JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. C4, PAGES 8663-8679, APRIL 15, 1997 Generation of edge waves in shallow water T. C. Lippmann Centerfor CoastalStudies,ScrippsInstitutionof Oceanography, La Jolla,Califomia R. A. Holman Collegeof OceanandAtmospheric Sciences,OregonStateUniversity,Corvallis A. J. Bowen Departmentof Oceanography, DalhousieUniversity,Halifax, Nova Scotia,Canada Abstract. Theoreticalgrowthratesfor resonantlydrivenedgewavesin the nearshoreare estimmedfrom the forced,shallowwaterequationsof motionfor the caseof a plane slopingbed. The forcingmechanism arisesfrom spatialandtemporalvariationsin radiationstressgradients inducedby a modulatingincidentwavefield. Only thecaseof exactresonance is considered, where thedifference frequencies ,andwavenumbers satisfytheedgewavedispersion relation(thespecific carrierfrequencies arenotimportant,onlytheforceddifferencevalues).The forcingis ex,'unined in theregionseawardof thebreakpoint andalsowithinthefluctuatingregionof surfzonewidth. In eachregion,theforcingis dominatedby the cross-shore gradientof onshoredirectedmomentum flux, exceptfor largeanglesof incidenceandthe lowestedgewavemodes.Outsidethe surfzone, thespatialandtemporalvariationof theforcingis determined by considering theinteractionof two progressive shallowwaterwavesapproaching thebeachobliquely.In the surfzone,incident waveamplitudesareassumed to be proportionalto the waterdepth. Thusinsidethebreakpoint, radiationstressgradientsareconstantandno forcingoccurs.However,at the breakpoint,gradients arisingfrombreakingandnonbreaking wavesareturnedon andoff (like a wavemaker)with ti•nescales and lengthscalesdeterminedby the modulationof the breakerposition. The tbrcingin thisregionis stronger,with inviscidgrowthratesresultingin edgewavesgrowingto the sizeof the incidentwavesof the orderof about10 edgewaveperiods,a factorof 2-10 timeslargerthanin the offshoreregion. Usinga si•nplepar,-uneterization for frictionaldamping,edgewave equilibrium mnplitudes arelbundtodepend linearlyontheratiotan]•/Ca , wherefl isthebeach slopeandCais a bouomdragcoefficient. For tan•/Ca about3-10,equilibrium amplitudes can be as •nuchas 75% of the incidentwavesover most of the infragravityportionof the spectrum. Whentheforcingis turnedoff, thesedissipation ratesresultin a half-lifedecaytimescaleof the orderof 10-30 edgewave periods. Introduction 1985; Oltman-Shay and Guza, 1987; Howd et al., 1991; Since the initial observations of Munk [19491 and Tucker [1950] much effort has been aimed at understandingthe origin and importance of low-frequency (relative to wind waves) surfacegravity wavesin shallow water. Field data obtainedon natural beaches haveshown thatlong-period (O(102-10 3 s)) infragravity motions often dominate power spectra in the inner surf zone and swash, particularly during storms when incident wave heightsin shallow water are severelylimited by breaking [Huntley, 1976; Huntley et al., 1981; Holman, 1981; Thornton and Guza, 1982; Holman and Sallenger, 1985; Sullenget and Holman, 1987]. The data also indicate that infragravity surface gravity waves are predominantly composed of edge waves, longshore progressive waves trappedto the shorelineby refraction,together with a smaller componentof leaky waves, normally reflected waves which escapefrom the nearshoreinto deepwater [Guza and Thornton, Copyright1997 by the AmericanGeophysicalUnion. Papernumber96JC03722. 0148-0227/97/96JC-03722509.00 8663 Herbets et al., 1995b]. The existenceof edge waves in nature has been known for some time. Bowen and (;uza [1978] and Holman [1981] suggest that the growth of edge wave amplitudes, an, is determinedby the coupling between the nonlinear forcing, F(a1, a2, x), arisingfrom an interacting incidentwavefidld andthecross-shore edgewavewaveform,On(x), (1) rYe f F(al,a2,x)On(x)dx g o where al and a2 are incidentwave amplitudes,rYeis the edge wave radian frequency, g is gravity, t is time, and x is the cross-shorecoordinate. Growth occurs when the forcing pattern is not orthogonalto the edge wave waveform. The details of the forcing are describedby the spatially varying form of F. In this work we derivean analyticexpressionfor F and numerically integrate the coupling integral to obtain an estimatefor the growth rate. The principalforcingis believedto be derivedfrom pairsof incident wind waves, which produce spatial and temporal 8664 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW variations in radiation stress gradients of the same timescale and lengthscaleas the edge waves. Outsidethe surf zone in intermediate and shallow depths, forcing occurs through a typical second-ordernonlinearinteraction[Gallagher, 1971; WATER and Mei [1981], they assumeconstantbreakingcriteria so that modulations in incident wave heights are manifested in fluctuations in the width of the surf zone. However, their in specifying the forcing functions all the way to the shoreline. Here we assume that amplitude variations are eliminated by saturationand no oscillatory forcing occurs. However,thereis a time dependentmomentumflux inducedby the modulation in breakpoint position associatedwith the variation in incident wave height [Foda and Mei, 1981; Symondset al., 1982; Symondsand Bowen, 1984; Schaffer, model is limited to only two dimensions (no inclusion of longshore variability), precluding the possibility of edge wave forcing. Nevertheless, the results of Symondset al. suggestthat edge wave generationby temporal variations in surf zone width is possible. Schaffer and Svendsen[1988] retain the ideasof Foda and Mei [1981] and Symonds et al. [1982], by allowing fluctuations in both breakpoint positions and incident wave energy inside the breakpoint, although they limit their 1993]. discussion to the two-dimensional case. List [1992] extends Bowen and Guza, 1978]. Inside the surf zone, difficulties arise Outsidethe surf zone (referredto herein as offshoreforcing) this mechanism in a numerical model to include arbitrary the generationmechanismevolves from the second-order topography and further includes incoming bound waves, forced (or bound) wave generated by wave groups in crudely•nodeledempiricallyusingfield data,as do Schafferand intermediatewater depths[Biesel, 1952; Longuet-Higginsand Svendsen [1988]. They forego a more theoreticalapproach Stewart, 1962, 1964; Hasselmann, 1962]. In two dimensions, the bound wave is, in principal, released at breaking and reflected at the shoreline; thus the incoming and outgoing waves form a complex pattern in the cross-shore,however, without the longshorecomponentnecessaryfor edge wave since it is not clear what the correct boundaryconditionsare for boundwave dynamicsin the shoalingand breakingregion, although it is expected the forced motions become nearly resonant in shallow water [Longuet-Higgins and Stewart, 1964; Okihiro et al., 1992]. Schaffer [1994] usesa WKB (optics) approachto describe generation. This idea was extendedto three dimensions by Gallagher [1971], who showed that nonlinearinteractions the incident waves and drives the edge waves with radiation between incident wave pairs could producelow-frequency stressmodulationsdeterminedby fluctuating amplitudes(i.e., oscillationswith a nonzero alongshorewavenumber. Bowen and Guza [1978] generalized Gallagher's model to include forcingof all modesand showedwith laboratoryexperiments that resonantresponsewas greaterthan forcedresponse.This same conclusionwas reachedby Foda and Mei [1981] in a fourth-orderWKB expansionof the momentumequations. In the surf zone, forcing occurs because an interacting incidentwave field producesspatial and temporalvariationsin locations of the position at which a wave breaks (herein referred to as surf zone forcing). Momentum is transferred from incidentto lower frequenciesthroughwave breaking,and the interactionsagain lead to fluctuations in the flow field with timescalesand spacescalesof the order of wave groups. This idea was first explored as a generatingmechanismfor long waves by Foda and Mei [1981] and Symondset al. [1982]. As will be shown later, there is also a contribution from nonbreakingwaves within the region of fluctuatingsurf zone width. In the work by Foda and Mei [1981], low-frequencywaves are generatedby wave-wave interaction assuminga fixed breakpoint position for all waves. This allows incident modulationsto progressto the shoreline,and edge waves are forced everywhere. In nature,however,the initial breakpoint of individual waves is not constant through time, nor is the breaker line uniform along the beach. Under most conditions, temporal and spatial variations in the width of the surf zone occur on large timescalesand space scales,of the order of infragravity scaling [Symondset al., 1982]. Although the basic physicsinshore from the breakpointis doubtful, Foda and Mei do consider resonant edge wave growth, also consideredhere, and thus a brief