betaFeedback

WorksPhysics and Astronomy

Novel Approximation of the Modified Mild Slope Equation

Abstract

The mild-slope equation and its various modifications aim to model, with varying degrees of success, linear water wave propagation over sloping or undulating seabed topography. However, despite multiple modifications and attempted simplifications, the different variants of the equation include multiple higher order terms involving the nonlinear water wave dispersion relation and thus remain analytic intractable. To further facilitate its use, we derive a drastically simplified alternative version of the modified mild-slope equation that bears striking resemblance to the linear shallow water equation while retaining all critical features of the original equation that enable it to be valid for a wide range of wave numbers and water depths. Direct comparison of the modified mild-slope equation and our simplified formulation indicates that the simplified equations agree with the modified mild slope equation at leading orders in the forcing frequency and local depth. Validations with multiple sets of benchmark wave scattering problems demonstrate that despite the clearly reduced complexity, the simplified equations were able to replicate the predictions of the modified mild slope equations over a wide range of wave numbers and surface topographies, including higher order resonant conditions.

Introduction

The study of surface wave scattering by seabed topography is vitally important for predicting coastal processes to help marine structures design, wave power extraction and habitat preservation, while also providing fundamental insights into wave energy dissipation and scattering mechanisms[1, 2, 3, 4]. The dynamics of water waves can be adequately described by Laplace’s equation in combination with other equations specifying boundary and/or radiation conditions[5]. However, unless the seafloor is flat, it is extremely challenging to find exact analytical solutions for this type of problems, and they typically only apply to simple geometries with horizontal or vertical boundaries. To overcome this limitation, approximations such as the linear shallow-water equations(SWE) or mild-slope equations(MSE) have been developed. These approaches simplify the problem by approximating the vertical dependence of the flow through depth-averaging of the original linear wave equations, thus making it feasible to study wave scattering over slowly varying topography[6, 7]. However, the SWE is only an accurate model when restricted to very long waves with respect to depth, such as tides or tsunamis, due to its neglect of depth-dependent wave dispersion relations. Simulations of realistic coastal wave dynamics instead rely on Boussinesq-type models which systematically incorporate higher order dispersive effects[8, 9, 10, 11, 12, 13], but at the cost of significantly greater mathematical and numerical complexity compared to the simplified SWE framework. On the other hand, the MSE, as originally discovered independently by Berkhoff[14, 15] and Smith and Sprinks[16] among others, aimed to overcome the limitations of the SWE and has undergone multiple modifications and improvements since its inception[17, 18, 19, 20, 21, 22]. Of particular note are the so-called complementary mild-slope equations(CMSE) derived by Kim and Bai[23], the extended mild-slope equations(EMSE) by Kirby[17], and modified slope equation(MMSE) introduced by Chamberlain and Porter[19] which improve the original formulation of MSE[14] for short wavelengths and in the presence of moderate to strong surface gradients such as sinusoidal ripples[24] or the resonant Bragg reflection by artificial bars[25]. The MMSE has found widespread success in application to surface wave scattering over sloping and undulating topographies[19, 26, 27, 28, 29, 30]. Due to its analytic complexity, the MMSE itself has also undergone multiple modifications and simplifications[20, 31] to facilitate application. In particular, a simplified version of the MMSE has been proposed by Porter[32] that amounts to an extension of the linearized shallow water equation to apply to variable depth topography and short waves. However, none of the above variants of the MMSE were able to attain both analytic simplicity with accuracy over a wide range of conditions.

For example, the extended linear shallow water equations by [32] were shown to be less accurate at large surface slope values, whereas even the simplified versions of MMSE[20, 21] still require the solution of the nonlinear wave dispersion relation as well as the higher order derivatives of the surface topography at each point.

This work introduces a remarkable simplification of the MMSE[19, 20] which further reduces its computational and numerical complexity by replacing the nonlinear implicit local wave dispersion relation with an explicit approximate expression that does not contain higher derivative terms of the bedform h(x,y)h(x,y). We demonstrate how the new equations can be elegantly reformulated to retain the familiar structure of linear SWE. This transformation offers two key advantages: Firstly, it retains the accuracy of the original MMSE over a wide range of wave-lengths, and secondly, it preserves the simplicity of standard SWE - facilitating theoretical analysis and numerical implementation based on existing SWE frameworks with negligible additional cost.

The Simplified Modified Mild Slope Equation

The Modified Mild Slope Equation

Our setup assumes incompressible, homogeneous and irrotational fluid motion over a bed of varying quiescent depth h(x,y)h(x,y), where xx and yy denote the horizontal Cartesian coordinates. zz is the vertical coordinate with the undisturbed free surface at z=0z=0 and is positive for upward locations. If the system is subject to forcing given by the velocity potential Φ\Phi with angular frequency ω\omega as below:

