ASTRODYNAMICS

@ACADEMIC PRESS, New York and London APPLICATIONS AND

ADVANCED TOPICS
1967

ROBERT M. L. BAKER, JR.

Computer Sciences Corporation
and University of California
Los Angeles, California

"
 COPYRIGHT @ 1967, BY ACADEMIC PRESS INC.

ALL RIGHTS RESERVED

NO PART OF THIS BOOK MAY BE REPRODUCED IN ANY

FORM, BY PHOTOSTAT, MICROFILM, OR ANY OTHER
MEANS, WITHOUT WRITTEN PERMISSION FROM THE
To my wife, Bonnie
PUBLISHERS, REPRODUCTION IN WHOLE OR IN PART
FOR ANY PURPOSE OF THE UNITED STATES GOVERN­
MENT IS PERMITTED.

THIS WORK REPRESENTS, IN PART, RESEARCH SPON­

SORED BY THE UNITED STATES AIR FORCE UNDER
AF49(638)-495 MONITORED AT UCLA BY THE
AIR FORCE OFFICE OF SCIENTIFIC RESEARCH. IT
ALSO REPRESENTS SOME RESEARCH SPONSORED BY
THE OFFICE OF THE CHIEF SCIENTIST, LOCKHEED­
CALIFORNIA COMPANY AND COMPUTER SCIENCES
CORPORATION.

ACADEMIC PRESS INC.

111 Fifth Avenue, New York 10003, N.Y.

United Kingdom Edition published by

ACADEMIC PRESS INC. (LONDON) LTD.
Berkeley Square House, London W.1

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 67-14535

PRINTED IN THE UNITED STATES OF AMERICA

 PREFACE

This volume represents a modification and extension of the advanced

sections of the original edition of An Introduction to Astrodynamics, pub­
lished in 1960. Fresh material is included on the subjects of electromagnetic
drag; general relativity; preliminary orbit determination and correction
procedure (for example, dealing with both minimum-variance filtering
and conventional least squares); extension of variation-of-parameters pre­
diction; regularization techniques; and so forth. Other sections (for example,
dealing with applications to lunar and interplanetary trajectories; general
perturbations; and so on) while covering standard material, offer a unique
arrangement of topics which benefits the teacher in that it has been developed
in the actual process of teaching from the original volume over the past
seven years. In this regard some ninety-six new exercises have been added.
It is the intention of the author to employ the introductory volume,
"An Introduction to Astrodynamics," as a prerequisite volume to this
advanced volume and it is employed in this fashion in the astronautical
courses offered at the University of California at Los Angeles. Thus,
the reader will find rather frequent cross reference to the "introductory
volume" within the main body of the text.
This volume is meant to什11 the gap between the engineering handbooks,
吹
which tend rapidly to become outdated, and the more sophisticated texts,
which tend either to be over the head of most senior university students or
to lack practical value. Emphasis is given to alternative procedures and the
limitations of each rather than to a collection of formulas to be utilized in
cookbook fashion or committed to memory. Each chapter includes a certain
amount of advanced exposition. Elementary or tedious background ma­
terial is presented in the appendixes. Most of the analyses are derived from
first principles and, although the emphasis in the text is upon practical
．．
VII
 ．．．
VIII PREFACE PREFACE ix

