Linear guided modes and Whitham-Boussinesq model for variable topogra

TL;DR

This paper compares numerical solutions using a simplified non-local Dirichlet-Neumann operator to known analytic solutions for linear water wave problems in channels with variable topography, demonstrating good accuracy especially for even modes.

Contribution

It introduces a simple approximation of the Dirichlet-Neumann operator for linear waves and validates it against classical analytic solutions for complex geometries and depth profiles.

Findings

The operator yields close results for even modes in bounded channels.

Discrepancies occur for odd modes near boundaries.

The model accurately predicts lowest trapped modes in variable depth domains.

Abstract

In this article we study two classical linear water wave problems, i) normal modes of infinite straight channels of bounded constant cross-section, and ii) trapped longitudinal modes in domains with unbounded constant cross-section. Both problems can be stated using linearized free surface potential flow theory, and our goal is to compare known analytic solutions in the literature to numerical solutions obtained using an ad-hoc but simple approximation of the non-local Dirichlet-Neumann operator for linear waves proposed in [vargas2016whitham]. To study normal modes in channels with bounded cross-section we consider special symmetric triangular cross-sections, namely symmetric triangles with sides inclined at $45^{\circ}$ and $60^{\circ}$ to the vertical, and compare modes obtained using the non-local Dirichlet-Neumann operator to known semi-exact analytic expressions by Lamb [lamb1932hydrodynamics], Macdonald [macdonald1893waves] , Greenhill [greenhill1887wave], Packham [packham1980small], and Groves [groves1994hamiltonian]. These geometries have slopping beach boundaries that should in principle limit the applicability of our approximate Dirichlet-Neumann operator. We nevertheless see that the operator gives remarkably close results for even modes, while for odd modes we have some discrepancies near the boundary. For trapped longitudinal modes in domains with an infinite cross-section we consider a piecewise constant depth profile and compare modes computed with the nonlocal operator modes to known analytic solutions of linearized shallow water theory by Miles [miles1972wave], Lin, Juang and Tsay [lin2001anomalous], see also [mei2005theory]. This is a problem of significant geophysical interest, and the proposed model is shows to give quantitatively similar results for the lowest trapped modes.