Φ(x,y,z,t)=Re⁡(ϕ(x,y,z)e−iωt),\Phi(x,y,z,t)=\operatorname{Re}\left(\phi(x,y,z)e^{-i\omega t}\right),

then according to linearized wave theory, ϕ\phi must satisfy2

∇2ϕ=0(−h<z<0),(1)\nabla^2\phi=0 \qquad(-h<z<0), \tag*{(1)}
ϕz−ω2gϕ=0(z=0),(2)\phi_z-\frac{\omega^2}{g}\phi=0 \qquad(z=0), \tag*{(2)}
ϕz+∇hh⋅∇hϕ=0(z=−h),(3)\phi_z+\nabla_h h\cdot\nabla_h\phi=0 \qquad(z=-h), \tag*{(3)}

where ∇=(∂/∂x,∂/∂y,∂/∂z)\nabla=(\partial/\partial x,\partial/\partial y,\partial/\partial z) and ∇h=(∂/∂x,∂/∂y)\nabla_h=(\partial/\partial x,\partial/\partial y) denote the total and horizontal gradient operators, respectively. The free surface elevation is given by ξ(x,y,t)=Re⁡(η(x,y)e−iωt)\xi(x,y,t)=\operatorname{Re}\left(\eta(x,y)e^{-i\omega t}\right) where

η(x,y)=iωg(ϕ)z=0.(4)\eta(x,y)=\frac{i\omega}{g}(\phi)_{z=0}. \tag*{(4)}

ϕ\phi is subject to either lateral boundary conditions or a radiation condition in the case of infinite extent, which will be of no further consideration for later part of this work.

The modified mild slope equations (MMSE) can be derived as a simplification and approximation for Eqn. (1)–(4) by eliminating the dependence on the vertical coordinate zz through depth averaging and the application of suitable variational principles[18, 19, 20]. They are formulated in terms of the free surface elevation as follows[19]:

∇h⋅u0∇hη+(k2u0+r)η=0(5)\nabla_h\cdot u_0\nabla_h\eta+(k^2u_0+r)\eta=0 \tag*{(5)}

where the local wave number kk satisfies the dispersion relation

ω2g=ktanh⁡(kh)(6)\frac{\omega^2}{g}=k\tanh(kh) \tag*{(6)}

dictated by the height h=h(x,y)h=h(x,y) at each point. The term r(h)r(h) has the explicit expression

r(h)=u1(h)∇h2h+u2(h)(∇hh)2,(7)r(h)=u_1(h)\nabla_h^2h+u_2(h)(\nabla_hh)^2, \tag*{(7)}

with

u0(h)=12ktanh⁡(kh)(1+2khsinh⁡(2kh)),(8)u_0(h)=\frac{1}{2k}\tanh(kh)\left(1+\frac{2kh}{\sinh(2kh)}\right), \tag*{(8)}
u1(h)=sech⁡2(kh)4(K+sinh⁡(K)){sinh⁡(K)−Kcosh⁡(K)},(9)u_1(h)=\frac{\operatorname{sech}^2(kh)}{4(K+\sinh(K))}\{\sinh(K)-K\cosh(K)\}, \tag*{(9)}
u2(h)=ksech⁡2(kh)12(K+sinh⁡(K))3{K4+4K3sinh⁡(K)−9sinh⁡(K)sinh⁡(2K)+3K(K+2sinh⁡(K))(cosh⁡2(K)−2cosh⁡(K)+3)},(10)\begin{aligned} u_2(h)&=\frac{k\operatorname{sech}^2(kh)}{12(K+\sinh(K))^3}\{K^4+4K^3\sinh(K)-9\sinh(K)\sinh(2K)\\ &\quad+3K(K+2\sinh(K))(\cosh^2(K)-2\cosh(K)+3)\}, \tag*{(10)} \end{aligned}

where we have used the abbreviation K=2khK=2kh. It has been noted that this formulation of MMSE can be transformed into the original mild slope equations by setting the term rr to zero at the expense of modeling accuracy for undulating topographies[19].

Porter [20] has demonstrated that the above formulation (5) for MMSE can be transformed into an equivalent equation of simpler structure as below:

∇⋅k−2∇ζ+{1−v(∇h)2}ζ=0,(11)\nabla\cdot k^{-2}\nabla\zeta+ \left\{1-v(\nabla h)^2\right\}\zeta= 0, \tag*{(11)}

where vv is given by

v(h)=3(2K+sinh⁡K)(sinh⁡(2K)−sinh⁡K)−3K2(cosh⁡(2K)+2)−4K3sinh⁡K−K43(K+sinh⁡K)4,(12)v(h)=\frac{3(2K+\sinh K)(\sinh(2K)-\sinh K)-3K^2(\cosh(2K)+2)-4K^3\sinh K-K^4}{3(K+\sinh K)^4}, \tag*{(12)}

and the free-surface elevation η\eta is related to ζ\zeta of Eqn. (11) as

η(x,y)=2cosh⁡(kh)⋅ζ(x,y)k(2kh+sinh⁡(2kh)).(13)\eta(x,y)=\frac{2\cosh(kh)\cdot\zeta(x,y)}{\sqrt{k(2kh+\sinh(2kh))}}. \tag*{(13)}

where it can be readily verified that in the shallow water limit h≪1h \ll1, we have η≈ζ\eta\approx\zeta.

