Waves

40 downloads 491 Views 1MB Size Report
At the top of the fluid layer, along the deformable surface (z = H+a), we impose that fluid ..... fluid mechanics, excep
Part II

PROCESSES

69

Chapter 4

Waves SUMMARY: The attention now turns toward specific types of motions that exist in natural fluid flows, beginning with waves. All waves need a restoring force, and to every restoring force corresponds a different type of wave: surface water waves under the action of gravity, internal waves under the action of buoyancy, topographic waves under the action of a vorticity gradient, etc. The study of wave dynamics also prepares for the study of instabilities, which in turn is a prelude to the study of turbulence.

4.1 4.1.1

Surface Gravity Waves Mechanism

Gravity waves on the surface of water are one of the most visible manifestations of fluid motions and one with which we all have a certain experience (Figure 4.1). The process at work is relatively easy to comprehend: A fluctuation causes water to rise above the equilibrium surface level, gravity pulls it back down because water is heavier than air, inertia acquired during the falling movement causes the water to penetrate below its level of equilibrium, and a bouncing motion results. The oscillation is similar to that of a spring that has been stretched and released. The ‘spring’ action in a surface water wave is gravity, hence the name of surface gravity wave. What is somewhat less intuitive is why gravity waves propagate horizontally. To understand this, one needs to consider the horizontal forces at play. When a parcel of water rises somewhere above the surface, the added weight of this water creates a pressure that is locally higher than normal, and this pressure anomaly accelerates (pushes, so to speak) the fluid away from that place and piles it up a little further, generating another surface rise some distance away. The net effect is a translation of the disturbance, hence a traveling wave. 71

72

CHAPTER 4. WAVES

Figure 4.1: Surface gravity waves on the sea approaching the coast. [Piha, New Zealand, photo Practical Ocean Energy Management Systems, Inc.]

Water motion under a surface wave is very nearly oscillatory, with almost no net displacement. Thus, surface waves, like most other fluid waves, are a mechanism by which the fluid moves energy from one area to another without involving any significant movement of the fluid itself. Energy and information are carried with the fluid acting as the support medium rather than as the messenger. That property has a fundamental implication: Surface waves by their very nature are unable to transport any mass, including dissolved pollutants and suspended matter. This fact is clearly manifested in the behavior of a floating object (such as an autumn leave on a pond) in the presence of surface waves: The waves pass by, but the object only bobs up and down. The energy carried by surface waves, however, must eventually be dissipated somewhere and will affect the water contents there. For example, wave energy can be converted into turbulent mixing under wave breaking, and the resulting mixing can stir the local water contents, such as pollutants, biological matter and heat. Wave energy can also be dissipated by bottom friction under wave-induced oscillatory flow, and this friction can in turn create a shear stress sufficiently strong to entrain sediments into suspension. In sum, waves do not contribute directly to transport and redistribution of fluid-borne elements along their travel but can be effective means by which a remote source of energy can affect the concentration of dissolved and suspended matter at a distant location. This remark holds true for most types of waves.

73

4.1. SURFACE GRAVITY WAVES

4.1.2

Linearization

Because of the oscillatory motions that they generate, surface gravity waves can be reasonably well described by a linear analysis. This is mathematically justified by restricting the attention to small wave amplitudes and weak accompanying motions. In the momentum equations (3.10)–(3.12), all terms linear in the velocity and pressure are then assumed to be small, while the nonlinear terms are assumed to be even smaller and therefore negligible. Among other terms, we neglect the first advective term u∂u/∂x next to the relative acceleration term ∂u/∂t, under the assumption that the velocity, length and time scales, U , L and T , meet the following criterion: U U2 0, Equation (4.44) admits periodic solutions cos(N t) and sin(N t) for h(t), indicating that the displaced parcel executes vertical oscillations at frequency N . Physically, when density decreases upward, the parcel once displaced upward is heavier than its surroundings, feels a downward recalling force, falls down, acquires a vertical velocity, overshoots its original position, becomes lighter than the ambient fluid, rises again and oscillates in this manner until friction eventually brings it to rest. On the contrary, should N 2 be negative, the particle does not return √ toward its original position but its displacement h(t) grows exponentially as exp( −N 2 t). Physically, the fluid is top heavy and gravitationally unstable. Typical values of the frequency N of stable stratifications correspond to periods 2π/N of a few seconds (atmosphere) or minutes (lakes). This is usually short compared to the period of internal waves, which implies ω < N .