comparisonwith their results is made later. Symondset al. [1982] considerthe problemof long wave forcing by time modulationsin surf zone width for the caseof a planeslopingbed and later includedinteractionswith barred topography[Symondsand Bowen, 1984]. In contrastto Foda the two incidentwaves are assumedto approachfrom the same direction). Numerical solutionsare considerablycomplicated by an angularbottomwhich necessitates a matchingcondition at a corner in the profile. Furthermore, Schaffer's offshore profile is horizontal; thus waves propagatingoffshore beyond this "shelf" are no longer refracted, thus limiting the possible edgewaveswhich can exist. Low-frequencyresponseis found through numerical calculation, with resonant modes determinedcritically by the single-incidentwave angle and the offshoreprofile configuration. This resultsin a very specific set of results only valid for theserestrictedset of conditions. In this paper we present a mechanism for the resonant forcingof edgewavesin shallowwater throughmodulationsin radiation stress gradients. We consider two interacting shallow water incident waves whose difference wavenumbers and frequenciessatisfythe edge wave dispersionrelation. The total forcing is found by integratingthe growth rate equation froin the shoreline to deep water (relative to wind waves). Contributionsto the forcing from inside and outside the surf zone are examined by separatingthe forcing integral at the breakpoint. We considerthe simplestcaseof a planar beach profile, where forcing in the surf zone is determinedby a time and spacedependentfluctuationin the width of the surf zone. In the next sectionwe review the theoryfor edgewaveson a plane sloping beach beginning with the forced shallow water (depth integrated),linearized equationsof motion, leading to an analytic expressionfor the initial (undamped) edge wave growthrate. Model resultsare then presented,comparingthe relative contributionsto the forcing from the componentsof the radiation stressand also the strengthof inviscid growth rates in the surf zone and offshore regions. Resultsare then discussed in terms of the validity of model assumptions, parameter sensitivity, and implications in field situations (spectralforcing). The effect of a linearizeddissipationterm on edge wave growth is discussed. In Appendix B the extensionof the calculationsto an incident wave spectrumis briefly discussed. LIPPMANN ET AL.: GENERA•ON OF EDGE WAVES IN SHALLOW WATER where L n is the Laguerre polynomial of order n. The approximate shallow water edge wave dispersionrelation is given by [Eckart, 1951] Model Edge Wave Theory The forced, shallow water (depth integrated), linearized equationsof motionincludinga linearizeddissipationterm are [Phillips, 1977] 2 cye- gke(2n+1)tanfi Growth m + 3t X-b--fxph - Sxx+ •+ Sxy+ - a, gTfy ph - •,u 3y xy Tyy s" -•v (2) 3t (h.)+ (8) Rates We seek an expressionfor the time rate of change of edge wavemodalamplitudes, or growthrate, 3an/3t. Substituting (3) (6) into (5) and allowing an to be a slowly varying function of time give 2a n 3a n -anZ(qJn)]e -ire -F and the continuityequation --+ 8665 (9) 3--'•qn +i2tYe 3y hu)-0 (4) where g-3xlh3•-•l+(CYe2-ghke where x andy are the cross-shoreand alongshorehorizontal Cartesiancoordinamswith x positiveseaward(for a right-hand systemwith z positiveupward),u andv are the corresponding life- key- Getß horizontalcomponentsof velocity, • is sea surfaceelevation, • is a friction coefficient (discussedlater), h is the still water In (9), and in subsequentequationsfor the growth rate, we depth,andp is the densi• of water. The equationshavebeen have droppedthe complex conjugatefor brevity, thus later we averaged over an incident wave period so that the time will take the real part to obtain the magnitudeand phase. The dependence is on infragravi W andlongerscales.Spqare the function Z(cpn) is the homogeneousequation for shallow radiation stresses of short (incident) waves introduced by water waves and vanishes for the case of resonance considered Longuet-Higginsand Stewart [1962, 1964] and describethe flux of pth directedmomentumin the qth direction. Combining (2)-(4) and temporarily dropping the damping term yield a single second-order(inviscid) equation in sea here. If we assumethat the edge wave growth rate is slow relative to its period, then the first term describing the accelerationin growth can be neglected. Thus an equationfor the initial edge wave growth rate is surface elevation [Mei and Benmoussa,1984], i2o'•,c-•--c),•e-iV•' -F (10) c-)271 c-)t 2 p• ax+ ay +•yy(ax+ ay (5) (describedlater), (ai)o, the normalizedgrowthrate becomes = Fxx+ Fyx+ Fxy+ Fyy= F whereFxx,Fyx,Fxy,andFyyarethecomponents of thetotal forcing, F, correspondingto the four secondderivativeson the right-handside. The forcingand frictionaldissipationare assumedto be of secondorder, and we ignore their effects on both the wave solutions and the dispersionrelation (which couldbe significantif the forcingand friction are large). For a planebeach,h= xtanfl, where/8 is the beachslope, the homogenous case(free waves)is satisfiedby edgewavesof the form [Eckart, 1951] rln(x,y,t) - •-anc)n (x)e -i(l•y-cr•t) +(*) 1 e3a• =(ai)ogtanfl(2n+l) -2rri I F0,•dx (ai)of •gt e-iv/• (11) 0 in which thenormalizing factor [•cpn2dx - 1/2k ehas been used.Thecoupling integralexpress4s therateat whichenergy is transferred from (6) complex conjugate of theprevious term,andi-4Z-f. The cross-shorestructureof the edge wave waveform, qS,•(x), is given by (7) the incident waves to resonant lower- frequencymotions. The inverseof the magnitudeof (11) is the number of edge wave periods necessaryfor the edge wave to grow to the size of the primary incidentwave. Coupling Integral where CYe=2nfeand ke=27r/Le are the edge wave radian frequencyand alongshorewavenumberOreandL e are the edge wave frequency and wavelength), n is the edge wave mode number,an are the complexmodal amplitudes(which can be resolved into a magnitude and phase), (*) indicates the c),• (x)- e-k•XL,• (2kex) The cross-shoredependenceis eliminated by multiplying by qS,• and integratingfrom the shorelineto infinity in the crossshore direction, as in (1). Incorporating the dispersion relation and dividing by the primary incident wave amplitude and Incident Wave Amplitudes In deep water, resonant interactions are quartic and only forced waves are produced by triad interactions. In shallower water, resonant triad interactions can occur, producing variation in radiation stresswhich can match the edge wave dispersion relation. This spatial and temporal modulation gives rise to the possibility of edge wave growth. Although the modulationscalesremain the same, the forcing function F, which dependson wave amplitude, varies as waves shoal in intermediatewater and eventuallybreak in the surf zone. As a consequence,there are two distinct forcing regions separated by the contourof the breakerline, xb(y,t) (shown graphically in Figure 1), and the coupling integral can be written 8666 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER OffshoreForcing Breakpoint •b '4'6•b I No Forcing Shoreline !•Ie ---keY-- (Je t Figure 1. Graphical representationof the edge wave forcing regionsrelative to an oscillatingbreaker line. Thevertical axisis Z= cre2x/gtan,B andthehorizontal axisis gre=key-Get.Theoffshore forcing region is defined as seawardof the •nostseawardbreakpoint. Inside the minimum breakpoint,no forcing occurs. The rangeof forcingby a modulatingbreakpointpositionis a functiono.__f incidentwave modulation,8. The regionswhere waves are alternatelyshoalingand breaking(between Zb and the oscillatingbreakerline) are labeled I and II, respectively. Go x!,(y,t ) the mean wave amplitudeand the modulation(aboutthe mean) is determinedfrom a2 [Symondset al., 1982]. Becauseof sharpdifferencesin the shoalingand breaking regions, the amplitudes of the incident waves are treated differently. In the offshoreregion the amplitudesare described by progressive shallow water waves over a plane sloping bottom; thus an interactingwave field has modulationsall the way to the breakpoint. Wave amplitudes in this region are referenced to a convenient location (e.g., the breakpoint). Inside the surf zone, wave amplitudesare stronglyattenuated shorewardby breakingand are describedas a linear functionof local depth, IF(al,a2,x)cPnCx)dxIFsz(al,a2,x)cpn(x)dx o o + IFoff(al,a2,x)cpn(x)dx (12) xt,(y,t) whereFszandFor f aretheforcinginsideandoutside the breakpoint, respectively. The forcing functions are determinedby the temporal fluctuations of radiation stress gradients,which are in turn determinedby •nodulationsin incident wave amplitude. It is the slowly varying breakerline, xb(y,t ), that creates first-order fluctuations in the surf zone coupling integral. If we assume that the breakpoint varies sinusoidally with gre tilne and lengthscales(describedlater) andfurtherseparatethe