Calvin H. Wilcox wether Sound is Propagation in Stratified FluidsApplied Mathematical Sciences | Volume 50Applied Mathematical Sciences . John: Partial Diferentiat Equations, 4th ed. (cloth) Sirovich: Techniques of Asymptotic Analysis. }. Hale: Theory of Functional Differential Equations, 2nd ed. (cloth) |. Percus: Combinatorial Methods. von Mises/Friedrichs: Fluid Dynamics. Freiberger/Grenander: A Short Course in Computational Probability and Statistics. Pipkin: Lectures on Viscoelasticity Theory. 3. Giacaglia: Perturbation Methods in Non-Linear Systems. Friedtichs: Spectral Theory af Operators in Hilbert Space. . Stroud: Numerical Quadrature and Solution of Ordinary Differential Equations. |. Wolovich: Linear Muttvariable Systems. Berkovitz: Optimal Control Theory. . Bluman/Cole: Similarity Methods for Diferential Equations, . Yoshizawa: Stabilty Theory and the Existence of Periodic Solutions and Almost Periodie Solutions. 45, Braun: Differential Equations and Their Applications, Srd ed. (cloth) 16. Lefschetz: Applications of Aigebraic Topology. 17, Collatz/Wetterling: Optimization Problems. 18. Grenander: Pattern Synthesis: Lectures in Pattern Theory, Vol 1. 19. Marsden/MeCracken: The Hopt Biturcation and its Applications. 20. Driver: Ordinary and Delay Differential Equations. 21. Courant/Friedrichs: Supersonic Flow and Shock Waves. (cioth) 22, Rouche/Habets/Laloy: Stability Theory by Liapunov's Direct Method. 23. Lamperti: Stochastic Processes: A Survey of the Mathematical Theory. 24, Grenander: Pattern Analysis: Lectures in Pattern Theory, Vol. I. 25. Davies: Integral Translorms and Their Applications. 26, Kushneri/Clark: Stochastic Approximation Methods for Constrained and Unconstrained Systems. 27. de Boor: A Practical Guide to Splines. 28. Keilson: Markov Chain Models—Rarity and Exponential 29. de Veubske: A Course in Elasticity, 30. Sniatycki: Geometric Quantization and Quantum Mechanics. 31. Reid: Sturmian Theory for Ordinary Differential Equations. 32, Meis/Markowitz: Numerical Solution of Partial Ditferantat Equations. 83. Grenander: Regular Structures: Lectures in Pattern Theory, Vol. I. 34. Kevorkian/Cole: Perturbation Methods in Applied Mathematics. (cloth) 95. Carr: Applications of Centre Manitold Theory. ESBS 0m vomsenn {continued on inside back cover)Calvin H. Wilcox Sound Propagation in Stratified Fluids 5 Springer-Verlag New York Berlin Heidelberg TokyoCalvin H. Wilcox University of Utah Department of Mathematics Salt Lake City, Utah 84112 U.S.A. ‘AMS Classification: 46NOS, 76005, Library of Congress Cataloging in Publication Data Wilcox, Calvin H. (Calvin Hayden) ‘Sound propagation in stratified fluids. ‘(Applied mathematical sciences ; v. 50) Bibliography: p. Includes index. 1, Sound-waves-Transmission. 2, Fluids-Acoustic properties. 3. Stratified flow. 1. Tite. If, Title: Stratified fluids. IL Series: Applied mathematical sciences (Springer-Verlag New York Inc.) ; v. 50. QAI.AG47 vol. 50 (QC233] 510s [534'.23] 84-1447 © 1984 by Springer-Verlag New York Inc. All rights reserved. No part of this book may be translated or reproduced in any form ‘without written permission from Springer-Verlag, 175 Fifth Avenue, New York, N.Y. 10010, U.S.A. Printed and bound by R.R. Donnelley & Sons, Harrisonburg, Virginia Printed in the United States of America, 987654321 ISBN 0-387-90986-9 Springer-Verlag New York Berlin Heidetherg Tokyo ISBN 3-40-90986-9 Springer-Verlag Berlin Heidelberg New York TokyoPreface Thix wonograph was begun ducing ay sabbatical year in 1980, when T vas 4 visiting professor at the University of Bona, and completed at che University of Utah in 1981. Proliminary studies were carried out during che period 1972-79 at Utah and while T held visiting professorships at the Untveraiey of Lidge (1973-74), the University of Stutegare (197% and 1976- 77) ond the Beole Polytechnique FEdGrale of Lausanne (1979). ‘Throughout this period ay research was supported by the U.S. Office of Naval Research. T should Like ¢o express here ay appreciation for the support of the Universities of Boun, Lidge, Stuttgart and Utah, che Ecole Polytwchnique Fédérale, the Alexander von Humboldt Foundation and the Office of Naval Research which made the work possible. My special thanks are expressed to Profesaors $. Chatter}i (Lausanne), H.C. Garair (Lig), - Lets (Bonn) and P. Werner (Stuttgart) for arranging ay visits to their universities. I also want to thank Professor Jean Claude Guillot of the University of Paris who collaborated with me during the period 1975-78 on the spectral theory of che Epstein operator. That work, and our many discussions during that period of wave propagation in stratified media, contributed importantly co the final form of the work presented here. calvin B. Wilcox Boon August, 1982Introduction Stratified fluids whose densities, sound speeds and other parameters are functions of a single depth coordinate occur widely in nature, Indeed, the earth's gravitational field imposes a stratification on its atnosphe oceans and lakes. It is well known that cheir stratification has a profound effect on the propagation of sound in these fluids, The most seriking effect 45 probably the occurrence of acoustic ducts, due €0 minina of che sound speed, that can trap sound waves and cause then te propagate hors— zoacally. the reflection, transmission and distortion of sonar signals by acoustic ducts {6 important in interpreting sonar echoes. Signal scattering by layers of microscopic marine organisus i important to both sonar engi- neers end sarine biologists. Again, reflection of signals from bottom sodinont Layers overlying a penetrable bottom are of interest both as sources of vnwanted echoes and in the acoustic probing of such Layere, Many other examples could be given. The purpose of Eh{s monograph 18 to develop from first principles « theory of sound propagation in stratified ftuide whose densities and sound speeds are easentially arbitrary functions of the depth. In physical cers, the propagation of both tine-harmonic and transient fields io analyzed. ‘The corresponding mathematical model leads to the study of boundary value problems for a scalar wave equation whose coofficients contain the pre~ scribed density and sound apeed functions. In