88

CHAPTER 4. WAVES

The set of equations is now reduced to (4.35) and (4.41) for the two unknowns u and w. Since these are linear and the coefficients are time independent, we seek a trigonometric solution of the type u w

= =

U sin(kx x + kz z − ωt) W sin(kx x + kz z − ωt),

with angular frequency ω and wavenumbers kx and kz in the horizontal and vertical directions, respectively. The equations relating the amplitudes U and W are obtained by substitution in the preceding equations: kx U + kz W 2

2

2

kz ω U + kx (N − ω )W

=

0

(4.46)

=

0,

(4.47)

which form a two-by-two system of equations for the the velocity amplitudes U and W. The system of equations for U and W admits non-zero solutions only if ω2 = N 2

kx2 . kx2 + kz2

(4.48)

This is the dispersion relation for internal gravity waves. It admits two frequencies for every wavenumber pair (kx , kz ): s kx2 , (4.49) ω = ±N 2 kx + kz2 from which it is clear that the absolute value of the wave’s frequency ω must be smaller than the stratification frequency N of the fluid. Should a periodic perturbation be imposed on a stratified fluid at an angular frequency higher than N , the fluid would be unable to respond in the form of wave radiation, and the energy would have to be dissipated locally by means of turbulence.

4.2.2

Internal seiches in a rectangular basin

Long internal waves, with wavelengths comparable to the length of the lake, are capable of reflecting back and forth between the extremities of the basin without appreciable damping, and the result is a standing internal wave, called an internal seiche. On occasions, under favorable wind forcing, an internal seiche can assume a very large amplitude. Figure 4.10 shows an example for Cayuga Lake, USA. Such dramatic oscillation in the basin evokes the concept of resonance. For a rectangular basin, that is, for an unrealistic lake with a flat horizontal bottom and perfectly vertical sides, the mathematical problem is separable and can be solved by modal decomposition. The boundary conditions require impermeability all around the domain, i.e., is u = 0 at x = 0, L in the horizontal direction and w = 0

4.2. INTERNAL GRAVITY WAVES

89

Figure 4.10: Temperature versus depth and time in Cayuga Lake (New York State, USA) during 11–26 September 2001. The swing in temperature values about every two days at depths of 12–20 meters are indicative of large vertical oscillations of the thermal stratification, generated by a basin-wide resonant internal wave (internal seiche). [Figure courtesy of Prof. Edwin A. Cowen, Cornell University]

at z = 0, H in the vertical direction. Here, L is the length of the lake and H is its uniform depth. We thus seek a solution of the type: u = w

=

 nπz  cos cos(ωt) L  H   mπx  nπz W cos sin cos(ωt) L H U sin

 mπx 

(4.50) (4.51)

which meets all boundary conditions as long as m and n are integers. Substitution in Equations (4.35)–(4.41) demands: U W + n = 0 L H U W nω 2 + m(N 2 − ω 2 ) = 0. H L This set of equations implies: m

(4.52) (4.53)

m2 H 2 . (4.54) m2 H 2 + n2 L2 In other words, a wave solution exists only if the frequency ω takes one among a set of discrete values (because m = 1, 2, ... and n = 1, 2, ...). Physically, this means ω2 = N 2

90

CHAPTER 4. WAVES

Figure 4.11: Schematic of the horizontal and vertical motions in the internal seiche of lowest frequency in a uniformly stratified lake with flat bottom.