Porter[20, 21] has shown that the term v(h)v(h) is bounded as 0<v(h)<0.030 < v(h) < 0.03 and is far smaller than its theoretical maximum value in most applications. Hence, the equivalent formulation of the MMSE given in Eqn. (11) can be significantly simplified by dropping the v(h)v(h) term, which then leads to:

∇⋅k−2∇ζ+ζ=0,(14)\nabla\cdot k^{-2}\nabla\zeta+ \zeta= 0, \tag*{(14)}

This equation can be readily seen as a generalized version of the linear shallow water equations which are valid under the condition kh≫1kh \gg1 and for which the dispersion relation is ω2/g=k2h\omega^2/g = k^2h. It incorporates the depth-dependent wave dispersion and is shown to agree with the original MMSE (5) to the leading order, making a valid approximation beyond the long wave limit for many practical problems[20, 21]. Hence, we will use the term MMSE to refer interchangeably to either Eqn. (5) or (14) in later parts of this work. However, Eqn. (14) still requires the evaluation of the wave number kk as a function of the local topography height h(x,y)h(x,y) at every point (x,y)(x,y), which involves solving the nonlinear dispersion relation (6) for water waves. In the following, we show how this equation can be further simplified by replacing k(h)k(h) with an explicit expression that can be more readily computed and analyzed.

Simplified formulation

Starting from the modified mild slope equation (14) introduced by Porter[20], our goal is to replace the the local wave number kk in Eqn (14) by a simpler explicit expression. We do this by following two approaches: First, we obtain the Taylor expansion for k(h)k(h) in terms of hh, and then we derive a second expression which agrees with the Taylor series to leading order but leading to a simpler expression when substituted into (14). We further show that both of these resulting formulations can be regarded as generalizations of the linear shallow water equation, similar to those introduced in Porter[32] but with simpler analytic structure.

Starting from the nonlinear dispersion relation k=k(h,ω2/g)k=k(h,\omega^2/g) for water waves given by Eqn. (6), we apply Langrange’s theorem on Taylor expansion for inverse functions[33] to obtain the following relation between the wave number kk, water height hh and ν=ω2/g\nu=\omega^2/g:

k(h,ν)≈k1(h,ν)=ν1/2h−1+13ν+⋯(15)k(h,\nu) \approx k_1(h,\nu)=\nu^{1/2}\sqrt{h^{-1}+\frac{1}{3}\nu+\cdots} \tag*{(15)}

It can be shown through simple calculations that the following alternative expression agrees with (15) to the leading order:

k(h,ν)≈k2(h,ν)=ν1/2(h−13νh2+⋯ )−1/2(16)k(h,\nu) \approx k_2(h,\nu)=\nu^{1/2}\left(h-\frac{1}{3}\nu h^2+\cdots\right)^{-1/2} \tag*{(16)}

Keeping the leading two orders of (15) or (16) and substituting them into Eqn. (14) to replace kk hence results in two possible simplified versions of the modified mild slope equation as given below:

∇⋅(ν−1h1+13hν∇ζ)+ζ=0,\nabla\cdot\left(\frac{\nu^{-1}h}{1+\frac{1}{3}h\nu}\nabla\zeta\right)+\zeta=0,

It can be readily seen that these two equations are generalizations of the linear shallow water wave equation, which is governed by the simple dispersion relation ν=k2h\nu=k^2h at the long wave or shallow depth limit and can hence be written as

∇(ν−1h⋅∇ζ)+ζ=0,\nabla\left(\nu^{-1}h\cdot\nabla\zeta\right)+\zeta=0,

We can see that the second version given by (SMSE2) contains a simpler mathematical expression than the first version (SMSE1) derived directly from the Taylor expansion of the dispersion relation (6). However, as can be seen from Eqn. (16), its range of validity is more restrictive since it requires that the forcing frequencies satisfy ν<3/h\nu<3/h.

In the following, we study the accuracy of each of these approximations compared to the exact dispersion relation (6), which gives an indication over which combination of wavelength and water depth the simplified equations will be accurate approximations to the slope equations given by Eqn. (14). In Figure 1, we compare both approximations (15) and (16) with the exact wave number k(h,ν)k(h,\nu) determined by the dispersion relation ν=ktanh⁡(kh)\nu=k\tanh(kh) for the two temporal forcing frequencies given by ν=ω2/g=1,3\nu=\omega^2/g=1,3. It can be seen by comparing Figure 1a and Figure 1b that at the lower frequency ν=1\nu=1 both approximations are accurate over a larger range of water depth values up until h=1.75h=1.75, whereas for the higher frequency ν=3\nu=3, the range of validity for both schemes shrinks to water depths of 0<h<0.350<h<0.35. In Figure 2, we display the maximal water depth hmax⁡h_{\max} for which both approximations will be accurate to within 10% of the exact value k(h,ν)k(h,\nu) as a function of the forcing frequency ν\nu. This confirms and quantifies the earlier observation that the range of validity for both simplified formulations decreases with growing forcing frequency ν\nu. This is not surprising, since, as a variant of the modified mild slope equation and a generalization of the shallow water equation, our formulation will be most accurate in the long-wave low-depth regime, similar as the other formulations of mild slope equations[19, 32]. We will show that this limitation does not prevent our proposed simplified equations from accurately modeling practical wave scattering problems that are most commonly studied in literature[24, 25, 34].

