Computation of Planetary Orbits

Author(s): Donald A. Teets and Karen Whitehead

Source: The College Mathematics Journal, Vol. 29, No. 5 (Nov., 1998), pp. 397-404
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2687254 .
Accessed: 11/07/2014 04:53
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to
The College Mathematics Journal.

http://www.jstor.org

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

of Planetary
Orbits
Computation
Donald A. Teets and Karen Whitehead
Donald Teets ([email protected])
received his
B.A. from the University of Colorado, his M.S. from
Colorado State University,and his Doctor of Arts from Idaho
State University. He has taught at the South Dakota School
of Mines and Technology since 1988 and has been chair of
the Department of Mathematics and Computer Science
since 1997. When he is not doing mathematics, he enjoys
backpacking, cross-country skiing, and rock climbing.
Karen Whitehead ([email protected]) received her
B.A. and Ph.D. from the University of Minnesota. She joined
the faculty at the South Dakota School of Mines and
Technology in 1981, serving as department head and
college dean before taking on her current duties as Vice
President for Academic Affairs. Her major avocation is
music: she is a substitute church organist and has sung in
student choral groups on campus for the past 14 years.

It is uncommon to find in the histoiy of science and mathematics great problems

whose solutions are accessible to first-or second-year undergraduates. The descrip?
tion and computation of planetary orbits provides a wonderful opportunity to study
just such a problem. Although techniques for computing orbits are certainly well
known and can be found in many textbooks on celestial mechanics [2, 5-8], we of?
fer here a concise introduction to the subject intended for students with only a strong
background in the calculus and analytic geometry. No background in astronomy is
assumed. The mathematical tools encountered in this problem include vector cross
products, dot products, equations of lines and planes in space, basic geometric and
trigonometric relationships, the trapezoidal rule, the polar coordinate equation of an
ellipse, area in polar coordinates, and a few other ideas that students should find
familiar. What a treat to find an application involving such a wide variety of topics
from a standard course in calculus and analytic geometry!
Our mathematical presentation has two limitations that we are aware of. First, our
goal is to present a method that is accessible to students with a modest background,
rather than a method for experts that will produce highly accurate results. The second
flaw is more fundamental: We assume that the planet's position with respect to
the sun is completely known at two different times. In practice, the problem of

obtaining two such position vectors from earth-based observations is substantial?it

was Gauss's solution to this problem (and the subsequent orbit computation for the
minor planet Ceres) that "made Gauss a European celebrity" in 1802 [1],
We state the problem as follows:
Given two vectors in M3 describing the positions ofa planet with respect to the sun
at two different times, compute parameters necessary to determine completely the
planet's position at any specified time.
We accept as given Kepler's three laws of planetary motion: a planet's orbit is an
ellipse with the sun at one focus; the planet's speed changes along the orbit in such
VOL. 29, NO. 5, NOVEMBER 1998

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

397

a way that the radial vector from the sun to the planet sweeps out area at a constant
rate; and the ratio of the square of the orbital period to the cube of the length of the
semimajor axis is the same for all planets in our solar system. Accessible derivations
of these laws can be found in [6]. Our problem is to show how the shape of the
elliptical orbit, its orientation in space, and the motion of the planet along the orbit
can all be determined from the two given radial vectors.
To develop the basic terminology of planetary orbits, our first
Basic terminology.
task is to establish a coordinate system in space, as illustrated in Figure 1.

/* First day
3L of spring

First day^\
of summer\

/sun
/

Figure 1

We shall use a rectangular coordinate system with the sun at the origin and with the
XF-plane chosen to be the plane ofthe earth's orbit, the so-called ecliptic plane. The
line through the earth and sun on the firstday of spring (in the northern hemisphere)
as the X-axis, and the F-axis is, of course, chosen perpendicular to the
axes are directed so that the sun, as viewed from earth, appears in
the positive X direction on the firstday of spring and in the positive Y direction as
is chosen
X-axis.