couplingintegral at •'b and expandin a Taylor'sseriesabout 1 ai =•-Th x <xb (15) where ), is a constantof O(1) [e.g., Thorntonand Guza, 1982]. For a plane beach,referenceamplitudescan then be given by '•t,, thento first order,(12) canbe written(AppendixA) 1 (ai)o=• ?xb tan/3 (16) where x•, is the position of the breakerline, a function of y 0 and t. +IFoff(ala2,x)cPn(x)dx Once the incident waves break, •nodulations in incident (13) amplitudes x/, Thus the first-order contributionsto edge wave growth occur through primary interactions from both breaking and nonbreakingwaves within the fluctuatingregion of surf zone width and through cross-interactionsof a pair of waves seawardof the average breakpointposition. In the following, the wave field is considered to be co•nposedof two waves,a primary wave with amplitudea I and longshorewavenmnberand frequency(kl, fl) and a secondary wave with mnplitudea2 and slightlydifferentwavenumberand frequency(k2, f2)- If we choose a2 = •a I vanish and the forcing becomes constant shoreward. The possibility for edge wave forcing occurs because the breakpoint position varies on timescales and spacescalesassociatedwith the incident beat. Thus the limits of integration in (12) for the surf zone integral will have modulationswith edge wave timescalesand spacescalesthat are deterIninedby the fluctuatingbreakpointposition(Figure 2). The •nagnitudeof the modulationis determinedby the maximum and •ninimum breakpoint amplitudes, ama x = aI + 8aI and amin = aI -c•al, respectively,so that the cross-shore rangeover which forcing occursis definedby (14) where • << 1 is a constant,then the primary wave determines whereZ'b is the meanbreakerposition. (17) LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER 8667 within the fluctuating region of the surf zone, which will the same frequency and wavenumber as the .'''''' I ''""""'f(ao(l+•)) oscillate at fluctuatingsurf zone boundary(given by the tiine varying a=Th/2 • z / f(ao) I contour of the breaker line). We choosethe velocity potential for progressiveshallow water waves over a plane sloping bottmn after Stoker [1947] I....... i ....... f(ao(1-fi)) • x•(l+fi) msl ,.x Figure 2. Graphical representationof the envelop of incident wave ainplitudes over a plane sloping bed as a functionof cross-shore distance.Surf similarity( a = 7h/2) is assumed once waves are breaking so that inodulations in incidentamplitudesproducevariationsin surf zone width. The vertical (elevation) and horizontal (cross-shore distance) diinensionsare arbitrary. (I)i --aig[Jo(Xi)cosyt'i+Yo(Xi)sin I//i] (22) where Xi_2( cri2x )112 gtanfi IIIi - (kiy- crit ) and Jo and Yo are the zero-order bessel functions of the first and secondkind, respectively,ki is the alongshore cmnponent of the incidentwavenumbervector, and the subscripti is an indexwhichrefersto individualincidentwaves. For Xi >3 (equivalent to H>-•(Ttanfi)2/2,so validfor nearlyall Parameterizing the Forcing oceaniccases),the besselfunctionsinay be approximated by Following Phillips [1977], the general forin of the radiation stressis given by ,q'pq --[3__jh C_)W 3X q (S pq Z-- [Stoker, 1947] Cos(X/--•)(23a) Jø (Xi)= '•i 2/¾2 • sin(X i-•) (18) where x is the horizontal spacevector in which subscriptsp and q denote either horizontal coordinate, (I) is the velocity potential of the incident waves, fi is the mean pressure,and Introducingcomplexnotationyieldsthe followingform for (I)i (hi--• aig(--2)l/2ei(X'-%+ cr i •,•rx i +(*) (24) 5pqis theKronecker delta.Themomentum balance hasbeen spatially averagedin the directionof the wave crests(allowing separation of the turbulent and wave terms) and temporally averaged over the incident wave period (allowing mean or slowly varying properties to be evaluated), denoted with the overbarsin (18). In shallow water •--pgz and assuming 7]= 0 at secondorder, (23b) The form of our velocity potentialis only strictly valid for waves approachingthe beachshorenormally. However, Guza and Bowen [1975] show that nonnormalanglesof incidence have only small effectson the forin of the velocity potential due to refraction decreasingthe angle of incidence and show that where shallow water solutions are valid, the solutions for Spq •ph•x• •x•+ •pq • • We consider the bichromatic case where (19) the incident wave field is composedof two discrete wave trains with slightly differentwavenumberand frequency, • - • + •2 shore-normal and obliqueincidenceangleare nearlythe same. Substitutionof (24) into (21) producesterms with sum and differencewavenumbers andfrequencies.The sumterms(highfrequencyforcedwaves) are unimportantto the generationof infragravity motions [Bowen and Guza, 1978] and are not consideredfurther. The difference terms describethe long timescalesand space scalesof the incident wave modulation and allow for the possibilityfor edge waves at infragravity (20) where the subscripts1 and 2 refer to the two incident waves. frequencies Substituting• into (19) produces Spq -Spq(*l*l)+2Spq(*l*2)+Spq(*2*2)(21) The first and last terms represent the self-self interactions which generate•nean flows and harmonics. The harmonicsdo not contribute to low-frequency forcing in either region and are therefore neglected. Letting the amplitude of a primary wave •i be much larger than any secondary waves •2 (equation (14)), then •1•1 terms are of O(1), while cross- interaction terms•1•2 areO(8) and•2•2 termsareO(82). The stressarising from •i•i interactionswill generate a setdown outsidethe breakpointand a setupinside the breakpoint Of = CY1-CY 2 (25a) [kfl-Ikl -k21 (25b) and where k! and k2 are the longshorewavenumbersof the incident wavesand the subscriptf indicatesthe forced differencevalues associated with the incident wave inodulation. For resonance to occur,(25a) and (25b) inustsatisfythe edgewave dispersion relation (8). Incident wave angles are asstunedto follow Snell's law for wave refraction and are chosenat the breakpointwhere the 8668 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER linear shallow water phasevelocity is simply a function of the local depth, c-•, _ sothat ki •yi2 ' g = sin O•o = tYisin(o•i)o (26) X4o X• 5 e +sina 3(2n+ 1)X• 222sin 3-Xe -i3X l+sina 12(2n+ 1)2Xe o• 1sin a2+1)(29) ß where ooindicates thedeep water condition and(o• i)oarethe wave angles of the individual waves taken at some reference depth, ho. If we further referenceai to the breakpoint,where ho = hi,, thenproductsala2 are givenfrom (24) by =7r (Sei(Xe-U/e) p (gtanfi o The three terms are derived from the componentsof the interaction radiation stress fromFxx, 2Fxy,andFyy. The three componentsof the interactionforcing are compared later. Contributions to edge wave forcingby nonbreaking wavesseawardof the breakpointbut within the fluctuating ]1/2 where the subscripto refers to values at the referenceposition region of surf zone width are consideredin the next section. Inserting(29) into the couplingintegralin (13) resultsin a complexexpression for the undmnped response of edgewaves andevaluating the appropriate derivatives give an analytic by oscillatingforcingseawardof the •neanbreakpoint. After (e.g., the breakp.oint). Now substituting (24) into (19), incorporating(26)-(27), form of the radiation taking the real part, expressionsfor the normalized initial stress growth ratemagnitude, Goff,andphase, Ooff, aregivenby (28) Spq =• (7off= 1 •an= 3•'•-• 1/2(30) fe(al)o at 4(2n +1)•(•bb)3(pn2+Qn2) where 3 2a• I lpq= 2a• 2al2+1 where Pn- I (32cos Xe- B1sin Xe)q),•XedXe JPq+a2 2ala 2+1 XeA -•(Xl 0•1X20•2) Kpq= •(Xlal _X2a2) Qn = I(BicosXe +B2sinXe)q)nXedXe Xb 0 1 B1 = 3-Xe 2+sino•1q-sina 2 - 2sina• sina 2 + 1 X•5 3(2n+l)Xe 2 12(2n +1)2 X• 3 XiX2 O.e2 x/¾2 Xe=2 gtan• B2- Xe 4 Surf Zone Response From Breaking and Nonbreaking Waves The cross-shoreintegral parameterizingthe surf zone contributionto the total edge wave forcing rangesfrom the ß 1• interactions,O(•5) termsare from •2 interactions, shorelineto the breakpoint,0_<x_<xb. FollowingSymonds et al. [1982], we do not allowmodulations in waveamplitude andO(•52) terms arefrmn•2•2 interactions. insidethe mini•num,•nostshoreward breakpoint by choosing Using typical values found in nature, f• =f2 =0.1 Hz, xb--lOOm, and tanfi=0.02, we find A is O(10 -3) at the ¾ to be constantfor all waves. Thus for a plane beach, •nodulationsin breakpointamplitudesgeneratespatial and breakpoint, and therefore terms in (28) containing A contributenegligibly to the interaction and are neglectedin temporalvariationsin the width of the surf zone. Symondset long wave the following. This implies that althoughinteractionsof the al. usedthis as the basisfor their two-dimensional incidentwavesdeterminethe responsefrequency,the specific model. Wave mnplitudesare expressedin terms of cross-shore carrierfrequenciesare not important. andsin(o•! )oandsin(o• 2)ohave been abbreviated o•!ando•2, respectively,for simplicity. In (28), O(1) terms are a result of Response Outside the Surf Zone The forcing function outsidethe mean breakpointposition is found directly by insertingterms of order •5in (28) (arising from cross-interactionradiation stresses)into the right-hand side of (5) and evaluatingthe secondspatialderivatives, breakpointposition using (16). The surf zone forcing functionis then foundby inserting(28) into the right-hand sideof (5) andevaluatingthe spatialderivatives.Sincewe are only interestedin first-orderapproximation, we needonly retainthe pri•naryself-interaction terms(AppendixA). Inside the breakpointfirst-order radiation, stressgradientsfor breakingandnonbreaking wavesaregivenby LIPPMANN ET AL.: GENERATION 3Bo.