Image not extracted

Figure 1. Comparison of the two different simplified approximations k1,k2k_1,k_2 given by (15) and (16) for the nonlinear wave dispersion relation k=k(h,ν)k=k(h,\nu) satisfying ν=ktanh⁡(kh)\nu=k\tanh(kh) for different temporal forcing frequencies ν=ω2/g\nu=\omega^2/g: (a) ν=1\nu=1, and (b) ν=3\nu=3.

Image not extracted

Figure 2. Maximal height hmax⁡h_{\max} as a function of ν\nu such that for all h<hmax⁡h < h_{\max}, the simplified expressions (15) and (16) produce approximate wave numbers k1,k2k_1,k_2 that deviates less than 10% from the exact value k(h,ν)k(h,\nu) given by ν=ktanh⁡(kh)\nu= k\tanh(kh).

Determination of Wave Reflection and Transmission

In the following, we only consider two-dimensional water waves without variation along the horizontal yy direction. We assume that the water depth h(x)h(x) is constant except within the limited region 0<x<L0 < x < L:

h(x)={h0∀x≤0,h1∀x≥L,h(x) = \begin{cases} h_0 & \forall x \le0, \\ h_1 & \forall x \ge L, \end{cases}

where h0h_0 and h1h_1 are constant heights of the surface bed on each side of the varying topography, and LL is the total length of the varying topography. Given the water depth function h(x)h(x), we can solve either one of the mild slope equations (5), (14), (SMSE1), (SMSE2) for the free surface elevation η(x)\eta(x) over x∈[0,L]x \in[0,L]. We then have to match the solution at the boundary points x=0,Lx = 0,L subject to continuity with the incoming and outgoing wave forms which are of the form:

η(x)={eik0x+Re−ik0x(x<0),Teik1x(x>L).\eta(x) = \begin{cases} e^{ik_0x}+Re^{-ik_0x} & (x < 0), \\ Te^{ik_1x} & (x > L). \end{cases}

where R,TR,T are the reflection and transmission coefficients, respectively, and k0k_0 is the incident wave number directly related to the forcing frequency through ν=ω2/g=k0tanh⁡(k0⋅h0)\nu= \omega^2/g = k_0\tanh(k_0 \cdot h_0). A more detailed description of the implementation of this procedure to determine the reflection and transmission coefficients can be found in the work by Chamberlain and Porter [26].

Validation

Roseau's Problem

As one of the rare examples in linear water wave theory that possesses an exact analytic solution, we present Roseau’s scattering problem[35] over a non-constant topography which is parameterized as z(ξ)=−h(x(ξ))z(\xi)=-h(x(\xi)) with ξ∈(−∞,∞)\xi\in(-\infty,\infty):

x(ξ)/h0=ξ−(2πβ)−1(1−hL/h0)ln⁡(1+e2βπξ+2eβπξcos⁡(βπ))(17)x(\xi)/h_0=\xi-(2\pi\beta)^{-1}(1-h_L/h_0)\ln(1+e^{2\beta\pi\xi}+2e^{\beta\pi\xi}\cos(\beta\pi)) \tag*{(17)}
z(ξ)/h0=−1+(πβ)−1(1−hL/h0)tan⁡−1{sin⁡(βπ)/(e−βπξ+cos⁡(βπ))}(18)z(\xi)/h_0=-1+(\pi\beta)^{-1}(1-h_L/h_0)\tan^{-1}\{\sin(\beta\pi)/(e^{-\beta\pi\xi}+\cos(\beta\pi))\} \tag*{(18)}

Thus h′(x(ξ))=−z′(ξ)/x′(ξ)h'(x(\xi))=-z'(\xi)/x'(\xi), and β∈(0,1)\beta\in(0,1) is a parameter that determines the degree of shoaling. The ratio hL/h0h_L/h_0 fixes the steepness of the topography. The wave reflection coefficient magnitude is then given by[32, 35]:

∣R∣=∣sinh⁡[(k0h0−kLhL)/β]sinh⁡[(k0h0+kLhL)/β]∣(19)|R|=\left|\frac{\sinh[(k_0h_0-k_Lh_L)/\beta]}{\sinh[(k_0h_0+k_Lh_L)/\beta]}\right| \tag*{(19)}

This describes a topography that stays at nearly uniform depth for all ∣x∣≫1|x|\gg1 and undergoes nonlinear steepening in the middle section as determined by the parameters hL/h0,βh_L/h_0,\beta. A limited portion of the profile z(x)z(x) for a Roseau bed topography with β=0.5,hL/h0=0.25\beta=0.5,h_L/h_0=0.25 is shown in Figure 3 with a horizontal shift such that the left-most boundary of the bed is located at x=0x=0.

Image not extracted

Figure 3. Limited portion of surface bedform for Roseau’s problem parametrized by Eqns. (17) and (18). The geometric parameters are β=0.5\beta= 0.5, hL/h0=0.25h_L/h_0 = 0.25. The xx coordinate has been shifted by 6 units to the right.