These

summer begins. The Z-axis is then chosen to form a right-handed coordinate system.
The angle i between the positive Z-axis and the vector n normal to the planet's
orbital plane (Figure 2) is called the inclination of the orbit. (Angles are measured

in degrees unless noted otherwise.) We shall assume that 0? < i < 90?; the actual
values of i for planets within our solar system range from 0? to about 17?. The planet's
orbital plane and the ecliptic plane intersect in the line of nodes, and, assuming that
the direction of motion is as indicated by the arrow, the point N on this line is
known as the ascending node. The angle DJ measured from the positive X-axis
to the line of nodes is the longitude ofthe ascending node. Next,
the angle u between the line of nodes and the major axis of the planet's elliptical
orbit is the argument of perihelion, which determines the orientation of the ellipse
counterclockwise

within the orbital plane. Note that perihelion is the point on the orbit closest to the
sun. The ellipse itself is determined by a, the length of its semimajor axis, and by e, its
eccentricity. Finally, the position of the planet on this elliptical orbit is determined (via
Kepler's second and third laws) by to, the time of perihelion passage. Collectively,
the six quantities i, Q, u, a, e, and t0 are called the elements of the orbit. Minor
variations exist in the choice of parameters that make up the set of orbital elements.
See [7] for a more complete discussion.
398

THE COLLEGE MATHEMATICSJOURNAL

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

Planet's orbital plane

Y
Ecliptic plane

Figure 2

Computation

of the Orbital

Elements

We now turn to the problem of computing the six elements of a planetary orbit from
the planet's positions r = (x, y, z) and r' = (xf, ?/, zf) at times t and t', respectively,
with t < t'.
i and ft determine the orbital plane.
To begin, we compute the
= r x r', and we observe in Figure 2 that
normal vector n = (77,1,77,2,77.3)

The elements

cost

(1)

from which the inclination i is easily obtained. (It is unambiguous

since we are
<
<
0?
i
90?.)
assuming
The angle Q is nearly as straightforward. Figure 3 shows (77,1,77,2),the projection
of n onto the XY-plane.
The vector pointing to the ascending node Af always lies
90? counterclockwise from (77,1,77,2),
as shown. If n\ > 0, as in Figure 3 (left), then
n

tan-i!^

+ 90o.
77.1

(2)

<nl>?2>
>x

>X

nl5n2)

Figure 3
VOL 29, NO. 5, NOVEMBER 1998

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

399

On the other hand, if n\ < 0, as in Figure 3 (right), we must add 180? to the result of
the inverse tangent computation to place the vector (ni, 712) in the proper quadrant.
In this case, we have Q = tan-1 (712/711)+ 270?.
The elements a and e determine the shape ofthe ellipse.
Having determined
the orbital plane, we turn to the problem of computing the parameters a and e,
which define the shape of the elliptical orbit within that plane. The polar equation
for this ellipse is
fc

(3)

1 + e cos /'
where
fc= a(l-e2),

(4)

and / is the true anomaly. the polar angle measured from the point of perihelion
on the major axis (see Figure 2). As before, we use / and f to denote the true
anomalies at times t and t', respectively (Figure 4).

Figure 4

A few more standard facts about ellipses will prove useful to us. If the length
of the semiminor axis is denoted by b and the center-to-focus distance by c, then
b = y/a2 ? c2, c = ae and the area of the ellipse is irab. It is straightforward to
combine these facts to obtain the alternative formulas KO?l2\fk and 7ra2\/l ? e2
for the area of the ellipse. The preceding facts about ellipses, including the polar
representation, can be found in most traditional calculus texts; for example, see [3].
Our strategy for determining a and e has three steps:
1. determine a value for fc;
2. use two different r values in (3) to produce two equations
e and /; and then
3. determine a from (4).

in the two unknowns

We compute fcusing one of the early methods given by Gauss himself [4]. Accord?
ing to Kepler's Second Law, the vector from the sun to a planet sweeps out area at a
constant rate. So if g denotes the area of the elliptical sector determined by the two

400

THE COLLEGE MATHEMATICSJOURNAL

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

vectors r and r' (Figure 4), At ? t' ? t denotes the elapsed time (in days) between
observations, and Tp denotes the period of the planet's orbit, Kepler's Second Law
gives us
no?l2\fk

g
At

(5)