that for a given lake of length L, depth H and stratification frequency N , there corresponds a discrete set of oscillation frequencies. Alternatively, to an externally imposed frequency of oscillation (due to tides or wind), there exists a discrete set of stratification frequencies N that cause the basin to resonate. The gravest mode, that corresponding to m = n = 1 (Figure 4.11), has a frequency given by: NH . ω = √ H 2 + L2 Since lakes are typically much longer than they are deep (L ≫ H), this can be approximated to: r H NH dT¯ = . (4.55) ω ≃ αg L L dz Of particular interest in environmental problems is the value of the horizontal velocity along the bottom. Indeed, resonance can generate large oscillations accompanied by strong bottom currents capable of eroding sediments and depositing them elsewhere. According to (4.50), the maximum horizontal velocity at the bottom is umax = |U |, which is related to the maximum vertical velocity along the side wmax = |W | by umax = (nL/mH)wmax , according to (4.52). In turn, the vertical velocity is the time derivative of the vertical displacement. If the range of vertical excursion (= twice the amplitude) along the side is ∆z, then wmax = ω∆z/2 and umax = (nLω∆z/2mH). For the gravest mode, the maximum horizontal velocity along the bottom is thus N ∆z , (4.56) 2 and it occurs in the middle of the lake. It is left to the reader to show that the maximum bottom velocity of all other modes (m and n arbitrary) is the same but occurs at other locations along the bottom of the basin. umax =

91

4.3. MOUNTAIN WAVES

Figure 4.12: Mountain wave over the Presidential range in New Hampshire manifested by undulating clouds. [Photo credit: AccuWeather.com]

If the basin is not rectangular, the situation is complicated, and the analysis lies beyond the scope of this book. For details, see Maas and Lam (1995) and Cushman-Roisin (2004). Data-model comparisons were performed for internal seiches in Upper Mystic Lake in Massachusetts (Fricker and Nepf, 2000) and Lake Geneva in Switzerland (Lemmin et al., 2005), which both showed that the modes excited by wind events tend to be the lowest modes and that the structure of these seiche modes is highly sensitive to both bathymetry and stratification. There is a tendency toward a narrowing of wave energy into beams with a resulting amplification of velocities along the beams and where they reflect on the bottom and surface.

4.3

Mountain Waves

Mountain waves, also called lee waves, are atmospheric internal waves generated by wind passing over a mountain ridge (Figure 4.12). Such waves can occasionally be accompanied by sudden and localized strong winds that can topple trees and structures (Gill, 1982, Section 8.8). As long as the wave remains of moderate amplitude to permit a description by linear dynamics, a stratified flow over a mountain, hill or land topography of any shape can be represented by means of a Fourier decomposition into sinusoidal waves. When this is done, each wave component has a given horizontal wavenumber component kx . Because the topography is fixed in space, so, too, is the wave, which means that its phase propagation speed c = ω/kx must be equal and opposite to the mean wind velocity U , and this sets the wave frequency ω to ω = − U kx .

(4.57)

At the same time, the dispersion relation Equation (4.48) must hold, from which

92

CHAPTER 4. WAVES

we can determine the vertical wavenumber component that the wave is required to have: r r N2 N2 − 1 = ± − kx2 . (4.58) kz = ± kx ω2 U2 Clearly, two possibilities arise depending on the sign of the expression under the square root. Either the expression is positive in which case kz is real, or it is negative and causing kz to be imaginary. In the first case, which is that of the longer horizontal wavelengths, the wave is undulating in the vertical (sine wave in z), and wave energy generated by the topography is propagating upward. By contrast in the second case, corresponding to shorter horizontal wavelengths, the wave is evanescent in the vertical (exponential decay in z), and the corresponding energy is trapped near the ground. The Fourier decomposition of an actual topography will necessarily include many wavelengths. While the longer wavelengths may fall in the first regime, the shorter wavelengths will fall into the second regime, and the overall wave structure can be quite complex. Figure 4.13 illustrates the case of a wave generated by an idealized, bell-shaped mountain, h(x) =

hmax , 1 + (x/L)2

(4.59)

which involves waves of both types. In stratified waters, internal waves are frequently generated by a current (tidal or not) passing over bumpy topography. The difference with mountain waves in the atmosphere is the presence of an upper surface against which upward propagating waves can reflect as downward propagating waves. This situation is very common in fjords where stratification created by fresher river water overlying saltier seawater is made to oscillate vertically by tidal currents over the entrance sill (Baines, 1982; Cushman-Roisin and Svendsen, 1983). Along the bottom, the buoyancy force responsible for internal-wave generation is proportional to: 1 dh h(x) dx