the formaliom adopted here these problens are {atinately related to the spectral analysie of 2 partial differential operator, acting in a Wilbert space of functions defined in the domain occupied by the Fluid. ‘The intended audience for this monograph includes both those appliod physiciate and engineers who are concerned with sound propagation in stratified Fluide and those mathematicians vho sre interested ia spectral analysia and boundary value problems for partial differential operarors.vill miempepucazon An atteapt to address simitaneously two such disparate groups must raise fhe question: is there a common domain of discourse? The honest answer to thie question is no! Current aathenatical Literature on spectral analyeis and boundary value problens is based squarely on functional analysis, particularly the theory of Linear transformations in Hilbert spaces. This eheory has been readily accesstble ever aiace the publication of M. Stone's AMS Colloquium volune in 1932. Nevertheless, the theory has not become a part of the curricula of spplied physics snd engineering and it 5 seldom seen in applied science Literature on wave propagation. Instead, that Literature is characterized by, oa the one hand, the use of heuristic non-rigorous arguneats and, on the other, by formal aunipclations that typically tavolve divergent series and Integrals, generalized functions of unspecified types ond the Like, ‘The differences in style ond method outlined above pose a dileama, Can an exposition of our subject be written that is accessible and useful ko both applied scientists and mathematicians? An attempt Is made to do this below by beginning each chapter wich 2 substantial sumary. Taken together, the summaries present the basic physical concepte and resules of the theory, formulated in the simplest and most concise form consistent with their nature. ‘The purpose of the sumeries is twofold. First, chey can be inter preted in the heurtstte way favored by applied physicists and engineers. When read in this way they are Independent of the reat of che text and Present a complete statonone of che physical content of che theory. Second, readers conversant with Hilbert space theory can interpret the summaries ae concise statewents of the prine{pal concepts and results of the rigorous mathematical theory. When read in this second vay, the sumaries serve ae lan introduction to and overview of the conplete theory.Contents Preface caneTER 1. cHuPTER 2. cumpree 3. cuvpreR 4. CHAPTER 5. APPERDIX 1. APPENDIX 2. REFERENCES INDEX wrnouertoN ‘THE PROPAGATION PROBLEMS AND THSIR SOLUTIONS Ls Summary 2) The Acoustle Propagator 5. Solutions with Finite Energy SPECTRAL ANALYSIS OF SOUND PROPAGATION IN STRATIFIED FLUIDS Le Sumery ‘The Reduced Propagator A, Solutions of the Byuatton aye = 6 Spectral Propertics of Generalized Eigenfunctions of A, ‘The Spectral Family of The Dispersion Relations ‘The Spectral Family of A Wornal ode Expansions for A Seni-Infinite and Finite Layers TRANSLENT SOUND FTELD STRUCTURE IN STRATIFIED FLUIDS 1. Summary 2. Normal Mode Expansions of Transient Sound Fielés 3) Transient Free waves 4] Transient Guided Waves 51 Asympeatic Energy Distributions 6, Senininfinite and Finite Layers SCATTERING OF SIGNALS BY INHOMOGENEOUS LAYERS Le Summary 2. Signals in Homogeneous Fluids 3. The Reflected and Tranenteted Signals 4] Construction of the Scattering Operator 51 The Scactering Operator and Signal Structure THE WEYL-KODATRA~ZTTCHMARSH THEORY STATIONARY PHASE ESTIMATES OP OSCILLATORY INTEGRALS ETH PARAMETERS Page a 13 1 1 a 2 32 33 3 55 n a5 90) 10% uw 125 12s 133 a 330) Se 156 LoL 261 187 270 178 18s 189 197 198Chapter 1 Introduction This monograph presenca 2 theary of the propagation of transient sound waves {a plane stratified flofds whose densities ply) and eound opeeds e(y) lave functions of the depth y. The main goal of the theory ts to calculate che signals produced by prescribed localized sources and to determine their asymptotic behavior for large times. The results include criteria for the occurrence of acoustic ducts that trap a portion of the signal and cause tt to propagate horizontally. The sathenatica) theory 4s based on a real valued function u(t.x), the acoustic pressure or potential at the tine t and the spatial point with Cartesian coordinates % * (,,4,,y)) which satisfies a scalar vave equation chat containe the fluid density p(y) and sound speed ely). A derivation of Chis vave equation from the Laws of fluid dynamics is presented here in order to clarify the hypotheses needed for te validity. Analogous deri~ vations may be found in the monographs of F. G. Friedlander [6}" and LM, Brekhovekthe [2] and in an avetele of T. Tolstoy [19]. A conpresssble, nonviscous, heat conducting fluid 1s considered, The states of the fluid ars characterized by the variables = fluld pressure = fluid density flutd velocity ~ fluid temperature (absolute ecate) an {3 | 5» tsuta enesopy ner ante mae Bach 48 @ function of the variables t and X. These variables aattety the “Numbers in square brackets denote references from the List at the end Of the monograph.“TWERODUCTTON equations ay an bs as oB-2v. avn =o, a £(p.P 42) = 0 and 49,055) = 0, where «sy +e denotes the material derivative. Equations (1-2) and (1.3) express the conservation of mass and Linear monentun, respectively, The f1elé @ = G(e,x) describes the external forces that act on the fluid, Equation (2.0 relates the variation of $ to the diffusion of heat in a medium with thermal conductivity k = KO). Equations (1.5) are alternative fore of the chermodyaanic equation of state. Sound vaves wilt be studied dn a static inhomogeneous fluid in which external forces are negligible. Thus {£ che Fluid parameters (1.1) for the seatic fluid are denoted by = subseript sexo then one has > $< 8, Meno, Steak a 2 8, Mw 0, Beno, z ae Be = 0 oF oy = py) a» op = 0 or py = Pye) ont a.10) 2+ QT) = 0. ‘The Last equation, together with suitable boundary conditions, determines 2 uatgue1. Introduction aw Ty = TO). Teer (1.5) implies chac cua) FB (2) 4P9 TD) = 0, whence aay Pe = const: and 99(X) may he determined by solving (1.12) with known po and Ty (X)- Finally, S9(%) te determined by an) 90949 48,00) © 0 ‘The equations governing