Tp

Let the period and semimajor axis for the earth's orbit be Te and A, respectively.
Then Kepler's Third Law tells us that
rp2
Hr
a3

rp2
(6)

^tA3

Using Te = 365.25 (days) and A = 1 (astronomical

(5) and (6) to get

unit), we can combine equations

Thus, to find k, we need only determine the area g from the vectors r and r' and
their magnitudes r and r'. Figure 4 (right) indicates how to approximate g from the
standard polar area integral using the trapezoidal rule:
(8)

\f'?tf?~\{f-i??-.
(Here the angles / and /' must be expressed
?
?
/<
f
f between r and rf satisfies 0 </'
//-/

in radians!) Assuming that the angle

tt, we have

(9)

^1(^)'

Thus k can be approximated directly from the vectors r, r', and the elapsed
by using equations (9), (8), and (7), in that order.
Once k is known, (3) gives us the values
k
e cos / =-1
r

and

k
e cos /' = ? ? 1,
r

time At

(10)

which in turn allow us to compute the ratio cos f /cos f. But

cos/' = cos(/+
/'-/)
cos /
cos /
= cosfcoS(f-f)-Smfsm(f-f)
cos/

(11)

= cos(//-/)-tan/sin(/'-/).
value oi f ? f from (9), we
/ in the proper quadrant! For
thus / is in the first or fourth
second or third quadrant.
is
e
from
it
trivial
to
and
then a is equally simple from
obtain
/,
(10),

Using the computed value of cos ff/cos f and the

easily solve equation (11) for /. Take care to place
if (k/r) ? 1 > 0, then (10) implies that cos / > 0;
? 1 <
0, then / is in the
quadrant. Likewise, if (k/r)
Knowing
equation

(4).

VOL. 29, NO. 5, NOVEMBER 1998

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

401

In Figure 2, we saw
The element uj orients the ellipse within the orbital plane.
that the angle between r and the vector (cosfi,sinfi,0)
pointing to the ascending
node is / + uj. Thus
r ? (cos Q, sin?2,0)

= r cos(/ + uj).

(12)

If the ^-coordinate of r is positive, then 0? < / + uj < 180?, and solving equation
(12) for uj is straightforward. But if z is negative, then 180? < / + uj < 360?, so
? cos-1
/ + uj = 360?
(cos(/ + uj)), from which we easily compute uj.
the position in the orbit as a function of time.
t0 determines
the
time
of
Computing to,
perihelion passage, while not difficult, requires two for?
mulas whose derivations are somewhat involved. We will present the formulas here
for use in the next section, then discuss their derivations later.
In Figure 5, a circle of radius a is circumscribed around the elliptical orbit with
semimajor axis of length a; their centers coincide. Whereas / is the true anomaly

The element

at time t, the angle E is known as the eccentric anomaly.

related by
tan ( ?- ) = ( ?r?
1

These

components

are

1/2
I

tan [ ^

) .

(13)

The eccentric anomaly E is in turn related to t0 by Kepler's equation:

E ? e sin E ??

2tt ,
-(t-to).

(14)

(Here E must be in radians.) Thus we obtain to by using (6) to compute Tp from

a; using (13) to compute E from /; and then using (14) to obtain t0. The orbital
elements are now complete.

The position on the orbit at any time can now be determined by computing the
eccentric anomaly E from (14), then the true anomaly / from (13).

Figure 5
402

THE COLLEGE MATHEMATICSJOURNAL

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

1. To illustrate the technique, let's compute the orbital elements of Mars.

Example
All the data in this section were obtained from [51.
We start with the position vectors r =
and r' =
(0.8118,1.2476,0.0062)
=
to
the
times
t
1996
(0 hours) and
July 4,
corresponding
(0.6485,1.3599,0.0125)
t' = July 18, 1996 (0 hours). Here are the steps in the solution:
and ||n||
1. Compute n = r x r' = (0.007164,-0.006127,0.294898)
Use (1) to obtain i = 1.83? and (2) to obtain Q = 49.46?.

= 0.295048.