To assess the fidelity of our simplified formulations for the modified mild slope equations, we computed the reflection coefficients RR for different incident wave numbers kk based on the procedure in Sec. II C using (SMSE1) and (SMSE2) and compared the results against both the exact value given by Eqn. (19) as well as the numerical results obtained from the modified mild slope equation(MMSE). The formulation given by (SMSE2) is only valid for forcing frequencies ν<3/h\nu< 3/h and the corresponding incident wavenumbers kk, so numerical results for (SMSE2) had to capped to kk values satisfying this constraint. Our main purpose here is to determine how accurately the simplified equations (SMSE1) and (SMSE2) approximate the MMSE given by Eqn. (14), which was shown to be nearly equivalent to the original MMSE in Eqn. (5)[20]. As can be seen in Figure 4, there is virtually no discernible difference between all three formulations except for the wave number range 1<kh0<31 < kh_0 < 3. It is especially encouraging to see that at the high frequency or steep slope limit, our simplified equation (SMSE1) was able to predict the convergence of the reflection coefficient to zero, as dictated by the exact result (19). This gives us confidence that our simplified formulations (SMSE1) and (SMSE2) are indeed an accurate representation of the MMSE over a wide range of wave numbers as encountered in practical problems.

Image not extracted

Figure 4. Reflection coefficient for the scattering problem over Roseau’s profile as a function of the incident wave number kk normalized by the maximal height h0h_0. Comparison of numerical results obtained from the simplified modified mild slope equations with Eqn. (14) and the exact analytical value.

Booij's Problem

A popular test problem that is widely used to assess the range of validity of different versions of mild-slope equations as a function of slope magnitude is introduced in the work by Booij[34]. In this configuration, plane waves are scattered incident normally on a depth profile which is reduced to one third of its original height through a planar slope of horizontal length LL, thus h1=h0/3h_1 = h_0/3 and h(x)=h0(1−2x/3L)h(x)=h_0(1-2x/3L) for x∈[0,L]x\in[0,L]. Booij calculated the reflection coefficients with the help of the original mild-slope equation developed by Berkhoff[14] as well as from the linear wave equations (1)-(4). In the following, we compare Booij’s data with numerical results of the reflection coefficient obtained through the following variants of mild slope equations, which are: 1. Berkhoff’s Mild Slope Equation (MSE)[14], 2. Porter’s Modified Mild Slope Equation (MMSE) given by (14)[20], 3. Porter’s extended shallow water equations (ESWE)[32], 4. Simplified Mild Slope Equation based on approximation (15) for local wave number (SMSE1) and 5. Simplified Mild Slope Equation based on (16) for k(h,ν)k(h,\nu) (SMSE2). Figure 5 displays the magnitude of the reflection coefficient ∣R∣|R| as a function of kLkL, where k=k(h0,ν)k=k(h_0,\nu) is the incident wave number. Booij’s results from solving the full linearized problem are also shown as "ground truth" to validate the other numerical results. In the steep slope range with kL<1kL<1, it can be seen that both simplified versions (SMSE1), (SMSE2) produce reflection coefficients that are nearly identical to those from the MMSE and are closer to data points from the linearized equations, whereas both the original MSE and Porter’s ESWE clearly show stronger deviations. This demonstrates that SMSE1 and SMSE2 are capable

Image not extracted

Figure 5. Reflection coefficient for the scattering problem over planar slope (Booij) as a function of the incident wave number kk normalized by the length of steepening terrain LL. Higher values of LL correspond to smaller slopes. Comparison of numerical results obtained from the simplified modified mild slope equations with other variants and the values from full linearized wave equations without depth-averaging.

of handling steep slope cases just as MMSE and are a better generalization of the shallow water equations than MSE and ESWE while attaining a simpler mathematical structure.

Scattering over Sinusoidal Ripple Beds

The ripple bed topography is another popular test case that has been extensively used to study the accuracy of different types of mild slope equations. It was well-known that both the linear shallow-water equation and wave scattering theory by Miles[36] as well as the original mild slope equation were unable to predict the correct reflection coefficients at the Bragg resonant conditions[25] for this type of topography, which was the primary motivation behind the derivation of more enhanced versions of mild slope equations[19]. In the following, we will evaluate the accuracy of our simplified formulations (SMSE1), (SMSE2) for the modified mild-slope equation(MMSE) compared to the original MMSE version[19, 20] at predicting scattering over such ripple bed topography. Experiments on water wave scattering have been carried out on a smoothly undulating bedform by Davies and Heathershaw[24] where the surface variation around a constant mean depth h0h_0 is given as h(x)=h0+δ(x)h(x)=h_0+\delta(x) with

δ(x)=Dsin⁡(ℓx)(0≤x≤L),(20)\delta(x)=D\sin(\ell x) \qquad(0\le x\le L), \tag*{(20)}

where h0h_0 is a constant and L=2nπ/ℓL=2n\pi/\ell. This means that the ripple topography is made of nn sinusoidal undulations of a single spatial frequency about the mean depth h0h_0. A sketch of this topography is displayed in Figure 6. Measurements for the reflection coefficient RR obtained in this experiment will be used to validate the accuracy of the simplified SMSE1, SMSE2 compared to the original MMSE.