sound vaves in a static inhomogeneous fluid will be obtained by @ formal perturbation method applied to the static solution just discussed. The sound waves wil} be agsuned to be excited by body forces with force density 15) where © is a parameter. Moreover, ¢ and G, will be assumed to be so small chat the resulting disturbance is adequately described by the first order terns 4s) a Teylor series expansion in €, Tho perturbation equations are obtained by substituting (1.15) and py tem Pp + em a1 1, + er, = 55 + eS; tn equations (1.2)-(1.5) aud dropping terns containing powers of © higher chan the fest. ‘Together with the snail displacenent assumption (1-16) che additional hypothesie will be made that the acoustic disturbances are so rapid that heat aéffusion is negligible during pawsage of the sound vaves, This assumption, which {8 called the adJabatic hypothesis, 48 equivalent to dropping the diffusion term in (1.4) by setting k= 0, Thus for sound4 1. nengopueszon waves one replaces (1.4) by aan be Inotead of Linearizing (1.17) directly one may take the materiel derivative of the equation of state #(p,0,8) = 0 and use (1.17) to get Beg Be Gas) 4, +o, Bao where @, = 2¢/3p and Gy = A¢/99. Substéeuting (1-16) tn (1,18) and eeatee ing only Linear terms in & gives (since Tp, = 0) So Be 4 gt[22t + py t) = Oy ae tM L5e + MM sO 88 = 9pDosP9(%) 5,00). (1.20) a8 = GyC00s05000.5400). his can be written ean) conf ° 2 ~ 2 = [22] om oo -- 2B, Ae the local speed of sound at X. Linearizing equations (1.2) and (1.3) yields 23 Bey + OR) =o G24) Py Bb + Gey = py B Combining (1-21) ané (1.23) gives the alternative equation as) Bi sto THLe Introduction . Finally, differentiating (1.25) wlth reupece co € and uelng (2.24) to etininate % gives a scalar vave equation for p, Ha - cog 0+ (Lon) 0.28) at = cos 7 [Lve,) =F where aan Pe -ctoy 0 8 ts determined by the prescribed source density &. TE 948, * 9G then one may Sntroduce an acoustte potential e(tsx) uch that a.28) 3 4 ay Hm =- Bee ‘Then (4.24) and (1.25) are satisfied if 29) Ha-toy 0+ (Lo) er with F > 96/t. te will be convenient to use this formulation below. A static fluid w{ll be said to be plane stratified if che flutd parameters depend on the single depth coordinate y. Tn this case, writing © * cG)s Pe * PG) in (2.29) one gets the partial differential cs Be oe Eee Hee ow & ) |= ereonian Sound vaves governed by (1.30) are studied in the remainder of this nonograph. The fluid media are agsuned to be unlimited in horizontal planes, 80 that ~® < x42, <%- The principal case treated ts that of a completely unlisited fluid for which -" <7 ©, This idealized problem provides a model for the analysis of sound propagation in portions of a Finite stratified fluid that are far from tho boundaries. ‘The cases of & somi-infinste layer, with 0 W and all uEK. It is also customary to normalize (1I(2)) by requiring continutty From the right: a.44) HQsO)s = We for all XE Rand ue x, but this is aor essential. ‘The spectral integest re 45) [£09 a0,A, Introduction where FQ) 1s 2 continuous complex-valued function of 2 ER, witl be foterpreted ag a Linear operator in, It will be defined in two steps. First, 42 > 0. The spectral family {11(\)) may be constructed by Fourier analysis, Indeed, Lf w€H then the Liste asp Ge) + ute aber J) oH ween Fee GON Tiga extete an X and one he a5 w@ = 2a aoe | et aemae, = else also in, and (2.53) tall = tell. ‘These results are Plancherel's theory of the Fourter transform. With this notation the spactral family of Aw -i in R? ts given by i | ° A Tt ds clear that in this case 1Q) varies continously with h. The dicferencial operator (1-31), acting in the vhole space K? or in a eemi-infinite or finice layer, is much closer to the second example than tho first. In particular, it alvaya has @ pure continuous spectrum. However, when p(y) and e(y) are varsable the structure of {I1Q)} may be complicated by the existence of acoustic ducts and corresponding fant lies of subspaces of H, OF course, the spectral resolution (1.36) 16 usefuh only when an explicit construction of {QJ}, such ae (1.50) or (1-54), can be found. This is achieved below for sound waves in stratified Fluids by means of the physically motivated concept of noraal nodes. Tho organization of the remainder of the monograph will now be deseribed. Chapter 2 presents construction of a selfadjoint realization of the acoustic propagator A defined by (1.51), followed by a solution ofLs Introduction a ‘the wave equation (1.30) by means of the funetionsd caleulus (1.47), (1.48) for A. The principal rewules of the monograph are developed in che long, Chapter 3 which presenta an explicit construction of the spectral family of A, under the condition that p(y) and o(y) have Limits in a suitable sense at y= 2. The construction 1s based on two families of generalized eigenfunctions, of normal soda functions. Physically, the first family describes the response of the fluid to an incident plane acoustic wave. The second fanily exists only vhon the flusd has an acoustic duct and represents guided aves that are localized near the duct. Chapter 4 presente a detailed analysis of the large-tine behavior of transient gound fields in stratified flvide, The calewlattons begin with 2 normal node expansion of the fields. ‘Then, under suitable restrictions on he behavior of p(y) and e(y) at y =, the large-tine behavior of the Helds is determined by the method of stationary phase. The convergence of v and ita First-order derivatives in H to the corresponding asymptotic wave functions 1s demonstrated and used to determine the asyuptotic discribution of energy for large eines. Chapter 5 presents an analysis of the acattering of acoustic signaie by an dnhonogensous stratified layer of finite thickness chat separates two homogeneous fluids. 