(4.60)

and is thus greater where the bottom slope (dh/dx) occurs closer to the surface. The reflection of internal waves against the surface may cause significant horizontal currents, and the spatially periodic structure of these creates a pattern of alternating convergence and divergence, leading to the appearance of visible lines, called slicks, resulting from the gathering of small floating debris or bubbles along convergence lines (Klemas, 2012). Likewise, reflection of the same internal waves along the bottom may influence the erosion/sedimentation pattern. This is particularly pronounced where the oblique propagation direction of the wave happens to match the bottom slope, causing a critical reflection (Cacchione and Southard, 1974; Puig et al., 2004).

4.3. MOUNTAIN WAVES

93

Figure 4.13: Structure of a wave generated by a uniform flow (U = 10 m/s, flowing from left to right) of a uniformly stratified fluid (N = 0.01/s) over a bell-shaped mountain profile with length scale L = 1 km. The upper panel depicts the vertical displacements (upward in blue and downward in red), the zeros of which are outlined with dashed lines. In a moist atmosphere, the zones of updraft may be associated with cloud formation. The lower panel shows the ground-level distribution of pressure and wind. [From Queney, 1948, with colors added]

94

CHAPTER 4. WAVES

4.4

Inertia-Gravity Waves

Inertia-gravity waves are long surface waves modified by the effect of the earth’s rotation. The theory proceeds as in Section 4.1.3 except for the inclusion of the Coriolis terms in the momentum equations and the assumption of hydrostatic balance, both attributed to the long length scale: ∂u − fv ∂t ∂v + fu ∂t 0

= = =

1 ρ0 1 − ρ0 1 − ρ0



∂p ∂x ∂p ∂y ∂p − g ∂z

∂u ∂v ∂w + + = 0. ∂x ∂y ∂z

(4.61) (4.62) (4.63) (4.64)

The hydrostatic balance (4.63) provides the structure of the pressure field: p(x, y, z, t) = pa + ρ0 g(H + a − z),

(4.65)

where a(x, y, t) is the surface elevation, expressed as a wave form: a(x, y, t) = A sin(kx x + ky y − ωt).

(4.66)

Since for long waves, the vertical velocity varies linearly over depth [see (4.18) in the limit of small kz and kH], is zero at the bottom [w(z = 0) = 0] and fluctuates with the surface displacement on top [w(z = H) = ∂a/∂t], its expression is z z ∂a = − ωA cos(kx x + ky y − ωt). (4.67) H ∂t H With the elimination of pressure p and vertical velocity w in terms of the surface displacement a, Equations (4.61)–(4.62)–(4.64) become w(x, y, z, t) =

∂u − fv ∂t ∂v + fu ∂t ∂u ∂v + ∂x ∂y

= = +

∂a ∂x ∂a −g ∂y 1 ∂a = 0. H ∂t

−g

(4.68) (4.69) (4.70)

It can be shown with some algebra that the wave form (4.66) for a and similar trigonometric functions for the velocity components u and v lead to two solutions for the frequency ω. The first solution is simply ω = 0, which corresponds to a state of equilibrium, called the geostrophic balance. This state is of profound importance in meteorology and oceanography but is not pursued here, as our present concern is wave motion.