•t,(Xe )o3 •Sxx breaking -- 2Xe 3x OF EDGE WAVES IN SHALLOW 3x- 4Bøx sin ctl Ibreaking 0Syy =0 3y integrationare symmetric;hence z'l =-z' 2 --z'. Cm= (32) m=l where • 'r2 iTM) 35pq e_imu/, r!(y,t) 3Xp 1 where z'1 and z'2 define the interval over which forcing occurs and (*) indicatesthe complex conjugateof the previousterms under the summation. The form of (32) is given for the breaking waves only; nonbreaking waves in the fluctuating region of surf zone width are rc out of phase with (32). Symondset al. presentan extensivediscussionfor finding the Fourier limits of integrationfor the two-di•nensionalcase. We extend their ideas to three dimensionsin the following. The mean breakpointmnplitudeis defined by the larger wave, and the fluctuation about the mean is defined by the amplitude lnodulation,B, suchthat by usinglinear superposition of two sinusoidal 38pqsinmz 3p m'r (36a) re=l, 2.... (36b) The first term in the series(Co) representsthe mean value over all •e space and subsequentlyvanishes when second derivatives are taken. For m > 0, the series consists of a primarymodulation(m = 1) and its harmonics(m > 1). Thusto lowest order (consideringonly forcing contributionsfrom the primary) the total forcing from breaking and nonbreaking waves in the fluctuating region of surf zone width is found (after evaluatingthe secondspatialderivatives) rsz--- --123ø ei(-'l•'•-IF") p (37) 1(3)sinx sintzlXt,2(cos•: sin•: sinx)} waves. Becauseradiation stressgradientsare discontinuousat the breakpoint, we represent (31) as a co•nplex Fourier Series. Extending Symondset al. [1982], Using (35), are Co = const (31d) depth is eliminated using (26) (the incident alongshore wavenumberis conservedacrossthe nearshoreregion). Since we are taking T to be constant,all modulationsinside the initial breakpoint vanish, and each component of (31) is constant. In essence,variationsin the forcing are defined by a wave maker type problem, where the forcing is turned on and off at the initial breakpoint(as in the work by Symondset al. [1982]). Periodic fluctuations in radiation stress gradients arise from fluctuations in the position of the initial breakpointon timescalesand spacescalesof the modulation. Alongshore modulations allow for the possibility of edge X--' Xb coefficients (31c) where Bo=pg(ytan fl)2/8. Thedependence ofwaveangle on 3Spq +(*) 3Xp =Cø +ZCmeiU/ (35) Symonds et al. [1982] show that for small • the limits of (3lb) the Fourier •)S xy 8669 =cosx/,-xt, = (31a) nonbreaking 3Syx =0 WATER ß 1+• z' 6(2n+1)z' z'2 •-i The first andsecondtermsarisefrom Fxx forcing by breaking and nonbreakingwaves, respectively. The relnaining three termsarisefrownFxy forcingby breaking wavesonly. The componentsof the forcing are comparedlater. The initial growth rate is found by inserting (37) into the first coupling integral and evaluatingin the salne manner as in the offshore region, except that the limits of integration are now given by (17). Taking the real part gives expressionsfor the magnitude, G sz, and phase, Osz,of the initial edge wave growth rate in the fluctuatingregion of surf zone width Gs z= 1 3a,• _ 127 2 2 fe(t,,'••o 3,- (2n +1)•t, 2(Mn +Nn )1/2(38) where X2 Mtl • waves f (DlCO S•-•-•b - D2 sin•'•-•b )cp nXt,dX t, X• X2 at,a• I1+tScøs!•,t,•e)] (33) Ntl • I (D2 cos•t, +Dlsin•X--•)cpnXt,dXt, X• whereXt, is themean X e value corresponding to x= •t,. We D1 = can express(33) in terms of breakpointpositionsusing (16) (34) whereZt, is the meanbreakpointposition. The upperlimit of integrationin (32) is determinedfroln the argumentof (34) sina•Xb2 sinz 6(2n+1) z ( 3•)sinz' z' Xt,2sintzl(COSZ 6(2n+1) z'' sinz') •-2 D2= 1+ . whereX1= Xt, - 8Xt, andX2 = Xt, + 8Xt,. 8670 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER Results surf zone mechanismsis shown in Figure 3. The magnitude The resultsare presentedin two parts. The first focuseson andphasefor the caseof (a 1)o= (ø:2)o= 10ø are plottedas a therelative magnitudes of FpqrPn whichcomprise thesecond- function of nondimensional35b(the legend defining individual orderforcingfunctions(29) and (37) from theoffshoreandsurf forcing componentsis given in the figure caption). For small zone regions,respectively. The secondpart comparesinitial 35b,Fxx dominatesthe forcing in each region. At higher35b, growthrate amplitudesandphasesin the two regions,(30) and longshorefluxes become relatively more important. Because (38), and for the total combinedforcing. Sensitivityto mode the forcing decays rapidly as 35b increases (for mode 0), number,n, and wave angle, (o•i)o, are examinedin eachpart. consideringonly Fxx gives a good first-orderapproximationto Model parameters used in the following analysis are the overall forcing. The forcing at higher modes,n- 1, 2, and tanfl-0.01, 7-0.42, 6-0.1, andxb =100 m, chosenas 3, is shownin Figure4 for (al) o =(o:2)o =10ø. Only the For convenience, incident forcingmagnitudes areshownsincethephases of Fpqarethe reasonable for field situations. wave angles were chosento be identical at the breakpoint. stoneas for mode0 (Figure3). The behaviorof the forcingis similar in each region, with Fxx dominatingall modesdue to Resultsare plotted againstthe nondimensional variable thedependence of Fxy(and Fyx)on(2n+l) -1 andFyyon 35l,= • g tanfl (39) The range of values plotted, 350<10, covers the range of typical infragravityfrequenciescommonlyobservedin nature. Forcing The Components relative strength of the components of F(al,a2,x)cPn(x) for mode0 edgewavesby the offshoreand 10-' (2n+l)-2. Since Fxxis insensitive to modenumber, the forcing at higher modeswill be nearly the same as for the low modes, suggestingthat all modes should be about equally forced. In general, the amplitudeof the forcing function in the surf zone is an order of magnitude larger than in the offshore region. Thus it is expectedthat this part of the forcing will dominate the growth rate. Since the forcing function in the surf zone is nearly independentof Zb (to first order), the Surf Zone Offshore 104 10'4 10-5 10-5 10-• 10-6 , 10-7 0 I 2 3 4 5 6 7 0 1 2 3 4 5 6 7 10 0 1 2 3 4 5 6 7 8 9 10 10 0 1 2 3 4 5 6 7 8 9 10 180 135 90 -45 -90 -135 -180 8 9 Zb Figure 3. Contributionsto the total edge wave forcing function, Fen , for a mode0 edgewave, (left) from the offshoreregion and (right) within the surf zone. The componentsof the forcing arisingfrom Fxx (dashed lines),Fxy(dash-dotted lines),Fyy(dotted lines),andthevector sums (solidlines)areplotted asa function of 35b =rYe2Xb/gtanfl (intheoffshore region, Fyx=Fxysothat2Fxyisplotted, whereas inthesurfzone, Fyx andFyyarezero).Results areshown for(al)o=(a2)o=10ø. Forcing amplitudes (m/s 2)ona logscale are shownin the top panels;forcing phase(degrees)on a linear scaleis shownin the bottompanels. Resultsare computed for tanfl= 0.01, y = 0.42, 6- 0.1, and xb = 100 m. LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER 8671 10-3Offshore k Surf Zone 1i'1-5 • - • Mode ! Mode 2 Mode 3 10 -• ,,//'•"•-' -• f 10-7I I '• I I I I I I I I 10-3 10• ] 10-s 10• . , 10-7 I I I I I I I I I I I I I 10-• 104 10-• •. .... •. /' 10 -* x /';' /' .I ,.:. ..... [..' 0 1 2 3 4 5 6 7 8 9 10 3 4 5 6 7 8 9 10 Zb Figure4. Forcing components arisingfrom Fxx(dashed lines),Fxy(dash-dotted lines),F (dottedlines), andthevector sum, F•Pn (solid lines), foredge wave modes 1-3,with(Crl) o=(or2) o=•o, plotted asa function of ZI,= Cre2Xb/gtanfi ß Contributions fromtheoffshore andsurfzoneregions areshown in theleft and right panels, respectively. Only the amplitudesare plotted since the componentphasesare identical to mode 0 (Figure 3; the total phase does vary slightly as a function of mode). Results are computed for tanfi =0.01, ?'