9 seateering operator is constructed and shoxn to determine completely the asymptotic fora of the reflected and cranonitted signals that are produced by a proscribed source localized in one of the honogensous fluids. ‘The monograph concludes with two appendices. Appendix 1 presents in 4 concise form, with references co the Iieerature, the Neyl-Kedaira~ Tatehmarsh theory of vingular Sturu-Liowille operators. These results provide che foundation for the proof in Chapter 3 of the completeness and orthogonality of the normal mode functions. Appendix 2 presents 2 version of the stationary phase estimate For oscillacary integrals containing parancters. ‘The uniformity with respect to the psraneters of these estimates is essential for the applications in Chapter 4, ‘The physical Literature on sound propagation in stratified fluids is very large. ‘The modern development date from the pioneering work of C. Ly Pekeris [27], published in 1943, tn which ply) and e(y) ere piecewise constant. Ip the interventag years hundreds of papers on the subject have appeared in che Journal of the Acoustical Society of America alone. Some of the matertal has been reviewed in che well knewn books of Ly M. Brekhovskikh [2] and I. Tolstoy and C. S. Clay (20).2 1. Rerpooverson ‘The work presented here differs from the earlier Literature in being based on concepts and results from the theory of Linear transformations in Hilbert spaces, Some of che many advantages of thie approach will be entioned here. First, the use of the Woyl-Kodaira-Titehmarsh theory of singular Sturn-Liouville operators permits a straightforward derivation of the completeness and orthogonality of the nornal made functions, without as 4 Wilbert space theory. Second, the use of the notions of convergence in hypotheses concerning p(y} and c(y)- Tits theory is intrinsically the sean-oquare and energy norms Ieads to results on the asyuptocic behavior For ¢ + © of transient waves that axe ot true ia the conventional sense of pointwise convergence. Finally, the introduction of the unitary scattering operator in Chapter 4 gives a particularly luctd form co the relationships among the incident, reflected and transmitted waveforms in the scatcering of signals » fan tnhonogencous Layer. Earlier applications of Hilbere space mechods co vave propagation ia stratified media were made by the author to the spacial case of the Pekeris| profile (23,26,27,28] and by J. C, Gulllot and Wilcox [7,8] and ¥. Dermenjian, Guillot and Wilcox [4] to the special case of the Epstein profile. For simplicity, displayed equations ere numbered separately in each chapter. The notation (j.k) alvays mena equation k of section § of the current chapter. The notation (4.j.1) {6 used to refer to equation k of section } of chapter 4,Chapter 2 The Propagation Problems and Their Solutions ‘The purpose of thie chapter La co present a mathematical formulation fang solution of the propagation probleas chat are studied in the subsequent chapters, The basic physical concepts and results are sumarized in 81. The remaining sections develop the sathenatical theory, together with rigorous proofs or references to the Literature. The case of an unlinited Fluid 16 treated here, The changes needed to treat seni~infinite and finite layers are indicated at the end of Chapter 3- 81. sumanr ‘The acoustic field in aa unlimited stratified fluid will be charec~ terized by « real-valued acoustic potential u(t,X) where X= Guy) = (ys) © R? denote the rectangular spatial coordinates of Chapter 1. Te partial derivatives of u wid]. be denoted by Dyas k= Oy 1s 2, 35 whore an Dy = 9/88, D1 = 97bmy Dy = /%R, Dy = ye LE GMs notation che wave egestion of Chapter 1 fakes ehe fore . ta e2 on + [by tu) = Hee an fe = fey 0 0+ [aby M4) = HED Where J = (D,4D,,D,) denotes the spatial gradient operator and P(t.) is a source density function. The acoustic pressure p and velocity field ¥ tn the fluid are given by [6] p-tiow a3) p= Dou and F = Sey respectively. 2u 2. ME PROPAGATION PROBLEMS AND THEIR SOLUETONE AS yet, no conditions other then positivity have been placed on the variations of p(y) and e(y). In the remainder of this work ie is assumed as @ minisel bypothesis that ae 0 < pg OY) Sey

eR eo she ten cone [Ly a} some ero nen with initial values f, g that satisfy (1.6), (1.7). The reduction of the principle is shown below.2.1, Seamary as the integral 9) EQ e) = f, Cyucew |? + FG) [Dpuces”) FF oT) ax, where df = dx,dxpdy, may be interpreted as the energy of the acoustic field win the ser KER at thne t. Te Le known that aoluttona of (1.8) satiety the conservation Lav (1-10) E(u,R 8)» ECOR?,0) whore E(u,R!,0) <4". Only solutions with finite coal energy are consi- dored below. A necessary and sufficient condition for u re have this property ie that the initial state have finite energy [22] aay J. {90 |? + &*G) gC0 [73 PEG) axe ® The ‘nietal value problem in {ts classical formulation (1.6), (1.8) Witt have & solution only ££ 0(3), 69), F(X) and g(X) are sutflctentty saooth. fowever, for arbitrary p(y), c(y) saristying (14) the problem say be shown to have a unique solution with finite energy vhonever the initial state fy has thie property. A proof 4s outlined in §3 below. A fornsh construction of the solution may be based on the operational calculus for the Linear operator &, defined by the differential operator aa) aun et) oo 0+ [Eds tu] acting in the ilbert epace € with sealar product i Ry TE the dousin of A is defined to be the set of al u @ euch that Vu € and 7 + tw) € chen A is a aeifadjoint non-negative operator. Horeover, a4) aCe) © eos Cw EM sin ea Dg 4 the solution of the initial valve problem piu + A 0, > 0, aus) WO) =f, uO) = ge16 THE PROPAGATION PROBLEMS AND THEIR SOLUTIONS Tr can be shown chat (1.14) is the solution with Finite energy whenever the Antetal values have finite energy. ‘The Snitiel value problem for (1.2) where P satisfies (1.5) and ute) 0 for & < T can be reformlated as ceR, ( pha + a0 + Flee (1.16) ec aay ween a [EM sin Oe) ae ea : Ft dg clear that for t > 0 this solution satisfies (1.15) with wo) oof 0M eas AAT ate F 18) ayutoy = [? feos aS Fe) ae 2 ‘The renaindor of the analysis will be based on the representation (2.14), Te will be convenient to rewrite it in the form a9) wCe) = Re (62,7 where v(t,X) is the complex-valued wave function defined by ve 2.20) weet) = EA and a.) press ales This representation is valid if f and g are real-valued and al £, £, g and AVE g are in. A rigorous interpretation of (1.14)-(1.21) may be based fon the calculus of selfadjeine operators in Hilbert spaces. Lt ig noteworthy that the sane formalism is applicable to the cases of semi~infinite and finite layers. For the senf-lafinite