95

4.5. ENERGY PROPAGATION The second solution is ω 2 = f 2 + gHk 2 , 2

(4.71) kx2

ky2 ).

where k is the magnitude of the wavenumber (k = + These waves are isotropic because their frequency depends only on the magnitude of the wavenemuber and not its direction, and are said to be superinertial because their frequency exceeds the inertial frequency f . These waves are also dispersive as their propagation speed r f 2 λ2 ω = ± gH + , (4.72) c = k 4π 2 is wavelength dependent, except in the limit of √ short wavelengths, so short that the Coriolis effect becomes negligible (f λ ≪ 2π gH). In this limit, we recover the shallow-water gravity waves of Section 4.1.5. For additional information on inertiagravity waves, the reader is referred to Section 9.3 of Cushman-Roisin and Beckers (2011).

4.5

Energy Propagation

An interesting aspect of wave propagation is that the energy carried by waves does not always travel in the same direction or at the same speed as crests and troughs. In a wave of a single wavelength, the energy density is uniformly distributed because of the pattern repetition every wavelength. But, in a wave consisting of multiple components, different wavelengths dominate at different locations, and the energy distribution is non uniform. See for example the ripple pattern of Figure 4.4 in which the local wavelength increases from center to rim and energy is radiated outward. In a multi-component wave, called a wave group, the trigonometric function sin(kx − ωt) is no longer applicable and needs to be generalized to sin α, where the phase α(x, t) is a more complicated function of space and time. Nonetheless, one can define a local wavenumber k and a local frequency ω as ∂α ∂α , ω = − , (4.73) ∂x ∂t to which correspond a local wavelength λ = 2π/k and a local period T = 2π/ω. The uniqueness of the function α in Equations (4.73) implies a relation between wavenumber and frequency: k =

∂ω ∂k + = 0. (4.74) ∂t ∂x Since the dispersion relation [such as (4.16) or (4.48) or (4.71)] prescribes a relation between frequency and wavenumber, it follows that ω = ω(k) and ∂ω/∂x = (dω/dk)(∂k/∂x), leading to:

96

CHAPTER 4. WAVES

∂k ∂k + cg = 0, ∂t ∂x

(4.75)

where the coefficient cg stands for dω/dk and is a function of k only. In terms of the local wavelength, we have ∂λ ∂λ + cg = 0, ∂t ∂x

(4.76)

with cg now taken as a function of λ. This equation is a transport equation stating that the quantity λ travels unchanged at speed cg . Each wavelength travels at its own speed cg (λ). Ahead and behind this wave are waves of the group with other wavelengths. Thus, the energy associated with wavelength λ travels at the speed cg =

dω dk

(4.77)

and not at the propagation speed c = ω/k of local crests and troughs. For this reason, a distinction is to be made between the two propagation speed: c is called the phase speed, and cg the group velocity. That the difference between the two speeds can be quite important is exemplified by deep-water waves. For these waves indeed, the frequency ω, given by Equation (4.23), is clearly a function of wavenumber k, with

c

=

cg

=

r g ω = k k r g 1 c dω = = , dk 2 k 2

(4.78) (4.79)

indicating that the energy travels at half the speed of crests and troughs. But, how is that possible? Close examination of the evolution of a ripple pattern (of Figure 4.4, for example) reveals that crests vanish by destructive interference and new ones form by constructive interference. An observer tracking by eye the progress of one crest will suddenly lose sight of it and be led to believe that the eyes were somehow confused between the crest that disappears and the following one, only to discover that this one, too, disappears! And, while crests disappear at the head of the formation, new ones keep appearing behind. In other words, individual crests overtake the group, and the group proceeds at a slower pace than individual crests. In the case of deep-water gravity waves, the group travels half as fast as crests within in. In two or three dimensions, there is a wavenumber in each spatial direction, and one forms the wavenumber vector ~k with components (kx , ky , kz ). The group velocity, too, becomes a vector ~cg formed with the derivatives of ω with respect to the wavenumber components (∂ω/∂kx , ∂ω/∂ky , ∂ω/∂kz ), that is, ~ k ω. ~cg = ∇

(4.80)

97

4.6. NONLINEAR EFFECTS

Figure 4.14: Structure of a ray of internal gravity waves. While the ray radiates energy toward the top right, crest and trough lines are parallel to the ray and travel down to the right. Thus, in an internal wave, phase and energy travel at a right angle of each other.