=0.42, 6=0.1, and xb =100 m. spectraldependenceof the forcing term will be determinedby the form of the edge wave waveform, having nodes at particular Zb. If the integrationof the couplingoccursover a narrow range of Zb, then the spectralshapeof the Onwill be stronglyreflected in the growth rate. The effect of (oci)o is examined in Figure 5 with function of Zb. As expected,the growth rates within the surf zone region are much larger (by a factor of 2-10) than in the offshoreregion. The growth rates are rapid, with a magnitude of about10-I across mostof therangeof Zb plotted.Because of the choice of normalization, the inverse of G gives the numberof edge wave periodsfor the edge wave amplitudeto (Crl)o=(cr2)o=1 ø, 10ø, and30ø. Resultsfor a mode1 edge grow to the size of the incident waves. Thus, assumingno wave are shownas a representative example. At small angles, dissipation,resonantlyexcited edge waves could grow to the Fxx dominatesthe forcing, being an order of magnitudegreater sameamplitudeas the incidentmodulationin as fast as 10 edge than the other terms. However, the forcing for low modes wave periods(and even fasterfor lower modesand smallZb). should beenhanced for steeper angles of incidence because Fxy The model predicts only initial, undamped growth rates andFyyhavea strong modalandangular dependence, although (essentially the rate at which energy is transferred from on shallow beaches, incident wave angles are not expected to incident to edge waves), and any reasonable damping be large owing to refraction effects. mechanismwill reducetheserates. The effect of dampingis discussed in the next section. Growth Rates Normalized growth rate magnitudes,G, and phases,0, for the offshore and surf zone regions (equations(30) and (38)) and the total (vector sum) are shownin Figure 6 for edge wave modes0-3 with (or1)o=(cr2)o =10ø. The amplitudesare plotted on a log scale and the phaseson a linear scale, as a The resultsfor the surf zoneforcingmechanismshowsharp valleys correspondingto nodes (zero crossings)in •Pn. The shapeof the edge wave profiles is similar to the growthrates for the surf zone mechanism. The corresponding valley in the offshoregrowthrate doesnot occurat the first edgewave node (Zb =1.5), nor is there any indication of strong nodal structure at higher Zb. This result is due to integrating the 8672 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER 10-3 Offshore Surf Zone 10-4 10-5 .J 10-6 10-7 I I I I I I I I I I 10-3 0-41 ./ 10-s c•-10 10 -6 10-7 I I I I I I 10-3 10-41 10-5 at- 30 10-6 10-7 I 012345678910 01234567891 )•b Figure 5. Effectof incidentwaveangle( (a I )o = (a2)o = 1ø, 10ø,and30ø) on the forcingcomponents for a mode 1 edgewave. Formatis the sameas in Figure4. E •0 ø ß • Mode 1 Mode 2 ] Mode 3 /' 10-2 1 I 'i,i / l'\/!' '•' !' 180 135 90- 450-45 - -90 - -135 - I I -180 I 012345678910 012345678910 I I I I I I I 012345678910 I I I I I I I I I 012345678910 Zo Figure 6. Normalizedinviscid growth rate (top) magnitudeand (bottom)phasefor edge wave modes0-3. Growthratesfor the offshoreregion(dashedlines;(30)), the surf zoneregion(dash-dottedlines;(38)), and the total(solidlines),areshown asa function of Zt,=Cre2Xb/gtan• ß Theinverse of thegrowth rateis the numberof edgewave periodsnecessaryfor the edgewavesto grow to the size of the incidentwaves. Results areshownfor (a 1)o=(a2)o = 10ø, tan•=0.01, ?'=0.42, 6=0.1, andxt,= 100m. I ! I LIPPMANN ET AL.: GENERATION ø OF EDGE WAVES IN SHALLOW WATER Offshore Surf Zone 10 -• Totals.., '"'"' 10-2 10-3 8673 _ I I I I I I I 180 - 135 90 45 0 -90 - -135 - -180 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Zb Figure 7. Effect of incidentwave angle on inviscid growth rates for a mode 1 edge wave. Resultsare shown for (left) the surfzone,(middle)offshore,and(right)totalfor (al) o =(a2) o =1 ø (solid lines), 10ø (dashed lines), 30 ø (dottedlines), and 45ø (dash-dottedlines). Format is the sameas in Figure 6. predicted growth rates are higher than can be expected in nature and places an upper bound on edge wave growth. We have considered the forcing due to an interacting incident wave field consisting of only two shallow water waves, yet in natural situations, incident wave fields are distinctly not bichromatic but rather consist of a spectrum of energy. If we have a continuous directional spectrum in n= 1 with (a I )o: (ø•2)o: 1ø,10ø,30ø,and45ø. The effectof o• is small for the directional range investigated, particularly nature, some (small) componentsof the forcing will satisfy for theoffshore regionwhere2FxyandFyyarerr outof phase. the resonancecondition for a number of different edge waves The angulardependenceis more pronouncedin the surf zone of variousfrequenciesand wavenumbers. In the offshore region where we have referencedamplitudes regionfor Zb >'" 3.0, increasingthe growthrate abouta factor of 2 from 1ø to 45 ø. Interestingly, in the offshore region, to the breakpoint, the spectral problem seems inherently increases in wave angle tend to reduce the growth rate (a linear where the superposition of many wave components consequenceof the rr phase relationships in the longshore leads to the possibility of resonant triad interactions forcing in the offshoreregion over a larger length of the edge wave profile. In the surf zone, the length over which the forcing occurs is very much smaller, determined by the incident modulation (•=0.1), and thus nodal points are strongly reflected in the growth rates. The effect of o• on growth rates is shown in Figure 7 for components of Fpq),whereas theopposite occursin thesurf occurringfor any numberof different (cyi-cyj, ki-kj) zone where the growth rate becomes larger as wave angle incident wave pairs. The forcing of a particular edge wave mode is just a linear sum of all possible interactions satisfyingthe edge wave dispersionrelation. Bowen and Guza [1978] discuss the implications of this resonant restriction increases (owingto increased contributions fromFxy). Discussion Growth Rates The predicted growth rates for the case of phase-locked forcing by deterministicwave trains are not unrealisticif we expect this mechanism to provide reasonable forcing of progressive edge waves under the stochastic forcing conditionsfound in nature [Holman, 1981]. The same type of result arose in the study of energy transfer into internal gravity waves from surfacewave packets. An initial study of Watson et al. [1976] showed a very strong forcing for the phase-lockedcase. Olbers and Hefterich [1979] redid the problem for a stochasticsurface wave field and found the strength of the forcing to be several orders of magnitude weaker thanpredictedby Watsonet al. This suggeststhat our and show that for narrow-beam incident swells, some frequency selection, with a strong modal dependence,may be expected. In the surf zone, the situation is more complicatedbecause amplitudesare functionsof local depth (and hencedistancefor monotonic beaches). However, for the bichromatic case, the forcing is dominated by F xx arising from the self-self interactionof the primary wave. If we consideronly this term, then the growth rate equationtakesthe form 1 c-)a ne-t•e. = -2rci xb (y,t) (ai)of ec-')t (ai)ogtanfi(2n+l) f•(xb)•n(x)dx 0 (40) 8674 LiPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER whereF(xt, ) is the meanforcingarisingfrom Fxx(•l•l). where G is found from (30) and (38) for the two forcing The operandof the couplingintegral is simply a functionof regions. Taking the magnitudeof the real part of (42) results the edge wave waveform with amplitude equal to F. The in a quadratic equationin 3an/3t which has solutionof the possibilityof edge wave forcing arisesif the temporal and fo rill spatial scalesof the oscillatingbreakerline match the edge •a n wave dispersionrelation. For the spectralcase, the breaker •=c, an[-l+f(an)] (43) line is describedby a summationof perturbations abouta mean breakpoint. Edge wave forcing results from a linear where combinationof various incident wave pairs contributingto the alongshoreperturbation,where the contributionfrom any 2•Cre 2 particularpair (with mnplitudesala2) canbe shown(Appendix c•- •[2+4ere 2 B) to be Axt, =Ho?,tanfl 4 ala 2cos(•l - •2) (41) where H o is the mean breaker height calculated over all incident wave spectralcomponents. Since zXx b is a functionof y and t, the Fourier transformof (41) resultsin a spectralform for the perturbation,now given in terms of longshorewavenumberand frequency. This leads to the attractive result that time and space dependent observations of the width of the surf zone can be related to the edge wave forcing by the lYequency-wavenumber spectrumof the wave breaking distribution. As indicatedby the growthrate equations(30) and (38), the behaviorof the forcing as a functionof the variousparameters turns out to be quite simple. Growth rate contributionsfrom the surf zone mechanis•nvary with Xl, in a way which largely follows the shapeof On. In the offshoreregion, the influence of mode numberon actualrates is largely throughthe variance distributionof Onin Xb space. That is, the rate of transferring energy from the incident field appearsto be about the samefor all modes,where, for highermodes,the energymust be spread over a larger cross-shoredistance. The only known measurements of growth rates for progressiveedge waves are from the laboratoryinvestigation of Bowen and Guza [1978]. Strict comparisonwith their results is