/ e ] dt = (t) = Im [ e m + m nt + m2 n Im iei(2l 2) dt 4 32 4 ( ) 3 2 225 3 15 = m + m nt + m sin(2l 2 ) + . (52) 4 32 8

730

THE QUASI-ELLIPTIC MOTION . . .

VOL. 50

VI. THE QUASI-ELLIPTIC ORBIT

The dynamical variables e(t) and (t) we have worked display a constant rotation of perigee of the Moon (i.e., the orientation of the ellipse) in the direction of an increasing , as well as its semiannual libration, with an accompanying pulsation in the size of the Moons orbital eccentricity, to which we alluded at the beginning. The m3 term in (t) is that found for the rst time after a very exacting work by dAlembert [6]. Together, the two secular terms represent approximately ve-sixths of the observed motion, and the perigee makes a full turn in the forward direction in T = 2 2 3, 827 n [(3/4)m2 + (225/32)m3 ]n (53)

days or about 10.47 years, 1.62 years o from the observed period of 8.85 years, not bad (t) in Eq. (45) considered the approximations made. Notice that the constant term of e originates only from the radial part of the perturbing acceleration, so the result (3/4)m2 nt provided by the rst lunar theorists (Newton, Clairaut, dAlembert, and Euler) shows that the combination of the inverse-square centripetal acceleration and the subtractive radial component of the perturbative acceleration yields a forward motion only half that observed. This is because the ratio between the second and the rst term of the secular part of (t) is, for the Moon, 7 75 m , 8 10 (54)

and this means that for the Moon the m3 term gives a contribution comparable to that of the m2 term, something totally unexpected, with the result that the rotation period of the lunar perigee is nearly halved, making it close to that observed. This explains the inability of the rst lunar theorists, people who grew up in the spirit of innitesimals, to predict such an occurrence before undertaking a more complete and detailed calculation. The relationship in Eq. (54) is small only when m is very small, at least an order of greatness lesser than it is for the Moon. For example, it is nearly negligible in the theory of the Jupiters four Galileian satellites, where 4 104 < m < 4 103 . The secular term, whose nature we have claried, is easily managed by considering it implicitly attached to the element , so that it can be formally discarded when considering the correction to the position vector of the Moon due to the variation of e, by agreeing that now e = e exp(i ), with + n t. Then 3 1 r = aeil ae(t) + ae (t)ei2l , 2 2 (55)

with e(t), e (t) given by the two last elements of Eqs. (46), (47) only. This way the quasielliptic position vector r q of the Moon that jointly describes the rotating ellipse, and the oscillations of perigee and eccentricity is found to be 3 1 45 15 r q = aeil ae + ae ei2l ame ei2l + ame ei(2l2l ) . 2 2 16 16 (56)

VOL. 50

MAURIZIO M. DELISEO

731

It should be noted that, taken by itself, r q is not an approximate solution of Eq. (35), as indeed can be inferred from the procedure we have followed, but it is rather only a signicant part of the complete rst-order solution of the equation. The knowledge of r q rq exp(iq ) allows the derivation of the correspondent polar coordinates rq , q of the Moon. The radial coordinate rq is given by rqr rq 15 q 1 e cos(l ) me cos(l 2l + = ), (57) a a2 8 We have added to the elliptic term a periodic variation of the length of the radius vector with an amplitude of about 2,960 km. For the polar angle q we get ( ) ( il il ) rq e rqe q = i ln = i ln rq rq [ (r ) rq ] q il = i il + ln e ln a a 15 ). (58) l + 2e sin(l ) + me sin(l 2l + 4 For both rq , q , the periodic m-part has a period of about 31.8 days. We have thus veried that in the angular motion of the Moon around the Earth the two largest deviations from the uniformity, depending on the rst power of the eccentricity, are in the order the equation of center and the evection (the rst has an amplitude of about 6.2815 , the second of about 1.2759 ). As regards the second approximation to the secular motion of lunar perigee in the literal form given by dAlembert, it has been interesting to see how, avoiding the detailed calculations the early lunar theorists had to run to obtain it for the rst time, the concept of eccentricity vector provides an easy way to nd a term that has historically represented an important support to Newtons law of gravitation.

References
[1] F. R. Moulton, An Introduction to Celestial Mechanics, 2nd Ed., Art. 197, p. 352, (Dover Publications, 1984). [2] M. C. Gutzwiller, Rev. Mod. Phys. 70, 589 (1998). [3] M. M. DEliseo, Am. J. Phys. 75, 352 (2007). [4] M. Nauemberg, Archive for History of Exact Sciences, Springer, Vol. 64, Number 3, 281 (2010). [5] M. M. DEliseo, J. Math. Phys. 50, Issue 2, 022901 (2009). [6] J. dAlembert, Recherches sur Dierents Points du Systeme du Monde Vol. 1, Art 66, 70 (1754).

You might also like