Image not extracted

Figure 6. Sinusoidal ripple bed for wave scattering experiment by Davies and Heathershaw[24]. Sketch reproduced from Liu, Liu, and Zhao[12].

In Figure 7, the magnitude of reflection coefficient ∣R∣|R| are plotted against 2k/ℓ2k/\ell, which represents twice the ratio of the incident wavenumber kk and the wave number ℓ\ell of the bed topography as given in Eqn. (20). For this particular case, the amplitude of the ripples relative to the mean depth is D/h0=0.32D/h_0=0.32, and the number of ripples in the bedform is n=4n=4. We compare the predictions obtained from solving (SMSE1) and (SMSE2) with those of MMSE and the experimental data. It can be seen that there is nearly perfect agreement between our simplified formulations and the original MMSE. In particular, besides correctly predicting the magnitude of the primary Bragg reflection at 2k/ℓ=12k/\ell=1, our numerical results from (SMSE1) and (SMSE2) were also able to capture the existence of higher order resonances at 2k/ℓ≈22k/\ell\approx2 which was known to be a deficiency of the original mild slope equation[19]. This shows that our formulations for the modified mild slope equations are an accurate approximation for the MMSE as given in (5) or (14), but require far less computational effort due to their greater simplicity similar to the linear shallow water equations.

Image not extracted

Figure 7. Reflection coefficient magnitude ∣R∣|R| of scattering over sinusoidal ripple bed as a function of the incoming wave number kk normalized by the wave number ℓ\ell of the surface undulation. Experimental data by Davies and Heathershaw[24] for the case D/h0=0.32,n=4D/h_0=0.32,n=4 are compared to the MMSE and our simplified formulations (SMSE1), (SMSE2).

Scattering over Artificial Bars

A variation of the scattering experiment by Davies and Heathershaw[24] has been carried out by Kirby and Anton[25] who placed periodically spaced artificial bars of height (δ=δ(x)>0\delta=\delta(x)>0) on a flat surface of depth h0h_0. In their study, the surface undulation δ\delta due to the artificial bars is given by a rectified cosine function as follows:

δ(x)={Dcos⁡(xbL(x−Nℓ))for Nℓ−bL2≤x≤Nℓ+bL20otherwise\delta(x)= \begin{cases} D\cos\left(\frac{x}{b_L}(x-N\ell)\right) & \text{for } N\ell-\frac{b_L}{2}\le x\le N\ell+\frac{b_L}{2}\\ 0 & \text{otherwise} \end{cases}

where N=0,…,Nb−1N=0,\ldots,N_b-1. Due to the rectification of the bedform’s cosine undulation, higher order harmonics beyond the original spatial period ℓ\ell of the artificial bars are being introduced, in contrast to the ripple bed topography of the experiment by Davies and Heathershaw[24] where the surface undulation only contains a single wave number. In the experimental setup as sketched in Figure 8, the mean water depth was h0=15cmh_0=15\mathrm{cm}. Height of the artificial bars was D=5cmD=5\mathrm{cm}, which leads to a ratio D/h=0.33D/h=0.33. This indicates a relatively large height of the bars and could be the cause of some deviations between predictions of mild slope equations and experimental data.

Image not extracted

Figure 8. Artificial bars in Kirby and Anton’s scattering experiment. Sketch reproduced from Kirby and Anton [25]

The bar length is bL=50cmb_L = 50\mathrm{cm}. Two different spatial periods in the surface undulation were tested, namely ℓ=80cm,120cm\ell= 80\mathrm{cm},120\mathrm{cm}, corresponding to cases where higher harmonics in the bedform play a role of low importance and higher importance, respectively.

Results for the case ℓ=80cm\ell= 80\mathrm{cm} are displayed in Figure 9. The reflection coefficient magnitude ∣R∣|R| is plotted against 2k/λ2k/\lambda, where λ=2π/ℓ\lambda= 2\pi/\ell is the wave number of the bedform. Experimental data from Kirby and Anton [25] are compared against predictions by the extended mild slope equation(EMSE)[17], the MMSE, and our simplified formulations SMSE1 as well as SMSE2. It can be seen that all variants of the mild slope equations were able to capture the first dominant Bragg reflection at 2kλ=12k\lambda= 1, whereas the MMSE and our simplified formulations were better at predicting the magnitude of the higher order resonance at 2kλ=22k\lambda= 2 compared to EMSE. Again, we notice that despite the obvious simpler formulation of SMSE1 and SMSE2, they all showed excellent agreement with results obtained from MMSE.

Image not extracted

Figure 9. Reflection coefficient magnitude ∣R∣|R| of scattering over rectified cosine bedform as a function of the incoming wave number kk normalized by the wave number λ\lambda of the surface undulation. Experimental data by Kirby and Anton [25] for the case of 4 bars with spatial period ℓ=80cm\ell= 80\mathrm{cm} compared to the MMSE and our simplified formulations (SMSE1), (SMSE2).