layer A} = (ky > Ob, Xe the Kilbort space with scaler product2.2, The Acoustic Propagater aw a.22) ads | BGT vay Pay oO) ax Be sed the doin of 4 is the set of all w EH auch Mae TUE % 4 Ju) €H and u satisfies the Dirichlet or Neumann condition at °. ‘The relations (1.14)-(1.21) thea hold as Defore. The analogous statenents hold for eho Finite layer RE = (XO < y Ch). §2. THE ACOUSTIC PROPAGATOR In this section a preciee definition of the acoustic propagator A is siven and a proof of its celfadjoineness in H is outlined. Here and throughout this work p(y) and e(y) are assumed to be Lebosgue measurable functions that satisfy the boundedness conditions (1.4). It fellows that ey 00 = [cg oon a F defines a measure in 4? thas 4s equivalent to Lebesgue measure (Lie. m 4s absolutely continuous with respect to dX and de/dk = &°%(y) 0G). Hence she Lebesgue space 2.2) Hm EetR,e FG) PAG) a {a a Wilbert apace with scalar product (1.13). Note that (1-4) inplics that ICs equivalent as a norsed space to the usual Lebesgue space Le(R4,dX), although they axe dlatince as Hilbert spaces. A selfadjoint realization of the differential operator A in be denoted by A, is obtained by defining the donain of A to be ea De) = AER) 9 fw > GOV Hye) where Lz(R') = Ly(R,dX) and aay LEY) = Let) 9 fu 6 Ga € EECRPDT 4s tho usual first Sobolev space [1]. ALl the differential operations ia (2.3), @.4) are to bo enderstoed in the sense of the theory of diacesbue tions, Te follows that rhe Linear operator A in dofined by (1.12). (2.3) sacteties ey Asaheo18 {THE PROLAGHETON PROBLEMS AND THETR SOLUTIONS where AY is the adjoint of A with respect to the sealar product (1.13). & proof of (2.5) may be given by the nethod employed in [26]. Alternatively, @ space (11, p. 322]. Indeed, if one defines a sesguilincar form A In Xt by 5) may be derived from Kato's theory of sesquilinear forms in Hilbert 6) Ba) = UR) Cx and an Aww =| G+ ov oon ax then 4t ds easy to verify that A is closed and non-negative, and chat A Le the unique selfadfoine noa-negative operator in H associated with A, As an sddiefonal dividend 4¢ follows fro Kato's second represencation theoren [a2 p. 332) chat D(A) © LEC) and for alt ve D(A") one has. 8 Ia? as dean = [| el? oo ax where [slg £4 the norm in X 83. SOLUTIONS WrTe FINETE ENERGY Te was shown {a [22] that the initial value problem (1.6), (1-8) hae a unique generalized solution with finite exergy (+ solution ¥FE) whenever the initial state £, g has finite energy; 4.e., (L-ll) holds. Taese conditions way be weieten, By (2.7), on ACE f) + llgl? = JAX of? + Il? < or, equivalently, 2) FeO), sex, Ie follows that the function u(c,X) defined by (1.14) is the unique solution FE and 6.3) we CRB 9 CR AERD), (ere CK(1,I0 is the set of k tines continuously differentioble X-valued function on I.)2.3, Solutions with Finies Enexoy co the acoustic fields studied below are defined by (1.14) where £, g setinfy (3.2). Te will be convenient to specialize further by writing u in the form (1.19)-(1.21), This Se correct provided £ and g are real valued, (3-2) holds and g © D(A). Of course, D¢s”"*) # H, in general. However, D(A!) is dense in H, by the spectral theorem. This fact can often be used to extend results derived from (1.19)-(1.21) to general solutions «FE.Chapter 3 Spectral Analysis of Sound Propagation in Stratified Fluids ‘the purpose of thia chapcer 4e to construct families of normal mode functions end te derive corresponding normal mode expansions for the acoustic propagator A, The basic physical concepts and results are sun~ narized in §1. A aore detatled formulation and complete proofs of ail che results are developed in tho remaining sections. This rather long chapter provides the analytical foundation for the solutions of the propagation problems that are developed in Chapters 4 and 5. Si. surmany ‘Throughout thie chapter ir ts assumed chat p(y) and e(y) saristy che boundedness conditions (2.1.4) and the four conditions: © Lo) = 891 ay <#y 2f" Jeon = ecm! ay 0 «2.19 for A> ct(im)y! aZtusd) © 19,0109)3.1. Sunny Pa aad Gay oo aan for A < cept [ex =F 2 [ 9.6.09 + tap. Ju particular, the solutions are osefilatory when A> ¢%( yo? and non-oseiLlatory wien A < cf(2#)u8. the first step in the analysis of A, is to construct special solutions of A,b = ¢ that have the asymptotic bebaviors 4,0 0d) = +oWh yt recon’, te0rush) = +oQh ye az) Leto. POD eal, y> = | { tk follows from an asymptotic calculation of the Wronsklans that 6, and 4, ate a solution basis for Ao = A$ when 2 # cM (OL*, while 6, ond 4, are a basis when A # Cau The nature of the apectrun and etgenfunctions of A, can be inferred from (1.12). Tt follows chat tf A < c#(6)u? then Ad = AP has bounded solutions only 1f ¢, and 6, are Linearly dependent. Thus a3 FO) = pt Wp .84) = 9 is an equation for the elgenvalues of 4,, where M denotes the Kronsksan and pW Se independent of y. The corresponding solutions aay Ueland = 9,60) bree ALGD) = OLE) 561% C))> where} = Ayu) As a root of (1.15), are square inteprable on R and honce ave eigenturctions of A, Horeover, A, can have no point eigonvaloes A> coyut, by (1.12). Thus o9(A,), the potne spectrum of A,, Lies in the interval (elp?,cf(@u8]. Getterta for op (Ay) to be empty, fintte or coustably infinite axe given in §6 below. Te VEIL be show thar the contiquous spectrum of A, 8 [e°()uP 2) and corresponding generalized eigenfunctions will be determined from (1.12).Ey 3. SPECTRAL ANALYSIS OP SOUND PROPAGATICA Por o(a)ué < \ < o(s)L* there is a single family of generalized efgen- functions of the form as) Wy aKSAD = 94649) oy CHsAD- For > cE(-a)u? there are tuo families defined by HOD = a0 6,000 16) VOD = aOR oC) ‘Te will he shown thar these functions have the following asymptotic forms. “14,0, WDY fen © ves aan 4G ~ a ANA DIY fae rr ta, 0409 $a, 00y e + Reid) © Lyte, (18) Vy Cott Dd ~ eg Gd? ao.ny G4) e : yr 14,60 $9,009 e + RG) © ryt a1) (ya, ~ 646 -t_ Giddy 4,60) © : ye fais Fade : yor, 2.20) $01) = 6G) s6.Gdy Dy e + RGD sytem Hore a. (us af (ude ay (usdds ag Cus) s GLC), 6,GLAD, CGA) Ress RyGnt), Ty (Usd) aad TGs) are functions of y and A that vill be calculated below. Faskiies of somal sode functions for A say be constructed from those for jp, by the rule (2.3). The folloving notation will be used,a2) YeGery Pad) = Cad? AP 4. Gy, [f.0, (yA? EO, (2.22) Wy Gsryspyhd = md? AP y. (yy [p],29, (242) € Oey 2.23) yee) = nt cP YP, PEM, R21, ‘The paramecer domains, Ry, %, are defined by ay) = 1D ER | FC) [PL (4.25) Ry = Cepsay © WP | cD pl? eRe eP*D IPl? (0.26) a+ (pew | Ip] eo, ez 1, where O, is the set of » > 0 for which A, has @ kth eigenvalue ‘the Umee families have different wave-theoretic interpretations that are characterized by thelr aayaptotic behaviors, Thus for (p,2) © 9 one has Horeag) | slerntag ty rye, aan 10%-Ly) e ’ ye Aptxtayy) he aan yoy ep 4) 0.28) Geyspd) © 2 L(prxta_y) Loree”) e the tye where q, = ails, R= RyClpL A, ete. Hance ¥jCey,Py8) behaves for y 740 Like an incident plane wave with propagation vector Ey = (p,-a)) plus a apecularly reflected wave with propagation wetor E, (0,9, vatle for y > ~= it behaves Like a pure transpitted plane vave vith propagation voetor &, = (p.