Normal mode decomposition and dispersive and nonlinear mixing in stratified fluids (Q2063895)

The authors start from the incompressible Euler equations \(\rho (\partial_{t}U+U\cdot \nabla_{X,z}U)=-\nabla_{X,z}P+\rho g\), \(\partial_{t}\rho +U\cdot \nabla_{X,z}\rho =0\), \(\nabla_{X,z}\cdot U=0\), \(t\geq 0\), \((X,z)\in \mathcal{S}=\mathbb{R}^{d}\times (-H,0)\), where \(U\) is the velocity field, \(P\) the pressure and \(\rho\) the density, which is not necessarily constant. The boundary conditions \(w|_{z=-H}=w|_{z=0}=0\) are imposed, where \(w\) is the vertical component of the velocity \(U\). The authors introduce equilibrium solutions \((U_{eq},\rho_{eq},P_{eq})\) which depend only on the vertical variable \(z\) of the form: \(U_{eq}=0\), \(\rho_{eq}=\rho_{eq}(z)\), \(\frac{d}{dz} P_{eq}=-\rho_{eq}g\), and observing that perturbations of such equilibrium solutions give rise to the propagation of internal waves, they look for exact solutions \((U_{ex},\rho_{ex},P_{ex})\) to the above Euler equations of the form: \(U_{ex}=\varepsilon U\), \(\rho_{ex}=\rho_{eq}+\varepsilon \rho\), \(P_{ex}=P_{eq}+\varepsilon P\), where \(\varepsilon >0\) is a parameter measuring the amplitude of the perturbation and \((U,\rho ,P)\) is the solution to the modified Euler equations \(\partial_{t}U+\varepsilon U\cdot \nabla_{X,z}U=-\frac{1}{\rho_{eq}+\varepsilon \rho }\nabla_{X,z}P-\frac{\rho }{\rho_{eq}+\varepsilon \rho }g\), \(\partial_{t}\rho +\varepsilon U\cdot \nabla_{X,z}\rho +w\frac{d }{dz}\rho_{eq}=0\), \(\nabla_{X,z}\cdot U=0\), \(t\geq 0\), with the boundary conditions \(w|_{z=-H}=w|_{z=0}=0\). The first main result proves a local existence under the hypotheses: \(0<\varepsilon \leq 1\), \(\nu >d+2\), \(\rho_{eq}\in W^{\nu ,\infty }(-H,0)\), and \(\rho_{0},U_{0}\in H^\nu ( \mathcal{S})\) are such that \(\nabla_{X,z}U_{0}=0\) and there exists \(\rho_{\min}>0\) such that \(\inf_{z\in \lbrack -H,0]}\rho_{eq}(z)\geq \rho_{\min}\) and \(\inf_{\mathcal{S}}(\rho_{eq}+\varepsilon \rho_{0})\geq \rho_{\min}\). The authors prove the existence of \(T>0\) such that for all \(\varepsilon \in (0,1]\), there is a unique solution \((U,\rho )\in C([0,T];H^{\nu }(\mathcal{S} )^{d+2}\)) to the modified Euler equations with initial data \((U_{0},\rho_{0})\). Taking \(\varepsilon =0\) in the modified Euler equations, the authors obtain a linearized approximation, for which they prove the existence of a unique solution \((U^{\mathrm{lin}},\rho ^{\mathrm{lin}})\in C([0,T];H^{\nu }(\mathcal{S} )^{d+2})\). They prove an estimate of order \(\varepsilon\) between the two solutions. They then show that the vertical velocity \(w\) satisfies \(\partial_{t}^{2}((\Delta +\frac{1}{\rho_{eq}}\partial_{z}(\rho_{eq}\partial_{z}\cdot )w)+N^{2}w=0\), where \(N=N(z)\) is the Brünt-Vaisälä frequency defined through \(N^{2}=-\frac{\rho_{eq}^{\prime }}{\rho_{eq}}g\), and they decompose \(w\) on the orthonormal basis \((f_{n})_{n}\in \mathbb{N}\) of \(L^{2}([-H,0],\rho_{eq}N^{2}dz)\) formed by the eigenfunctions of the Sturm-Liouville problem \(\frac{d}{dz}(\rho_{eq}\frac{d}{dz}f_{n})+\frac{ \rho_{eq}N^{2}}{c_{n}^{2}}f_{n}=0\), \(f_{n}(-H)=f_{n}(0)=0\). They successively deeper analyze the cases \(N\) constant and positive and \(N\) non constant. They finally consider the shallow water regime, further assuming the strong Boussinesq assumption that is that the density is constant. The paper ends with some further considerations, for example taking into account the Coriolis effects and the case of nonsteady waves involving Kelvin-Helmholtz instabilities.