Figure 10 displays reflection coefficient results for the second case where the spatial bar period is increased to ℓ=120cm\ell= 120\mathrm{cm}. The same set of equations as before were used to compare against experimental data. Similar like the previous case, it is evident that all variants of the mild slope equations were again able to accurately predict the primary dominant Bragg reflection at 2kλ=12k\lambda= 1. However, the MMSE and our simplified formulations were again more accurate at capturing the magnitude of the higher order peak at 2kλ=22k\lambda= 2 compared to EMSE which, despite its greater complexity, underestimated the reflection coefficient magnitude. For the range of incident wave numbers studied here, the simpler formulations SMSE1 and SMSE2 all showed nearly perfect agreement with results from MMSE, with differences only starting to become noticeable at wave numbers k>2.35k > 2.35.

Image not extracted

Figure 10. Reflection coefficient magnitude ∣R∣|R| of scattering over rectified cosine bedform as a function of the incoming wave number kk normalized by the wave number λ\lambda of the surface undulation. Experimental data by Kirby and Anton [25] for the case of 4 bars with spatial period ℓ=120cm\ell= 120\mathrm{cm} compared to the MMSE and our simplified formulations (SMSE1), (SMSE2).

Conclusions

This study has successfully developed a simplified yet remarkably accurate approximation to the modified mild slope equations (MMSE) for modeling water wave scattering phenomena across a wide range of conditions. The newly derived mild slope equations (SMSE1,2_{1,2}) demonstrate that mathematical simplicity does not need to come at the expense of physical accuracy as they achieve nearly identical results to the original MMSE for the most important scattering problems, while its reduced complexity offers significant practical advantages.

The main benefit of (SMSE1,2_{1,2}) lies in the fact that they do not require the exact evaluation of the nonlinear wave dispersion relation and many other related correction terms at each point, which is necessary in the oroginal MMSE. However, despite the clear resemblance of (SMSE1,2_{1,2}) to the linear shallow water equations (SWE), it is capable of maintaining accuracy from shallow to intermediate water depths and across a broad spectrum of wavelengths. Hence, our newly developed equations combine the best of two worlds: They share the same mathematical simplicity of the linear SWE while offering a wider range of validity beyond the existing generalizations of SWE, which typically fail outside narrow parameter ranges. Through validation of multiple benchmark test cases for wave scattering problems, we found that the simplified equations were able match the predictions by MMSE while eliminating unnecessary mathematical complexity.

The practical benefits of this simplification are manifold. First, the streamlined mathematical structure facilitates deeper theoretical analysis, making it possible to extract fundamental insights that were previously obscured by computational complexity. Second, the equations’ elegant formulation enables the derivation of exact analytical solutions for problems that previously required numerical approximation - opening new avenues for understanding wave-topography interactions. From an implementation perspective, the model’s computational efficiency and compatibility with existing numerical schemes make it particularly valuable for both research and engineering applications..

Looking forward, this work establishes a foundation for several promising research directions. One important extension would be to generalize our simplified mild slope equations to multiple dimensions, which appears to be a straightforward task. These novel equations can also provide an ideal starting point for developing more efficient numerical methods for coastal wave prediction.

In conclusion, this study demonstrates that through the adoption of simple approximations to the MMSE, we can achieve both the accuracy and wide applicability of the MMSE as well as the simplicity of the linear SWE. The resulting framework not only advances our theoretical understanding of wave scattering phenomena, but also provides engineers and researchers with a powerful new tool that combines physical fidelity with practical usability. Future work will explore applications of this approach to more complex wave environments and three-dimensional scattering problems.