-g.). The incident and transmitted plane waves can be shown to satiafy Sneli's law n(™) sin 6() = n(-) sin 6(-) where 6(=) and 8(-=) are the angles botwwen the y-axts and Ky and E,, eenpectively, end n(s=) = Us). 6 GuyspsA) has @ sistlar interpretation.26 3, SPECTRAL RWALYSIS OF SOUND PROPAGATION For (9,2) € My one has srg) iors) vey eo (pl 2.29) ¥Coyspst) = ™ oe my lO”, yom. Hence for y + te jy (4.9452) behaves Like a incident plane vave plus a specularly reflected wave while for y > -= ir is exponentially danged. ‘This 4s analogous to the phenomenon of tatal reflection of @ plane wave in 4 homogeneous sediun of refractive index n(«) = ee) at an interface with 2 medium of index n(-2) = e"?(-=) ) > n(—). For 9 EM & > 1, one has Eel) yee “OU ote Ts se, a 30) Ceara) ~ SP pee 2a . HHonce the functions yj (x.¥,p) can be interpreted as guided waves that axe exapped by total reflection in the acoustic duct vtere cy) < c(t). They propagate in the direction f = (p,0) parallel co the duct and decrease ‘exponentially with distance from it. The cooffictents Ry Ry and Ty, Ty in (1.27), (1.28), (1.29) may be interpreted as reflection and transmission coefficients, respectively, for the acattering of plane waves by the stratified fluid. They will be showa to satisty the conservation lave fe ects) 3D) : in? + . la =. pc) pes) ‘The three familtes ¥,, Y_ and Yy cepresent, collectively, the response of the stratified fluid to incident plane waves exp {i(pex-4y)}+ (Pag) & RY to ace this consider the mappings (Pa) = HC) = (ragCla] OD), AD a3) (2.9) = XC) = (eaglal WD, (AE (a) © 0) * Gre] DI, (AdBet, Stmmary 27 4H, is a analytic transformation of {2 onto the cone (2.33) {@.a | a> a tpt where (a3) a= Cel /e@))* = 1) 2 0, Similarly, j {s an analytic transformation of 9 onto the cone 2.35) co = Koad | efo,a) | eM 4 a, OHO, Ogre cy, a etloalol OD Gee, for y+ 4m and { (a) € & Quay *%Gsy,P.a) ~ 20,9) { ry oth atClel Ady, (oa) © Cy, AOD) 4g GO, GQ) Cc3.1. Sumazy 23 for y +=, In §9 it is shown that one way take: (2 c(@) 0 ©), (ua) € 6, UC, (2.46) e(o.9) = any? c=) pM? 2), (Oa) © 6. Another fanily of normal node functions for A Ls defied by 45) Obayed = BETH D, Cag) ee. It 1e clear chat A$. = Mpyq) @ and «a.48) #Gag.ea = 27 FP 4 cy,0,4) where an 0.7.0 = LORD: ‘he asyaptotic behavior of # for y > 2 may be derived from (1.42), (1.43), (1.44). Tr 4s given by OMY 4 Fz MOM, 0,9) € 6, ACpextay) 48) #Guy.p.q) > elp.ant e 4B, bP Op,9) € oy, Ft Gl MP, eg ec, for y 7 4" and F fodelO, gy e (1.49) Caarspead ~ ofpsad { Hy lP# IPL N7, fon) € 6, fOr) 6 OHM, Ger, for y +=. These relations clearly inply that ¢_(e.y,pyq) 18 not simply a multiple of $,(i¥sPra). By contrast the guided mode functions have the symmetry property30 3. SPECTRAL ANALYSTS OF SOUKD PROPAGATION 4,50) Youre) = RG, kL, because the fictions ¥(y,|pl) are real-valued and depend on p only stiroush [ol ‘The completeness of the family 1.0 sly, Leet consequence of the invariance of the wave equation under } is derived from chat of ss} $n $9, The existence of the evo fantlies 6, and 4 ia a reversal. Ia Chapter 4 the family $_ ie used to construct asyuptotic solutions for E+ 46 of the propagation problem. ‘The normal aode expansions vhich are the main results of his chapter will now be formulated. For clarity, the case in which A has no guided odes is described first. The general case is described at the end of the section, ‘The normal mode expansions For A are in essence Hilbert space expan stones that Le, they converge ia 2 mean equare, of Hilbert space, sense. ‘The space K in which A acte was defined by (2.1.13). The Lebesgue spaces 1,(Q of square Integrable functions oa QC K™ are also needed. 12(Q) is a Wilbere opace vith scalar product aan wo «| TOT oy a ‘ whore a * dp,.+.de, 48 the volume elenent in 2”. 1€ A haa no guided sodes then the fanity (6,(1,7) 1 PER!) is a complete fanily of normal mode functions for A, This statement means that avery b 6X has a nique generalized Fourier transform fy © (8) such thaz as 8,0) = ceDttw [ | FIER no 7) oo at et G53) Ie = FAL, cary and a) way sein f eye 8 Equation (1453) generalizes the Paroeval relation of Fourier snslysie. Equation (1.52) 1s a condensed nocation for the assertion that if {K,) ts any nested sequence of compact sets in R? such that UK, = RY, and if3.1, Summary a nD, KEK, 55) 4,00 = oo, Kew Ky then the integrals c.5e) | GER noo Fe) ona = | EA new <*oro tren NR? Nk, are finite for evory # € 8? and define functions fy, © Ls(R*) such that the sequence (hg) #8 convergent in L;(R). fy de by definttion the Limit of his sequence: Equation (1-54) has the analogous interpretation. Thus it represenes an arbitrary h € K as a superposition of normal node functions. (1.54) ts useful because f¢ provides @ spectral representation of A. This neans that 1h As An the domain of A, 90 that ANE X, then NP) fy(P) © F(R) and as amen = 10> Bow where (2) 18 defined by (1.41). This property is used in chapter 4 to solve the wave equation for A tin vhat followe the X-1im notation is sometimes suppressed for the sake of sinplicity, Sovever, incegrale such as (2.52) and (2.54) are alvoys to be interpreted an Htlbert space limita. Of course, for special choices of 1 Se ay happen chat the {nteprale also converge pointwise. However, even in these cases 1¢ Ls the mean square convergence that se the nore relevant