difficult since their discussion was limited to Zb = 0.25, and furthermore,they have 6= 1.0, thus violating our assumption of small-amplitude modulation. Their observedgrowthrate is over an order of magnitudeslowerthan predicted by the model. The effect of viscous damping is likely to be important in the laboratory case, and since the scalesof the lab study are much different from thosetypically found in nature, no si•nple comparisonis readily made. In addition, any reasonabledamping mechanismis likely to be different in the two forcing regions and may well have characteristicswhich lead to preferential damping of high f (an)=•--• e + 1+;[2 Ir2an2 The caseof interestis the positiveroot, corresponding to energytransferfrom theincidentwavesto theedgewaves. We havealreadyassumed thatthe frictionaldissipation hasonly a negligible effect on the wave solutionsand dispersion relation; thus;t2<<4ere 2. Integrating (43)andconsidering only edge wave growth give a transcendentalform for the amplitude an=Coe-Clteq If(an)dr (44) where co is an integrationconstantand for growthto occur f(an) •nustbe real and positive. This equationcannotbe solvedanalytically;however,we can examinethe amplitude decay(dissipation) whentheforcingis turned off (IGI=0) and theexpected amplitude atequilibrium ( 3an/3t= 0 ). For the casewherethe forcingis turnedoff, (44) becomes an _ -e _l__At i4•2 o'e 2 e ISl-0 (45) where (an)ois the initial edge wave amplitude at t=0. Takingthe magnitudeof (45), the decaytimescalenormalized by the edgewaveperiodis found t__= -2'n[a•l(a•)o](46) It still remainsto parameterize the damping.We takea very simpleformfor •, followingLonguet-Higgins [1970] ;t=Caluø•l h0 (47) whereho is the depthat the breakpoint , Cotis a bottomdrag frequenciesand low modes [Bowen and Guza, 1978' Holman, 19811. coefficient of order10-2-10 -3,anduo is themagnitude of the Damped breakpointgiven by linearorbitalvelocityof theincidentwaveof heightHo at the Growth Rates Up to this pointwe haveconsidered only the undampededge wave growth rate. Here we briefly examine the effect of frictional damping. If we retain the frictional term in (2) and (3), the normalizedgrowthequation(11) takesthe form uø=2[ 2 o (48) Using H o = 7Xotanfi gives t -4'n[an/(a•)o](tan (2or e+i;t)--•-+/•CYea n(a1)o O'e2 TT = 77r3/2 (• Xt,'/2 (49) (a•)oXtanfl(2n+l) i2rc FCndx-G (42) A plot of tit e as a functionof Zb for various values of tanfi/Cct with an/(an)o =0.5 is shownin Figure8. Thehalf- LIPPMANNET AL.:GENERATION OFEDGEWAVESIN SHALLOWWATER Sensitivity to Model 8675 Parameters The dependence of growthrateson Zb, tan/t, and creis essentially combined into the nondimensionalscaling [•/Ca 10 parameter ,•b(equation (39)). Thusmeasured spectra obtained 30- in the field can be interpretedin termsof samplingposition andlowest-order profilecharacteristics. The effectof varying 20- : any particularcombination of parameters is easilydeduced. Additionally, because theforcingis dominated by Fxx,incident wave angleshave little influenceon the growthrate, except •0 '2 for low modesand large anglesof incidence. t/T• In themodel,y = 0.42 is assumed constant, consistent with 1 0 1'0 field data [Thorntonand Guza, 1982;Sallengerand Holman, 1985], and entersthe growth rate equations(30) and (38) linearly. Sinceall reportedvaluesof yare O(1). varyingyis notexpected to significantly influencethefinal results.Thus Figure8. Normalized edgewavedecaytimescales (equation the formulationof the growthrate (in both the offshoreand (49))asa function of Zt,= O'e2Xl•/gtan, B, forvarious values of surf zone regions)is dependenton only one free parameter: tan[¾C a . Thevertical axisisin number of edgewaveperiods the incident modulation, & Allowing the modulationto get for the edgewaveamplitude to decay50% (an/(an)o =0.5) aftertheforcingis turnedoff. Results areshownfor ?'= 0.42. muchlargerthanabout0.1 is notaccounted for exactlyby the model, where we have assumedsmall 8, so that incidentwave travel times areshort compared totimescales associated with life decayscaleis moderately insensitive to Zt,. The ratio the modulation[Symondset al., 1982]. Since terms in the radiation stresscontainingA (equation tan/•/C a is a roughindication of theQ (resonance) of the (28)) are small comparedto other terms, the forcing is of incidentfrequencies.Hencethe system, with highervaluescorresponding to weaklydamped effective!yindependent conditions with smalldragcoefficients.Steepbeaches tendto havesmallerdissipation ratesthanflat beaches, intuitivelyin incident frequencies serve only to provide a necessary (Gf, kf) interaction which matches theedge wavedispersion in incident wavevelocitypotentials, {I)i{I) j agreement withconceptual damping mechanisms [Bowenand relation.Products Guza, 1978; Komar, 1979; Holman, 1981]. For resonant (equation (23)),havean (XiXj)-1/2dependence, which products of incident frequencies o'io' j. However, we systems, half-lives canbeashighas10-30edgewaveperiods. contains At equilibrium,we can estimatethe size of edgewave amplitudes relativeto incidentwavesfrom(42) haveremovedthe dependence on incidentfrequencies in the formulationof S((I)i(I) j) by choosingincident wave amplitudes relativeto the breakpoint, whereproductsaiaj have an(XiXj)•/2dependence. Substituting (27)into(21) a•=r4[G•l fi);(t,l/2c-)an =0(50) 3/2ltan eliminates dependencies on (J'io'j. Still, the resonantresponseassumedin the modelrestricts This form can also be obtainedfrom the simple argumentthat incidentwavepairsto differencefrequencies andwavenumbers whichsatisfythe edgewavedispersion relation. Bowenand Guza [1978] discusstheseresonantrestrictionsin terms of incidentwave anglesandfrequencies.They showthat only a f(a,•) in (43) mustbe realfor growthto occur. A plotof a,•/(al)ois shown in Figure9 asa function of Zbforvarious valuesof tan•/Ca andwith Isl estimated conservatively fromFigures6 and7 to be about0.05. Guza and Thornton [ 1982]foundedgewavermsamplitudes in therun-upof a near planar beachas high as 75% of the incidentwave rms amplitude. HolmanandSallenger [1985]alsoobserved edge 3 13/ca 2.75 lO 2.5 9 2.25 wavesof thismagnitude on a barredbeachundera widerange 2 of conditions. The modelsuggestsamplituderatiosof this 1.75 magnitude for tan•/Ca values ranging from 3 to 10. Comparison with field resultsis not strictlyquantitative an/(aOo•.5 becausewavesin natureare nevertruly bichromatic,although it is notedthatKomar [1979] suggested that tan/•/Ca ratios of thismagnitude arereasonable for meancurrents onbeaches often found in nature. The dampingparameterizationused in (47) is only 8 7 6 1.25 4 1 •.•_•_-•-- 0.75 3 =_-------2 0.5 1 0.25 applicable to wavesin thesurfzone.It is notknownwhatthe damping mechanism is in theoffshore region.Herbetset al. Zb [1995a] show that the ratio of shorewardto seaward propagating freeinfragravity wavesdecreases asseaandswell Figure 9 Normalizededge wave equilibriumamplitude energyincreases, suggesting thatthedampingof edgewaves (equation i50))asafunction ofZb=C•e2Xb/gtanl • forvarious on the shelfdepends on the incidentwaveenergylevel, in valuesof tan/J/C a . Edge wave amplitudeshave been qualitative agreement withthesurfzonedamping mechanismnormalizedby the incidentwave amplitudeat the breakpoint. Results areshown for 7-0.42 andIGI-0.05. examined. 0 0 1 8676 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER finite number of incident frequencypairs satisfy the resonant condition for a narrow beam incident swell. This places a restrictionon the possible(o.1, o'2) combinationswhich could theoretically excite edge waves. Shallow Water Assumption In our formulation of the interaction radiation stress, the incident wave surface elevation is approximated by the shallow water Bessel function solutionfor progressivegravity waves over a sloping bottom. Inside the surf zone this formulation seems very reasonablesince the water depth is alwaysshallowwith respectto incidentwave wavelengthsand amplitudesare given as a functionof the local depth. Outside the variation in breaker position due to changing wave conditionsis ignored. In effect,Foda and Mei assumethat T is always larger when the incident waves are large. Field evidence [Guza and Thornton, 1982; Sallenger and Holman, 1985] suggeststhat T is reasonablyconstantfor any particular beach and the position of the breaker is the property that varies as a function of incident wave height. This is a central point in our calculation. Foda and Mei [1981] also compute growth rates for the interaction which vary over 2 orders of magnitude as a functionof the parameter the surf zone, however, the use of the shallow water solution (which gives resonanttriad interactions)is only valid for a limited distanceoffshore (determinedby the beachslope) and This parameterarisesfrom a particularscalingwhich is not the natural scaling for the shallow water equationson a sloping depends on the incidentwavefrequency.In