= 0.131946 radians. From (8) compute g = 0.147964,

2. From (9), compute f'-f
and then from (7) obtain k = 1.509860.
3. Using k, r, and r', (10) gives us cos /'/cos / = 0.147660, and from (11) we obtain
/ = 1.416096 radians. Now (10) yields e = 0.0932 and (4) gives a = 1.5231.
4. From (12), obtain u = -73.64?,

or equivalently (add 360?) u = 286.36?.

5. Use (6) to find Tp = 686.57 days, then (13) to compute E = 1.324532 radians,
and finally (14) to obtain t ? to = 134.86 days. Simple counting (or an almanac
or a spreadsheet with calendar functions) yields t0 = February 20, 1996 (at
approximately 3.4 hours). The computation is complete!
For comparison, the correct values are i = 1.849?, ft = 49.557?,
e = 0.093405, uu = 286.502?, and t0 = February 20, 1996.

a = 1.523692,

2. Assuming that we have accurate data to begin with, the most obvious
Example
weakness in the method is the estimation of k via the trapezoidal rule. As the interval
?
f increases, we lose accuracy in this estimation. The following table extends
f
?
f on the computed elements.
Example 1 to show the effect of larger intervals f
The value of t is the same as before, but now we allow t' (and thus /' ? /) to
increase:

and r' = (0.5500,

For the planet Venus, r = (0.3304, -0.6482, -0.0279)
=
=
with times t
-0.4757, -0.0382)
July 18, 1996 (0
July 4, 1996 (0 hours) and t'
hours). Calculate the orbital elements.

Exercise.

Answer: The correct elements are i = 3.394?, ft =

0.006773, co = 54.89?, and t0 = March 22, 1996.
Derivation

of Kepler's

76.680?,

a =

0.723332,

e =

Equation

Our last task is to develop Kepler's equation (14) and the related formula (13) more
fully. Since both derivations are readily available in textbooks (e.g. [7, 8]), we shall
VOL. 29, NO. 5, NOVEMBER 1998

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions

403

be brief. For (13), start with the half-angle formula

tan

1 ? cos E
o ?E
=-?.
2
l + cos?

Next, observe from Figure 5 that

all -e2)
cos /.
1 + e cos /

a cos E = ae + r cos / = ae+Solving the second

tion yields

equation

for cos E and substituting the result into the firstequa?

/1 ? e \w /1 ? cos J
9E
f
tan2 ? = '
1 + e J \ 1 + cos /
and another application of the tangent half-angle formula finishes the job.
For (14), we start by finding the area S in Figure 5, either by computing the polar
integral
,2\ \2
q(l
ff (
d9
2
27o
Jo V1
Vi + e^ cos ^ /
1

(hard) or an equivalent integral in rectangular coordinates (easier).

get S = \a2\/\
e2(E
esinJ^). By Kepler's Second Law
S
?
t to
which easily produces

value.

Kepler's

In any case, we

7ra2VT^2

equation

T*v
when we replace

S with the computed

References
1. WalterBtihler,Gauss: A BiographicalStudy,SpringerVerlag,New York,1981.
2. J.M. A. Danby,FundamentalsofCelestialMechanics,MacMillan,New York,1962.
3. Ross L. Finneyand George B. Thomas,Jr.Calculus, 2nd ed., Addison-Wesley,
Reading,MA, 1994.
4. Carl FriedrichGauss, SummarischeUbersichtder zur Bestimmungder Bahnen der Beiden Neuen
HauptplanetenAngewandtenMethoden,Werke,6, 148-165.
on thePersonalComputer,Springer-Verlag,
and Thomas Pfleger,
New
5. OliverMontenbruck
Astronomy
York,1991.
Associationof America,Washington,
6. HarryPollard,CelestialMechanics,Mathematical
DC, 1976.
7. JohnPrussingand Bruce Conway,OrbitalMechanics,OxfordUniversity
Press,New York,1993.
8. A. E. Roy,OrbitalMotion,2nd ed., Adam HilgerLtd.,Bristol,1982.

404

THE COLLEGE MATHEMATICSJOURNAL

This content downloaded from 62.204.192.85 on Fri, 11 Jul 2014 04:53:27 AM

All use subject to JSTOR Terms and Conditions