Internal gravity waves are peculiar in the sense that their energy propagates vertically in the opposite direction of their crests and troughs. Indeed, according to (4.49), the vertical phase speed is N kx ω = ± kz kkz

(4.81)

∂ω N kx kz , = ∓ ∂kz k3

(4.82)

cz = while their vertical group velocity is cgz =

where k 2 = kx2 + kz2 . In contrast, the horizontal wave speed and group velocity share the same sign, and it can be shown (Problem 4.7) that the group velocity vector is aligned with the crest and trough lines, implying that phase and energy propagate at a right angle of each other (Figure 4.14) For more on group velocity, the reader is referred to Lighthill (1978, Section 3.6), Kundu (1990, Chapter 7, Section 9) or Cushman-Roisin and Beckers (2011, Appendix B).

4.6

Nonlinear Effects

The preceding wave theories were all predicated on the assumption of a weak amplitude in order to justify linear dynamics and thus permit a relatively easy mathematical treatment. But as anyone who has ever contemplated surf on a beach, it is clear that waves can acquire finite amplitudes, depart in shape from simple sinusoidal curves, sometimes to the point of tipping over and rolling into surf (Figure 4.15). Needless to say, nonlinear wave theory has received considerable attention, at the cost of major mathematical difficulties. For surface gravity waves, a long-wave, finite-amplitude theory can be developed in the case when the wave amplitude a is on the order of H 3 /λ2 in which H is the resting water depth and λ the wavelength. This leads to the following for a(x, t) first obtained by J. Boussinesq in 1872 (see Long, 1964):

98

CHAPTER 4. WAVES

Figure 4.15: Rolling wave upon approaching a beach.

# "  2 ∂2a ∂a ∂2a 3a ∂ 2 a H 2 ∂ 4 a 3 . − gH = gH + + ∂t2 ∂x2 H ∂x H ∂x2 3 ∂x4

(4.83)

Among other solutions to this equation, one is the so-called solitary wave which has a single crest and travels without changing shape over time, at√ a speed dependent on its amplitude and a little greater than the linear-wave speed gH. This particular solution also possesses the property of being “a highly favoured type of disturbance” (Long, 1964) in the sense that it is often the shape ultimately assumed by a weakly nonlinear wave after a certain period of adjustment. This situation arises after reaching a balance between two opposite tendencies. Because the highest point of a disturbance moves faster than others, there is a tendency for wave steepening. This is represented by the two nonlinear terms on the right of Equation (4.83). At the same time, because longer waves propagate faster than shorter ones, according to the dispersion effect represented by the last, linear term of (4.83), the waveform spreads and tends to flatten. The combination of wave steepening and wave flattening gives rise to an equilibrated wave shape, the solitary wave. The preceding equation and its solutions, however, fail to apply to breaking waves so often seen on beaches, such as the one caught by photograph for Figure 4.15. The modeling of breaking (or rolling) waves can only be done with a nonhydrostatic numerical model, and the interested reader is referred to the books by Mader (2004) and Lin (2008). Finite-amplitude internal waves, too, have been studied (Turner, 1973, Section 3.1). As their surface cousins, internal waves possess solitary forms in which a balance is struck between nonlinear steepening and linear dispersion.