the offshore beachwe have considered(equation(39)). As a consequence, region,(24) givesan x'1/4 dependence on waveamplitude. Foda and Mei have computedvalues of E2 which are very Thus productsof incidentwave amplitudesdecayin this region small, ranging from 1 to 3. If we consider a 60s beat, 10s asx'1/2,whereasthey shouldbecomeconstant in deepwater incident wave, and slope of 1'100, then E2=16. As growth where the wave profile we have assumedis only approximate. ratesincreasevery rapidly with E2 in Foda and Mei's results, As there are no resonanttriad interactionsin deep water (only we might expect very large growth rates for values typical of forcedwaves are generated),the magnitudeof the forcingdies open coast beaches. However, precise values cannot be away quickly as incidentwavesapproachdeepwater [Okihiro compareddue to their complex (fourth order) representationof et al., 1992]. the forcing inside the surf zone. The growthrate equationalso includesconsideration of OnSince Ondecaysexponentiallyoffshoreas a functionof mode number, the offshore integral for low modes is not biased Conelusions significantlyby reducedcontributionsin intermediatewater. A theoreticalmechanismfor driving resonantedge waves in Therefore, when evaluating the model, the upper limit of the nearshore is derived from the forced shallow water integrationcan be reducedfor low slopingbeacheswithout equations. Forcing integrals are separatedinto an offshore substantially underestimating the growth rate. This approximationis good for shallow beacheswhere edge wave region outside the breakpoint and within the region of length scalesare small; for steep beachesthe applicationis fluctuating surf zone width. Contributions to the surf zone region occur from primary self-self interactions for both questionable.The approximationis also more accuratefor the case of low modes which have a relatively rapid offshore breaking and nonbreaking waves, whereas in the offshore region, forcing is from crossinteractionsof a pair of waves. decay' for higher modes with slower decay scales, The strength of the forcing in each region is based on contributionsare more severely biased. Comparison With Foda and Mei [1981] Foda and Mei [1981] consider the problem of long waves generated by a normally incident swell which has a small alongshorevariation which is fixed in space but has a slow modulation in time. This variation in wave height could be thought of as two incident waves of the same frequency approachingthe beach from equal, but opposite,anglesto the normal. The modulation is then the beat frequencybetween these waves and the normally incident wave. This is therefore a rather particular case of the general problem which we are considering. Their discussionwas further limited in that the dominant (largest) wave is normally incident. However, Foda and Mei have carried through a very sophisticatedanalysis which includesresults for the case where the long waves grow to magnitudesof the same order as the incident modulation. They can thereforediscussthe processesthat eventuallylimit edge wave growth. However, the complexity of the calculationrather precludesany simple interpretationof these results. A further complication is that, for the case of breaking waves, Foda and Mei [1981] have useda representationof the breaker condition in which the breakpointis constantand the perturbationamplitude (modulation) extendsto the shoreline; amplitude modulations which arise from an interacting bichromaticwave field. Surf zone forcing is derived from momentumfluxes inducedby temporaland spatialvariations in initial breakpointamplitudes,expressedfor the planebeach caseas three-dimensional modulationsin surf zone width (first suggestedby Symondset al. [1982] while focusingon the two-dimensional problem). The nonlinearforcingis provided by theunbalanced gradient in radiation stress, Spq.Following Phillips[1977],a formforSpqdueto thenonlinear difference interactionof two incidentwavesapproaching the beachat an angle is derived. The model indicatesthat the forcingarisingfrom the crossshorecomponentof onshoredirectedmomentumflux provides the major contributionto the edgewave forcing,particularly for small anglesof incidence,higher edge wave modes,and lower frequencies.Increasingincidentwave angletendsto reduce the contributionin the offshoreregion for all modes becauseof the relative phase relationshipsof the forcing components,whereasthe growth rate inside the surf zone is enhancedtending to favor the breakpointmechanismin the overall growth rate. We find that the strengthof the surf zone generation mechanismis 2-10 times greaterthan in the offshoreregion for parameterrangesof particularinterest. Thus considering only the surf zone componentmay provide a reasonablefirst- LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER order estimate of the growth rate. The strength of the growth rate is inversely proportionalto edge wave mode number and directly proportional to the incident wave modulation, with a linear dependencein each region. Initial inviscid growth rates are found to be quite rapid, with edge waves amplitudes growing to incident wave mnpntut•esof the ---•-- of •n •.•_•. wave periods. Including a simple parameterizationfor bottom frictional damping and using inviscid growth rates to describe the energy transferfrom incident to edge waves allow predictions of edge wave equilibrium amplitudes consistent with field The first integral on the right-hand side of (A3) is zero since radiation stress gradients are constant inside the mean breakpoint. Thus, after evaluatingthe secondintegralin (A3) •b(1+•cos•e ) f Fsz,Ondx =Q(•b +•b cos •e)-Q(•b) where Q(x) is the unspecifiedresult of t•e integration. Expanding(A4) in a Taylor'sSeriesabout•t, p:•,tluces •b(I+'•COS •) •neasurements of Guza and Thornton [1982] and Holman and 3Q(.•b ) I Fsz On dx =•bcos I//e•)-•• 0 , for tanfi/Cd valuesof about3-10 (esti•nated from equilibrium for the surf zone, where the combination linear sum of all resonant interactions. Appendix A: Expansion of First-Order Coupling Integrals Edge wave growth rates are given as a function of the coupling integral between the wave forcing, F, and the edge wave waveform, On 2 3x2 +.. ' (AS) whichcanbe writtenin termsof the integrandof (first term on right-handside of (A2)) evaluatedat -•t, plus higher order harmonics 7b(1+,•cos• ) f Fsz CPndx =•b COS I//eFsz ('•b)On (•b) o of various incident wave pairs contributesto the alongshoreperturbation of the oscillatory breaker line. The contribution to spectral forcing by any particular pair is shown in principal to be due to the (A4) 0 Sallenger [1985], who found shorelinemnplitudesas high as 75% of the incident waves. Edge wave dissipationrates in the surf zone are found when the forcing is turned off and suggest half-life decay scalesof the order of 10-30 edge wave periods conditions). Dissipationrates outside the surf zone are not estimated since the appropriate damping mechanism in this region is not known. The extensionto spectral forcing is shown to be possible. The development in the offshore region can be based on a linear sum of all possible co•nbinations of wave triads satisfyingthe edge wave dispersionrelation. The same is true 8677 +O(• 2coS2•e)+... (A6) Thus for the first integral on right-hand side of (A2) we need onlyconsider theforcing at the•nean breakpoint, Fsz(jb), whichhastermsarisingfrom self andcrossinteractions of •1 and •2. The •• and •2•2 interactionsproducecoupling termswhich containprimary•nodulations (cos•e)of order8 and•3 respectively, whereas •2 interactions produce additional harmonic reruns (cos2•e)oforder•2. Considering only first-order, primary modulations (arising from •• interactions),the surf zone coupling integral is approximated by • (l+•co• % ) 0 •FszOnd••bCOS•eFsz(OlOl , •b)On(•b) (A7) where (I)1 and (I)2 are velocity potentialsof the incidentwaves which have mnplitudesa• and a2 -aa•, respectively. The coupling integral canbeseparated at thebreakpoint, xb(y, t), into surf zone (first term on right-hand side of (A2)) and offshore (second term on right-hand side of (A2)) integrals with the forcingdefinedin eachregionby F sz andFoff, respectively, •, () As in the surf zone region, the offshore coupling integral (secondterm on right-hand side of (A2)) can be separatedat Xb, • Xb • •FoffOndx • FoffOn • +• FoffOndx (h8) • (•+aoo•v,•) f FOndx I Fsz•)ndx q- I Foff CP ndX o o (A2) 7b(1+?; oos% ) The second integral on the right-hand side of (A8) has modulationswhich arisefrom only the •2 interactionsand is numericallyevaluatedin the main text (•• and•2(I) 2 interactions produce only nonvarying components through xb(y,t)-•b(l+r•coSgte)andthe functional dependencies where the breakpoint has been defined by have been droppedfor brevity. The surf zone integral (first term on right-handside) in (A2) can be further separatedby the mean breakpoint, Y'b, into a nonvarying component from the shoreline to -•b and a component arising from modulations in the breakpoint position this integral). The first integral describesthe contributionto edge wave growth by nonbreakingwaves in the fluctuating regionof the surf zone. Evaluatingthis integralproduces I FoffOndx --g(.