te problens of vave propagation. To desctibe the general case where A has guided modes let Ny ~ 1 denote the mmber of diatiner guided normal mde fmetfons hy, Zor A where 25 My £40. Thon (1.52) still holds for every he sand in addition for every k such that 1 £k < Ny there exists a unique function fy & Ls(0,) such thee se) BG) = Let@y-rim [ TEGEET nao FO) 0 nak. x Moreover, the Parseval relation (1.53) is replaced by Whe = Bl “Sie (59) the = ILE gy to WEE e- lbas * 2 ep and the normal mode expansions asserts that the Limits2 3, SPECTRAL ANALYSIS OF SOUND PROPAGATION @..60) np = Hate | ayeney Rycerer sd ae noo estim [4G Rodan, 1k ee 7 Net 2.82) henge Y wel where 4f Ny ~ +9 then the Last sum 4s convergent in X, The representation (1.52), (.58)=(1.62) 1s a spectral representation for A in the sense that Lf h 4s am che domain of A then (1.57) holds and in addition @.63) Cand (P) A CIPLD By Ce) For L Sk my whore AGP) = ALDI) Yap) ‘The functions (_,9y 42. functions for A The normal node expansion for this family in defined by (1.52), (1.58)-(1.62) with , and fy replaced by ¢ and fh, The functions {oe Wistest**) form atl another complete cet. The normal moda expansion for ehis family is derived in §9. } form a second complete set of nornal mode $2. THE REDUCED PROPAGATOR A, ‘The Sturm-Liouville operater A, defined by (1.5) has a selfadjoint Fealtzation, also to be denoted by A, dn the Hilbert space an HE) = gie*O) ODE). the domats of hy, S9 che see an ray -H@ofe: 2 fro ger}. sue properties a) at 2 Bes where At is the adjoint of A, in X(R), ean be verified by showing that A, fe the operator in X(R) associated with che sesquilinear form A, in 3(R)2:3, Solutions of the Rauation a, = $9 2 (2.4) D(A, = LOD < CD =f (8a, valet as) Aon = | {Ea vs} toner, the spectral analysis of A will be derived below from that of A,. The ain stepa of the analysis are the following, First, conditions (2.1.4) and (1.1) ate used to construct che special solutions of a, = Ao defined bby conditions (1-12). Second, these solutions are used to construct an eigenfunction expansion for A,. The construction 1s based on the Weyi~ Kodaira-Titchmarsh theory of singular Sturm-Liouville operators. Finally, the expansion for Aj, and Fourier analysis in the variables x1, mj ate used to construct @ spectral representation for A. This mathod has been applied co the special cases of the Pekeris and Epstein profiles [8,26] vere explicit representations of the solutions of A, = Ab by means of slonentary fonctions are available. This the main technical advance in the work presented in thie chapter 49 the coastruction, for the class of density and sound speed profiles defined by (2.1.4) and (1.1), of solutions of Ao = 2 that have prescribed asyaptotic behavior for y > # and euttielent regularity in the parameters 2 and y to permit application of the methods of [8,26]. $3. SOWTLONS OF THE RAUATION A = 6b ‘The special solutions #)(yst43) G + 1,2,3,4) desertbed in $1 are constructed in this section, Analytic continuations of these fuiccions co complex values of 2 are used in 8 for the calculation of the spectral, fantly of Ay. Hence the nore general case of solutions of Ap = G9 with 6 € will be treated. ‘he equation 4, = £9 comot have solutions in the classical sense untess e(y) and ply) are continuous and continuously differentiable, respectively. A suitable clas of solutions is deseribed by the following definition in which ACG) denotes the see of ali functions thar are absolutely continuous with respect to Lebesgue measure in the interval Pepe Definition, A function § t I= (a,b) + C ks said to be a solution of ay 4,90) = -c2yHoG GG) 41D)" ~ vO} = cOO?3 SPECTRAL ANALYSIS OF SOUND PROPAGATION a the interval I (where @* = do/ay) 1f and only if GD © AC(E), p19" € ACCT) and (3.2) holds for almost ald y € 1. The following notation will be used fn the definition and construction of the special solutions #y(ysu55)- For each « > 0 Lo = tc | meg <4), 3 Ree) = fe | Re Ge, Re) = Re) O(c } ttm 5 2 Oo}. ‘The defimicions (1.10), (1.21) will be extended as follows. ag G0) = ERG) = G4) Ha/4 < arg aCby5) < 1/4 5 Reccemyny ah.) = £4,040 ond abies) = Gu? = oP ey) 3) anf < are ating) < mf 5 € L(y 4h) + -t ayn. the results of this section will now be Cormuiated. Under hypotheses (2.1.4), (1.1) on p(y), o(y) there ender Functions 6.9 $y RRR, x CUCODD UY RCOEID) + 6, 4 = 12s shore R= {u | u 2 01) such that for every fixed (44,5) © R, * CEM U RCCODD)» (F446) Ae @ solution of (1) for ye R and Je 1,2 ana3.3, Solutions of the Equation Ayo = cb 38 1G) = exp CaLGDe} LL + 0) on yo4, OG) Toye) = 0) alte) exp Ca} ucry}f2 + of] and bast) = exp (na) G.t)y [t+ 0) G8) _— eG) Ofc) = 0) ahs) exp Ca} @,cy)tL + 0) Sintlarly, there exist functions a9) 4 RAR, LOCI U RECN) 7G, F344, such thet for every fixed G5) © RL * C(oCE) U REMI, 64670840) ts a solution of G.I) for y@ Rand j = 3,4 and .GU.0) = exp fal(u,cyH{L + 0(2)] G19) yom ND oSlraw ed = OH GE) exp La! GHry)LD + 00) and Ge Grtc) = exp (o91G1,cy}12 + 0(1)] Gay yoo POD aE) = PDA HE) exp L4G, D9} + of} ‘The following three corollaries describe the dependence of the solu- tons $5 (sis5) on the parameters v asd 5. The FunceLons ¢,(ysus0) satisty ou tye ote & ofa g fone | ¢ been} for 3 = 1, 2 and enn Lote e fk «Uf tan [ce reocnos] 4 3 20 4. Moreover36 SPECTRAL ANALYSTS OF SOUND PROPAGARTON dye ota € | U too | ee econ), G14) ve U tao | ce eterno}. 1 ta oape cf ofr UJ (wo |e rem], 50 ou oteze efx L taro | ce Rtn] od cotollary 3.3. for each fixed (98) € RXR, the mappings G15) 6+ Gunes 67 eG) Home are analytic for = 3, FE LECH U REECE, 415) J #2 5S LCeGonny U RCC, 28s Ge KCoCmD U REDDER, 4, 6 LeeCanD U ReeCmpun**, whore R(«) AE = Rik) 9 fz | Atm g > 0). Corollary 3.4. The asymptotic estinates for 9, and 09} of theorem 3.1 bold uniformly for (u4£) in any conpact set T, such that for selec U £On0 | 2 € Llc URI, vi G27) pea teL) one [se cer o Reon), us \ sae Lf tao | cee » Coon). ib _ sea Une) | c€ beady) U Re}. 