application rather than elaborate mathematical proofs, the text does in­ Baker, Joan Boyle, Linda Foltz, and Janet Clodfelter, who typed many of
elude over fifty derivations of fundamental relationships. At the end of the rough drafts. P. R. Peabody, Kurt Forster, Geza Gedeon, Bernard
each chapter a list of key equations is presented and as a further aid to the Cohlan, Dave Pierce, Ted Moyer, Joseph Ball, Donald Lamar, Pedro
reader a summary table of the astrodynamic constants more frequently Escobal, Paul Koskela, Edward Pitkin, J. L. Junkins, W. E. Nally, Robert
employed in calculations is given on pages 441 and 442. The glossary of R. Lockry, and Howard Dielmann provided me with technical material
terms provides a self-contained key to all of the astronomical nomenclature and constructive suggestions for many of the subsections. Professor Samuel
utilized in the text and each definition includes a reference to the section Herrick, especially, has contributed both directly, through specific sugges­
where the term is treated in the text. (The glossary of symbols, found in the tions and theories, and indirectly, through the years of training that he has
introductory volume, is not repeated since it applies equally well to both provided the author and many of the other contributors to the text.
volumes.) The reader is not expected to have had a prior background in
classical mechanics or astronomy and, for that matter, sufficient prerequisites Los Angeles, California R.M.L.B., Jr.
for the book would be freshman and sophomore courses in physics or engi­
neering, mathematical courses in integral calculus, and a familiarity with
the introductory volume. For the more advanced reader the rather com­
prehensive list of references should serve as a key to the literature (up to
1967) for the purpose of further research and self-study.
Because an annotated table of contents of the text is available separately,
little detail is required in the discussion of the text's scope here. Chapters
1,3, and 4 discuss the orbit·determination and orbit prediction. The、nu­
merical examples found here serve as a guide to the practical utilization
of the analysis. It is emphasized that each and every orbit determination
procedure includes a complete illustrative numerical example. These are
indispensable aids to the proper understanding and implementation of the
procedures. Chapter 2 involves a detailed accounting for the various per­
turbative forces that influence spacecraft motion including a very detailed
examination of radiation pressure and drag perturbations. Because the text
is meant to be a practical one, both for teachers and for practicing aero­
space scientists, the subject of the application of all of the proceeding ma­
terial in this and in the introductory volume is paramount in importance
and the last two chapters treat applications to lunar and interplanetary
orbits and trajectories. Each chapter contains exercises of various degrees
of difficulty, some of which lead to extensions of the,_theory.
At the University of California, Los Angeles, the advanced text has
been employed in connection with one undergraduate and one graduate
course. The undergraduate course in orbit determination utilizes Chapter 1
for the first half of the quarter and a class-project orbit-determination
numerical study for the second half. The graduate class draws from se­
lected sections of Chapters 2 through 6.
The author wishes to acknowledge the extensive services of Merrilee
Vold, who organized and retyped the final manuscript and of Bonnie
 1.4 FINAL PROCESSING OF OBSERVATIONAL DATA 89
IMPROVEMENT
88 ORBIT DETERMINATION AND
In matrix notation it is also possible to define the already mentioned
where 2
standard deviations of the parameters, api'This is accomplished by ex­
叭
tracting the inverse matrix of the least-squares normal equations (Eqs.
Wz2
． ( 1.107)]. Such a matrix is termed the "covariance" matrix and is a free by­
．
． product of least-squares inversion. Of course, the confidence one has in
W今
W( the api is dependent upon the correctness of the assigned weights and the
， validity of the assumption that the observations are uncorrelated and only
．
2
contain random errors. Because of inevitable systematic errors these叨
WN
lead to an overly optimistic estimate of the accuracy of the parameters.
ght ·left as an exercise
ed least-squares form is It is emphasized that this formulation of the least-squares procedure
Th e ver ifi c atio n of this wei
points up the fact that the old observed-minus-computed residuals do not
for the reader. need to be "remembered" by the computer as the new residuals are simply
by the coefficient of 11归 accumulated in the A勹z matrix elements.* Also, one needs only (n2 + n)/2
Next divide each equation through
..·C(t) Ap.
t)
matrix elements in the matrix as the other elements are redundant (that

如 ｀ 丿 S p, I (�勹 Opd ( A,,, 卜

+ is, it is a symmetric matrix); thus least squares is eminently suitable for
high-speed digital computers, and not necessarily large ones, as the com­

妇(�) /1p, + (�) Llpz ... + (�) Llp; +···+ (�

) Llpn puter's size is governed by the number of p/s (parameters) that we wish to
+
improve, not the number of observations. It might be useful to launch into

rn
．
the entirely analogous filter theory approach; but since filters are more
. : ．

处｀』 Op, C 』 Opd 4 (�;:) Ap, C(-U:) Ap.

applicable to navigation, we shall hold such a discussion in abeyance until
+
Chapter 5.
. A word of caution should be added to the foregoing discussion. It is not
．

(?) Llp; + ...+(五） Llpn,

ordinarily "easy " to invert the least-squares equations, that is, (ATA)主
．
:

(�!勹 (?)
妨N
. This difficulty arises in spite of the fact that we usually have a very large
Llp, +· · +
如= Ap 1 + number of residuals and hence a strongly over-determined system of equa­
can be expressed as: tions.This problem, as already noted, is the result of the quality, not the
which are the same as Eq. (l.l04), and quantity of data—quality here not in the sense of accuracy (in which sense
it was used previously), but in the sense of proper correlation to the param­
eters whose improvement is being sought. As an almost trivial example,
3

．．（。也fop;) ... consider the fact that no matter how many times or how accurately one
．
．
measured the geocentric equatorial x and y coordinates of a satellite, one
． could never precisely define its z coordinate.
5
?5,

...(d虾I饥） ...
'5(
t

In actually computing (A1'A)-1, one often scales the problem in such

贪
a way that the longest diagonal of the matrix consists of ones. If the off­
or symbolically
AL1x=L1z. diagonal elements were nearly zero, then the matrix inversion would be
above through by A气 "strong." In practice such elements may have values as high as 0.95 and
To form a square matrix, multiply the
芝戈＄
，

the inversion is therefore usually quite "weak," that is, the solution exhibits
§ ，
F…:�： ；;

ATA Llx= AT Liz but few significant figures. As already noted, the suggestion by Anderle
．？义戈

or
Llx = (ATA)-1A Liz
T
．．

* A1'A elements are, of course, changed if improved partials are formed.

q. (!.113).
by inversion, which is the same as E
．．．
．二
咖
＿
＿ ＿咖
•••
•
• 」-