99

4.6. NONLINEAR EFFECTS

Problems 4-1. After a few glasses of whiskey, a retired captain recounts a very stormy night in the South Pacific when the waves were so fierce that his 45-m long ship with its bow and stern spanning the distance from trough to crest was rocked back and forth in less than three seconds. Can this possibly be true? 4-2. In which two limits do surface gravity waves have their wavelength proportional to a power of their period (i.e. λT n )? What are then the values of the exponent n? 4-3. Swell is a surface wave generated in the open ocean by a wind storm. Consider a swell of 15-m wavelength in the middle of the Atlantic Ocean heading for Miami Beach. As the swell encounters shallower waters along its path, its nature changes from deep-water wave to intermediate-water wave and ultimately to shallow-water wave. At which respective depths do these changes occur? If the change from deep-water to intermediate-water wave occurs after 1200 km of travel, how old is the swell by then? Finally, at which depth will the swell propagate twice as slowly as it did in the deep ocean? [Hint: While the swell undergoes transformation, its frequency ω remains unchanged.] 4-4. By taking the limit kH → 0 in Equations (4.17) and (4.18), show that the horizontal velocity u associated with a shallow-water wave is nearly depth independent. What is the vertical structure of the accompanying vertical velocity w? 4-5. At very short wavelengths (millimeters to centimeters), the restoring force of water waves is surface tension, and the waves are called capillary waves (Figure 4.16). At slightly longer wavelengths, surface water waves are intermediate between pure capillary and pure gravity waves, and the theory leads to the following dispersion relation: ω =

p

k (g + γk 2 ) tanh(kH) ,

(4.84)

in which γ is a parameter related to the surface tension, a joint property of the two fluids in contact at the surface. Because the gravitational acceleration g = 9.81 m/s2 and the surface tension parameter γ = 7.38 × 10−5 m3 /s2 for water-air contact, the γk 2 correction becomes important only at relatively short wavelengths. (a) For which wavelength λ does the surface-tension correction term become equal to the gravitational term? (b) If the water depth H = 3 m, should the waves at the preceding wavelength be considered as deep-water waves, intermediate waves or shallow-water waves?

100

CHAPTER 4. WAVES

Figure 4.16: Capillary waves riding on gravity waves.

(c) Simplify the dispersion relation accordingly. (d) Derive the expressions for the phase speed c and the group velocity cg from this simplified dispersion relation, and then form the ratio cg /c of the two. Plot this ratio as a function of wavelength and explore the two limits of only surface tension (pure capillary waves) and only gravity (pure gravity waves). (e) What is a main difference in energy propagation between capillary and gravity waves? (f) After throwing a stone in a pond and while watching the waves radiate outward, where should you see the capillary waves, near the center or at the outer edge? 4-6. What are the periods of the first three seiche modes of Lake Erie, which is 388-km long and 18.9-m deep in average? 4-7. What is the magnitude of the oscillatory bottom current if a wind storm over Lake Erie (see previous problem) generates the second seiche mode with a trough-to-crest height of 60 cm? 4-8. In a 28-m deep lake during summer, the temperature varies gradually from a high of 25◦ C at the surface to a low of 17◦ C at the bottom. This stratification supports internal waves, which are manifested on the surface by a pattern of distorsions (slicks) of small wind-induced waves. Observations reveal a repeating slick pattern of 163 meters along the main axis of the lake and traveling horizontally with a speed of 22.6 cm/s. What is the corresponding vertical wavelength of the internal waves? Compare this to the water depth. Is there any relation that strikes you that could have a significance? 4-9. Norwegian fjords are former glacier valleys now flooded by seawater. There are therefore long and narrow. The freshwater discharged by lateral rivers patially mixes into the seawater, creating a vertical stratification with decreasing

4.6. NONLINEAR EFFECTS

101

salinity upward. At the open end, sea tides force motions at fixed frequencies, which inside the fjord generate internal waves at tidal frequencies, called internal tides. Consider Skjomen Fjord near Narvik in northern Norway, with length L of 25 km, average depth H of about 110 m, stratification frequency N of about 2.0 × 10−3 s−1 , and in which the semi-diurnal tide generates motions with a 12.42-hour period. What are the horizontal wavelengths of the first two vertical modes (those with half vertical wavelengths respectively equal to the depth and half the depth)? Can these waves fit in the length of the fjord? 4-10. Derive dispersion relation (4.71) for inertia-gravity waves. Are the velocity components u and v in phase, out of phase, or in quadrature (phase shift of 0◦ , 180◦ or ±90◦ )? 4-11. Show that the energy of internal gravity waves travels in a direction parallel to the lines of crests and troughs.