150 he special solutions 9,(y,u,5) are not, ta goneral, uniquely deter= ined by the asyeptotic conditions (3.7), (3.8), (3-10), (3.11). Indeedy Af Re @hGsA) > 0 (resp, Re Gi(u,A) <0) it is clear chat any multiple of2.3, Solutions of the Bnuation a9 = CF a , (resp. 41) can be added to 4, (resp., $,)- A similar remark holds for $y and @y. However, for «: kG € Ca subdominant solution Cone with ‘minimal growth at y = © or y = ~*) 45 unique. In partfeular, since Re qh(iig) > 0 for © L¢ece=)) 3.18) O48) 20 for GE RGc(=u) Re 42012) £0 for 5 © Rou) one can prove Corollary 3.5. the solution ¢ 4 uniquely determined by (3.8) for ali ce Léc)u) UR (eu). Similarly, $5 te uniquely determined by (2.20) for 5 © L(ot-")¥) U RCel—=)H), @ de uniquely determined by (3.7) in R(c(=)u) and $, is uniquely determined by (3.12) in R*CeC-)u). nen Re t= e2Cs0) ut Theorem 3.1 provides no information about the asymptotic behavior for y > #8 of solutions of A, = Gf- However, positive results can be obtained by strengthening hypothesis (1.1). The following extension of @ known result (14, p. 209} will be used in $4. Theorem 3.6. Assure that p(y) and c(y) sattafy hypothesis (2.1.4) and a8) [Floor - pe! ay cm Jy? foo ~ ee} ay <= Then there exist functions (3.20) Rx RFR JR, such that for every pe Ry the pair $,(y,u), $2(y.u) 4s 2 solution basis for Ae = conte, oxGnu) = 14 0a) en yore 7) HOw = 0 and baQaud * pldy{t * 0] (3.22) yas, eG) Gow) = 1+ oD28 3. SPECTRAL ANALYSIS OF SOUND PROPAGATION Laprange's formule for A, say be written Ps -teyygt (3.23) FP toa, - tay) Pore" Ona & LeHGD = IED) cn where (3.26) [49169 = oD") HOD ~ EET) HOD. In particular, if ¢ and ¥ are solutions of 6 ~ 6, AW = &Y om an interval E then [G61(9) ¥ const. on I and [6)](9) © 0 on I Af and only 4 and Y are Linearly dependent there, By combining these facts and ‘Theorem 3.2 one can show that 18, 686,0) O20 50,09] © -2 07 @) QhGL.0), 3.25 Tag G ust) 94GB] = -2 0? "90005 which imply Govoltary 3.7. The pate 4, (¥,150)s $2 (4,0) 48 solution basis for Ao = 56 For all (uss) € RL * (L(CODu) U R(elOD). Similarly, the pate 8,670.0, 6,(.u.0) 48 a solution basis for all (145) © R, x (L¢eC#u) U Reem). ‘This completes the formulation of the results of 83 and the proofs zt by an equivalent first order system. The latter can be regarded az a perterba~ will now be given. The method of proof involves teplacing 4,6 = tion of che corresponding lime ystems for y > 8, In this way integral ‘equations are established for solutions with prescribed asymptotic behavior for y + or y += and chose equations are solved by classical Banach space methods. This techaique for constructing solutions with preseribed asymptotic behavior 12 well know ~ see for example (5, p. 1408] aod (16, Ch. VIET. A Elese order systen equivalent to A,p = Gd. If 6(y) ts say solution of G2) on an interval I and if yO) = 007 (3.26) 2G) = POD BO” then ys Ye © ACCE) (ef. (3.2)) and3.3, Solutions of the Eauttion 4,¢ = 54 33 Wf) = eG) vO) aan WO) = poh? - 5 FGI] HO) for almost avery y € I, Thus the colum vector ¥(y) with components ¥, (9), ¥e(y) 1s a solution of the first order Lincar systox G28) VO) = MOL VOD where ° PG) (3.29) Moe) cb? - ce Fon) e Conversely, 1f ¥ € AC(I) {8 a solution of (3.28), (3.29) and ff 9G) = G9 then $ £5 a solution of (3.1). the solutions of Theorem 3.1 vill be constructed by incegestiag (3.28), (3.29) the Linty system for y > 4” ond its solutions. By replacing Cy), eG) tn (3.28), (3.29) by PC), €(=) one obtains the system 3.30) 9G) = MyDD HOD where ° 1) a Moise) = | Pepe = oe ° My(uig5) hes diseince eigenvalues 416145) Rl). The colums of nahtng) for Ge LC ud (3.32) BG = f°) ahh 0) eT) and aze corresponding eigenvectors. ence oa MyG45) BUYS) = BO DOLD40 2. SPECTRAL ANALYSIS CP SOUND PROPAGATION were yG.0) ° 3.) bone) = . ° Hayne) system (3.30), (3.31) may be integrated by the substitution 6.35) v= Bu,Oe Tt follows that 2"(y) = DUu,5) 29), whence 210) =e, exp fahu,sdy? 3.36) [2 +e ew baG.00 and therefore VG) * e, emp (A,Gndy} +e, exp GahGnddy? ean HO) = 8) aL Oe, exp Cah OL0y) - C2 exp fa} Gieod¥)) where cy, cg af constants of integration. Application of perturbation theory, System (3.28) may be regarded as a perturbation of the Limit eystem (3.20). ‘Thus ££ N(y,3)0) As defined by G38) MOE) = Myon + NOD then ° a) 39) Non = vw a6) +6 4,6) oy where 410) = 9) = pO (3.40 gO) = 6G) = OD oo) a0) «87 @) FED -3.2, Solutions of the Bqvation Ayo = o> a ote that each of these functions 48 in 1,(%9,%) for every yy © R. Por 4,9) this is part of hypothesis (1.1). For a;(y) and a(y) te follows from (2.1.4) and (1). For example, one can write (3.41) a.) = (977) pG)G) = PO)? which exiles a: a9 a product of a bounded measurable function and a Function im ty (¥ys")« On cosbining (3.28), (3.38) and naking the substitution (3.35), one Finds that (.28) 1s equivalent to the system Gan 210) = DOS) 20) + BOVE) 2G) where aay ECysMyE) © BYE) MEYSHAE) BOE) hhas components that axe in Ly(yo,°). Solutions of (3.42) will ve constructed which are asymptotically equal, for y > 4, co the solutions (2.36) of at = DUE)e. Proof of Thoorem 3.1. ‘The remaining eases can be proved by the sane method. Solutions of (3.42) The proof will be given for the function # only. are related to the corresponding solutions of (3.1) by dea te 3.48) ppt = 0 ©) aha, ~ a2) a = hGH 2)- ‘hus WILE be a golution of (3-1) thet sacisfies (3.7) £f 2 1s @ solution of (3.42) that satisfies (a5) 21 = exp Calyinys 2g = exp fatyln, and (3.46) iG) = 1+ 01), m6) = 01) for y +. Bquations (3.42) and (3.45) Implya2 3. SPECTRAL ANALYSIS OP SOUND PROPAGATION Bry my + Bia tye G87) Wye -Pahny +E my + he me and hence by tntogration p nore ef 209 nore (3.48) re ' ay) = exp C2 aly) eg + |) exp 2 ally 9D) £5077 nyorayt % where ©, C2, You ¥; axe constants and the summation convention has been used J is summed over j= 1,2). Gonseruetion of $1 for C€ Le(ey 5 © L(x). Thus te construct a solution of ( By (3.18), Re q}u,c) > 0 for alt .47) char satiaties (3-46) te 4s natural to choose c, = 1, ej * 0, yy = +” and yy finite im (3.48). This gives che system of integral equations f gg say MO) = E= | B56 nore Gan y nor = PP ese G2 go = 99) By) ayorrey" » Ke da msturel to study system (3.49) 4m the space 3.50) X= oRCLyy =) 0°) bf two-component vector functions of y shose components are contiruous and bounded on y; {y <*. X 48 © Banach space with nor asp In} = sup Cin G91 + Ino). ye ‘the system has the form here (9) and 4° are column vectors with components (ni(y).n2(y)) and (4,0), respectévely, and the atrix kemel K(y.y') is defined by