• b)- g(.• bq-(•bcos life) (+aco f FszrPndxIFsjCPndx+ IFsjrPn dx 0 0 (A3) (A9) 8678 LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER where R(x) is the unspecifiedresult of the integration. and the secondterm representsthe perturbationof the wave Expandingin a Taylor's seriesabout Sb and keeping only first-order,primary modulationsgive height about its •neanvalue. The problemcan be simplifiedif the perturbationof the waveheightaboutthe meanvalueH o is not too large(it is, of course,necessarilylessthanone). So expanding(B5), f FofflPndx--• bcos lff eFoff (0101 , -•b )(Pn(.•b ) • (l+aoos•,,) (AlO) xb The total couplingintegral(A2) canthusbe approxi•nated ai i IF•ndX-•bcos lff e•n(•b )[Fsz(0101, •b)- Foff(0101 , '•b)] o +leor(OlO2, x)O(x)a _-Ho(x) 1+i j •iai2 (All) whichshowsthe first-ordercontributions to edgewave growth occur through self-self interactionsfrom both breakingand nonbreakingwaveswithin the fluctuatingregion of surf zone width, and through cross interactionsof a pair of waves seawardof the averagebreakpointposition. Since the signs (B6) Thereis a potentialforcingtermfor edgewavesfrom each of the secondtermsin the expansion cos(%•j)- cos[(//-//)x+(k i- kj)y-(cr i- cr i )t+cpi - cpj ] of FszandFof f areopposite in thefluctuating regionof surf zonewidth,the •1•1 forcingcontributions from breakingand nonbreakingwaves reinforce. (B7) with resonantforcingpossibleif ki - kj = ke andcri - crj= cre Appendix B: Breakpoint Forcing in a Wave Spectrum where (cre,ke) satisfy theedgewavedispersion relation. Within the spectra,the interactionbetween several different sets of incident waves can produce the same values of If the seasurfaceelevationrI is describedby 1l - • ai(x)cos I//i (BS) (B1) (or e, ke), andthenetforcing isthesumofallthese potential inputs. i Now the wavebreakswhen H = Th, so the meanbreaker where V/i=lix+kiy-Git+ cpi, then the significant wave height is 4 ti•nes the standarddeviationof rI, and the mean position hb is given by hb-Ho(xo)/7 wave height H o is given by (B9) and for a plane beach, the mean breakpoint is xo- Ho(xo)/(ytan/•).Theperturbation in breaker height then leads to a variable breakpointposition in both the longshoredirection and in time, where The variationin wave height,givenby the envelopeof the 1 xb(y, t)-•H(x, 7tanfi wave groups,can be defined and is often calculatedin terms of the low-passedvalues of the squareof the surfaceelevation, where "(X,t)--2'•'[( T•2 >] 1/2 (B3) where { } isthelow-passed value; then from (B1), . Ho •1+ T'tanfl t) Z Zaiajcøs(lffi-Iffj) i J Z ai (BlO) i which is a linear co•nbinationof a large number of contributions to the longshoreperturbationfrom pairs of [1+o42 )] incidentwaves.The contribution fromanyparticular pair is then Axe, =Ho7tanfi 4 ala2 cos(•l - g2) (Bll) the low-pass filter re•noves the high harmonics and sum interactions, so that ai2 i j cos( - Acknowledgments. This research was sponsored by the National Science Foundation undercontract OCE-8007972-01 andby theOffice Naval Researchunder grant N00014-90-Jll18. This work was (B5)ofcompleted while T.C.L was supported by a NationalResearchCouncil Postdoctoral Fellowship. We thankBob Guza and the reviewersfor LIPPMANN ET AL.: GENERATION OF EDGE WAVES IN SHALLOW WATER 8679 theirinsightfulcomments whichled to a correction to the surfzone Longuet-Higgins,M. S., and R. W. Stewart,Radiation stressesin water waves: A physicaldiscussion,with applications,Deep Sea Res.,11, forcingmechanism andin generalgreatlyimprovedthe paper. Joan 529-562, 1964. Semlerprepared thetypeset manuscript. Mei, C. C., and C. Benmoussa,Long waves induced by short-wave groupsoveran unevenbottom,J. Fluid Mech.,139, 219-235, 1984. Munk, W. H., Surf beats,Eos Trans. AGU, 30(6), 849-854, 1949. Okihiro, M., 1<. t. tauza, and R. J. Seymour,Boundinfragravity waves, References J. Gcophys.Res.,97(C7), 11,453-11,469,1992. Biesel,F., Equations generales au secondordrede la hotfieirreguliere, Olbers, D. J., and K. Herterich, The spectral energy transfer from surface waves to intemal waves, J. Fluid Mech., 92, 349-379, 1979. Houille Blanche, 7, 372-376, 1952. Bowen,A. J., and R. T. Guza, Edge wavesand surf beat,J. Geophys. Oltman-Shay, J. M., and R. T. Guza, Infragravity edge wave observationson two California beaches,J. Phys. Oceanogr.,17(5), Res., 83C4), 1913-1920, 1978. 644-663, 1987. Eckart,C., Surfacewaveson water of variabledepth,WaveRep. 100, Phillips, O. M., The Dynamicsof the Upper Ocean, 2nd ed., 336 pp., Scripps Inst.of Oceanogr., Univ.of Calif.,La Jolla,1951. Foda, M. A., and C. C. Mei, Nonlinearexcitationof long-trappedwaves CambridgeUniv. Press.,New York, 1977. Sallenger, A. H., and R. A. Holman, Wave energy saturation on a by a groupof shortswells,J. FluM Mech.,111,319-345,1981. Gallagher,B., Generationof surfbeat by nonlinearwave interactions, natural beachof variable slope,J. Geophys.Res., 90(C6), 11,93911,944, 1985. J. FluM Mech., 49, 1-20, 1971. Guza, R. T., and A. J. Bowen, The resonantinstabilitiesof long waves Sallenger,A. H., and R. A. Hobnan, Infragravity waves over a natural barredprofile,J. Geo?hys.Res.,92(C9), 9531-9540, 1987. obliquely incidenton a beach,J. Geophys. Res.,80(33),4529-4534, 1975. Schaffer,H. A., Infragravity waves inducedby short-wavegroups,J. Fluid Mech.,247, 551-588, 1993. Guza, R. T., and E. B. Thornton,Swashoscillationson a naturalbeach, J. Geophys.Res.,87(C1), 483-491, 1982. Schaffer, H. A., Edge waves forced by short-wavegroups,J. Fluid Mech.,259, 125-148, 1994. Guza, R. T., and E. B. Thomton,Observations of surf beat,J. Geophys. Res.,90(C2), 3161-3172, 1985. Schaffer,H. A., and I. Svendsen,Surf beat generationon a mild-slope Hassehnann,K., œ)nthe non-linear energy transfer in a gravity-wave beach, in Proceedings of the 21st International Conference on spectrum, 1, Generaltheory,J. FluidMech.,12,481-500,1962. CoastalEngineering,pp. 1058-1072,Am. Soc.Civ. Eng.,New York, 1988. Herbets,T. H. C., S. Elgar, andR. T. Guza, Generationandpropagation of infragravitywaves,J. Geophys.Res.,100(C12), 24,863-24,872, Stoker,J. J., Surfacewavesin water of variable depth,Q. App. Math., 1995a. 5(1), 1-54, 1947. Herbets, T. H. C., S. Elgar, R. T. Guza, and W. C. O'Reilly, Symonds,G., and A. J. Bowen, Interactionsof nearshorebars with Infragravity-frequency (0.005-0.05 Hz) motionson the shelf, II, Freewaves,J. Phys.Oceanogr,25(6), 1063-1079,1995b. Hohnan,R. A., Infragravityenergyin the surf zone,J. Geophys.Res., 80(C7), 6442-6450, 1981. Hohnan, R. A., and A. H. Sallenger,Setup and swashon a natural beach,J. Gcophys.Res.,90(C1), 945-953, 1985. Howd, P. A., J. M. Oltman-Shay,and R. A. Holman, Wave variance partitioningin the troughof a barredbeach,J. Geophys.Res., O0(C7), 12,781-12,795, 1991. incomingwavegroups,J. Geophys. Res.,89(C2), 1953-1959,1984. Symonds,G., D. A. Huntley, and A. J. Bowen, Two-dimensionalsurf beat: Long wave generation by a time-varying breakpoint, J. Geophys.Res., 87(C1), 492-498, 1982. Thornton, E. B., and R. T. Guza, Energy saturationand phase speeds measuredon a naturalbeach,J. Geophys.Res.,87(C12), 9499-9508, 1982. Tucker, M. J., Surf beats: Sea wavesof 1 to 5 minute period, Proc. R. Soc. London A, 202,565-573, 1950. Huntley, D. A., Long-periodwaveson a naturalbeach,J. Gcophys. Watson, K. M., B. J. West, and B. I. Cohen, Coupling of surfaceand Res., 81(36), 6441-6449, 1976. internalgravity waves' A mode couplingmodel,J. Fluid Mech., 77, 105-208, 1976. Huntley,D. A., R. T. Guza,and E. B. Thornton,Field observations of surf beat, 1, Progressiveedge waves, J. Geophys.Res., 80(C7),. 6451-6466, 1981. A. J. Bowen, Depamnentof Oceanography, DalhousieUniversity, Halifax, Nova Scotia, Canada B3H 461. Komar, P. D., Beach-slopedependenceof longshorecurrents,J. Waterw. Port Coastal Ocean Div. Am. Soc. Civ. Eng., 105(WW4), R. A. Hohnan,Collegeof Oceanand Am•osphericSciences,Oregon 460-464, 1979. StateUniversity,()C AdministrationBuilding 104, Corvallis,OR 97331- List, J. H., A model for the generationof two-dimensionalsurf beat,J. 55O3. Geophys. Res.,97(C7),5623-5635,1992. Longuet-Higgins,M. S., Longshorecurrentsgeneratedby obliquely T. C. Lippmm•n,Centerfor CoastalStudies,ScrippsInstitutionof Oceanography,University of California, San Diego, 9500 Gihnan [)rive, La Jolla,CA 92093-0209,(e-mail:tlippmann@ucsd.edu. incidentseawaves,2, J. Geophys.Res.,75(33), 6790-6801, 1970. Longuet-Higgins, M. S., andR. W. Stewart,Radiationstresses andmass transport in gravitywaves,with application to 'surfbeats,'J. Fluid Mech., 13, 481-504, 1962. (ReceivedJanuary5, 1994; revisedJune24, 1996; acceptedSeptember27, 1996.)