Declaration of Interests: The authors report no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  1. [1]1R. G. Dean and R. A. Dalrymple, Water wave mechanics for engineers and scientists, Vol. 2 (world scientific publishing company, 1991).DOI
  2. [2]2M. W. Dingemans, Water wave propagation over uneven bottoms (in 2 parts), Vol. 13 (World Scientific, 1997).
  3. [3]3Y. Li and C. C. Mei, “Bragg scattering by a line array of small cylinders in a waveguide. part 1. linear aspects,” Journal of Fluid Mechanics 583, 161–187 (2007).DOI
  4. [4]4X. Garnaud and C. C. Mei, “Bragg scattering and wave-power extraction by an array of small buoys,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 466, 79–106 (2010).DOI
  5. [5]5C. C. Mei, M. A. Stiassnie, and D. K.-P. Yue, Theory and applications of ocean surface waves: Part 1: linear aspects (World Scientific, 2005).
  6. [6]6K. O. Friedrichs, “Waves on a shallow sloping beach,” Communications on Pure and Applied Mathematics 1, 109–134 (1948).DOI
  7. [7]7J. J. Stoker, Water waves: The mathematical theory with applications (Courier Dover Publications, 2019).
  8. [8]8D. H. Peregrine, “Long waves on a beach,” Journal of fluid mechanics 27, 815–827 (1967).DOI
  9. [9]M. Brocchini, “A reasoned overview on boussinesq-type models: the interplay between physics, mathematics and numerics,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 469, 20130496 (2013).pubmed.ncbi.nlm.nih.gov/24353475
  10. [10]Q. Chen, “Fully nonlinear boussinesq-type equations for waves and currents over porous beds,” Journal of Engineering Mechanics 132, 220–230 (2006).DOI
  11. [11]F. Shi, J. T. Kirby, J. C. Harris, J. D. Geiman, and S. T. Grilli, “A high-order adaptive time-stepping tvd solver for boussinesq modeling of breaking waves and coastal inundation,” Ocean Modelling 43, 36–51 (2012).
  12. [12]W. Liu, Y. Liu, and X. Zhao, “Numerical study of bragg reflection of regular water waves over fringing reefs based on a boussinesq model,” Ocean Engineering 190, 106415 (2019).DOI
  13. [13]Y. Xiong, S. Tang, and L. Zhu, “Improved boussinesq-type equations for flexural-gravity waves,” Physics of Fluids 37 (2025).DOI
  14. [14]J. C. Berkhoff, “Computation of combined refraction—diffraction,” in Coastal Engineering 1972 (1973) pp. 471–490.
  15. [15]J. C. W. Berkhoff, “Mathematical models for simple harmonic linear water waves: wave diffraction and refraction,” Ph. D. Thesis (1976).
  16. [16]R. Smith and T. Sprinks, “Scattering of surface waves by a conical island,” Journal of Fluid Mechanics 72, 373–384 (1975).DOI
  17. [17]J. T. Kirby, “A general wave equation for waves over rippled beds,” Journal of Fluid Mechanics 162, 171–186 (1986).DOI
  18. [18]J. Miles, “Variational approximations for gravity waves in water of variable depth,” Journal of Fluid Mechanics 232, 681–688 (1991).DOI
  19. [19]P. Chamberlain and D. Porter, “The modified mild-slope equation,” Journal of Fluid Mechanics 291, 393–407 (1995).DOI
  20. [20]D. Porter, “The mild-slope equations,” Journal of Fluid Mechanics 494, 51–63 (2003).
  21. [21]D. Porter, “The mild-slope equations: a unified theory,” Journal of Fluid Mechanics 887, A29 (2020).DOI
  22. [22]T.-W. Hsu, L.-H. Tsai, and Y.-T. Huang, “Bragg scattering of water waves by multiply composite artificial bars,” Coastal Engineering Journal 45, 235–253 (2003).DOI
  23. [23]J. W. Kim and K. J. Bai, “A new complementary mild-slope equation,” Journal of Fluid Mechanics 511, 25–40 (2004).DOI
  24. [24]24A. Davies and A. Heathershaw, “Surface-wave propagation over sinusoidally varying topography,” Journal of Fluid Mechanics 144, 419–443 (1984).
  25. [25]25J. T. Kirby and J. P. Anton, “Bragg reflection of waves by artificial bars,” in Coastal Engineering 1990 (American Society of Civil Engineers, 1990) pp. 757–768.DOI
  26. [26]26P. Chamberlain and D. Porter, “Decomposition methods for wave scattering by topography with application to ripple beds,” Wave Motion 22, 201–214 (1995).DOI
  27. [27]27H.-W. Liu, J. Yang, and P. Lin, “An analytic solution to the modified mild-slope equation for wave propagation over one-dimensional piecewise smooth topographies,” Wave Motion 49, 445–460 (2012).DOI
  28. [28]28H.-W. Liu, D.-J. Fu, and X.-L. Sun, “Analytic solution to the modified mild-slope equation for reflection by a rectangular breakwater with scour trenches,” Journal of Engineering Mechanics 139, 39–58 (2013).DOI
  29. [29]29H.-W. Liu, H.-D. Zeng, and H.-D. Huang, “Bragg resonant reflection of surface waves from deep water to shallow water by a finite array of trapezoidal bars,” Applied Ocean Research 94, 101976 (2020).DOI
  30. [30]30H.-W. Liu, W.-J. Xie, J.-J. Xie, and D. Ning, “Zero reflections of water surface gravity waves induced by a finite periodic array of artificial bars,” Physics of Fluids 37 (2025).DOI
  31. [31]31H.-W. Liu and X.-M. Zhou, “Explicit modified mild-slope equation for wave scattering by piecewise monotonic and piecewise smooth bathymetries,” Journal of Engineering Mathematics 87, 29–45 (2014).DOI
  32. [32]32R. Porter, “An extended linear shallow-water equation,” Journal of Fluid Mechanics 876, 413–427 (2019).DOI
  33. [33]33J. L. Lagrange, Nouvelle méthode pour résoudre les équations littérales par le moyen des séries (Chez Haude et Spener, Libraires de la Cour & de l’Académie royale, 1770).
  34. [34]34N. Booij, “A note on the accuracy of the mild-slope equation,” Coastal Engineering 7, 191–203 (1983).
  35. [35]35M. Roseau, Asymptotic wave theory, Vol. 20 (Elsevier, 2012).
  36. [36]36J. W. Miles, “Oblique surface-wave diffraction by a cylindrical obstacle,” Dynamics of Atmospheres and Oceans 6, 121–123 (1981).DOI

Paper details

vlnc:W_cx3c2br2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents