betaFeedback

WorksPhysics and Astronomy

Space-time resonances in the spatiotemporal spectrum of nonlinear dispersive waves

Abstract

In weakly nonlinear dispersive wave systems, long-time dynamics are typically governed by time resonances, where wave phases evolve coherently due to exact frequency matching. Recent advances in spatio-temporal spectrum measurements, however, reveal prominent features that go beyond the predictions of time resonance theory. In this work, we develop a theoretical framework to interpret these signatures by identifying and characterizing an alternative mechanism: space resonances. These arise when wave packets share the same group velocity and remain co-located, leading to long-lived interactions. We further show that gauge-breaking terms in the Hamiltonian give rise to space resonances supported on negative frequencies. By combining sea-surface elevation data, numerical simulations, and analytical theory, we derive the leading-order spatio-temporal spectrum for weakly interacting water waves, providing a unified explanation for its observed features.

Introduction

For non-linearly interacting waves, a central question is to determine which interactions govern the evolution at different time scales. A natural lens for addressing this question is the spatiotemporal Fourier transform which reveals how energy is distributed in wavenumber and frequency.

In the case of small amplitude dispersive waves, long-time behavior is often attributed to time-resonant interactions between Fourier modes. NN waves of the form A(kℓ)eiω(kℓ)tA(k_\ell)e^{i\omega(k_\ell)t}, ℓ=1,…,N\ell= 1,\ldots,N, where ω(k)\omega(k) denotes the linear dispersion relation, are said to be time-resonant if:

k1±k2±⋯±kN=0,ω(k1)±ω(k2)±⋯±ω(kN)=0,k_1 \pm k_2 \pm\cdots\pm k_N = 0,\qquad\omega(k_1) \pm\omega(k_2) \pm\cdots\pm\omega(k_N) = 0,

for some combination of signs. When these conditions hold, oscillatory phases cancel, interactions do not average out, and energy may transfer persistently among modes.

To further analyze this problem, it is now possible, due to recent experimental and numerical advances, to measure wave fields simultaneously in space and time, enabling the computation of the spatio-temporal spectrum: the wavenumber–frequency (denoted by (k,σ)(k,\sigma)) spectrum via a Fourier transform in both variables. These data reveal, however, that the Fourier transform of a small-amplitude water-wave surface η(x,t)\eta(x,t) is not spread uniformly, nor confined to the linear dispersion curve alone. Instead, energy localizes sharply along additional surfaces in (k,σ)(k,\sigma) that cannot be predicted from time-resonance conditions alone, see Figure 1 and [1–7]. Identifying the mechanism that selects these localization surfaces is therefore essential for understanding which nonlinear interactions actually dominate the dynamics.

Spatio-temporal spectrum of a linear combination of surface elevation and velocity potential on the surface, $|a(k,\sigma)|$

Figure 1. Spatio-temporal spectrum of a linear combination of surface elevation and velocity potential on the surface, ∣a(k,σ)∣|a(k,\sigma)|, from the numerical simulation of water wave equations. Quantities in the figure are non-dimensional, corresponding to rescaled quantities in the simulation. Five branches discussed in the text are marked in the figure, including the main branch (fitted by σ=g∣k∣\sigma=\sqrt{g|k|} for negative kk), branch A (σ=g∣k∣\sigma=\sqrt{g|k|} for positive kk), branch B (σ=2g∣k∣\sigma=\sqrt{2g|k|} for negative kk), branch C (approximately by σ=cgk\sigma=c_g k near the origin) and branch D (σ=−g∣k∣\sigma=-\sqrt{g|k|} for positive kk).

Surface gravity waves represent one of the most celebrated and enduring problems in fluid mechanics [8]. Their equations are remarkably rich, capturing phenomena from gentle ripples to tsunamis, yet notoriously difficult to analyze due to their full nonlinearity and nonlocality. The subject has a long distinguished history, spanning oceanography (Hasselmann’s kinetic theory [9]), mathematical physics (Zakharov’s Hamiltonian formulation [10]), and rigorous analysis (from Stokes’ expansions [11] to modern results on well-posedness and asymptotics see [12, 13] and references there-in). Despite this maturity, the mechanism responsible for selective amplification of some non-resonant branches in the case of many interacting waves remained unidentified.

In this manuscripts we study analytically the spatiotemporal spectrum and reveal that this localized structure arises from a second class of long-lived interactions, which we call space-resonances. Two interacting wave packets may exchange energy coherently if they share the same group velocity, without being time-resonant. Such packets remain co-located in space, creating a sustained interaction that leaves a clear spectral signature. These interactions concentrate energy along non-linear branches in (k,σ)(k,\sigma), away from the main dispersion curve where time-resonances reside. Space-resonances thus explain the strongly enhanced, selective appearance of certain non-resonant harmonic surfaces observed in the spectrum.

While gravity waves are our central example, our framework extends naturally to nonlinear optics, plasma waves, Bose-Einstein condensates, and other dispersive systems. We first develop the theory in a simplified model of nonlinear wave interaction, obtain an analytic leading-order formula for the spatio-temporal spectrum, then apply it to water-wave data and verify its predictions. We present numerical explorations of other systems as well.

In nonlinear systems, the spectrum distinguishes free modes, which are waves that share the same signature as the linear dispersion relation, from bound modes, which arise through nonlinear interactions and deviate from the linear dispersion relation. These two classes of waves produce distinct signatures in (k,σ)(k,\sigma)–space. A commonly adopted approach for analyzing these modes is to apply perturbation theory and normal form transformations to eliminate non-resonant terms from the system [14–16]. One then obtains new variables representing the free wave field, which oscillate at the linear frequency (with nonlinear corrections) and whose dynamics are governed by resonant and near-resonant interactions, described within wave turbulence theory by the Wave Kinetic Equation.

In [14], Zakharov analyzed surface gravity waves using the perturbative and normal form approach described above. For deep water, exact three-wave time resonances are excluded by the concave nature of the dispersion relation, so a normal form transformation is applied to obtain the dynamics of the free wave field. From this framework, the spectral wave action density associated with the physical water wave variables (surface elevation and velocity potential) can be analytically reconstructed starting from the spectrum of the free waves [16]. Although the bound modes themselves are not resonant, their inclusion in the normal form transformation modifies the statistical properties of the wave field, influencing many statistical observables [16]. In this procedure, all non-resonant modes are treated on equal footing and are therefore expected to appear uniformly in the spatio–temporal spectrum. However, subsequent experimental studies revealed that some non-resonant modes are more strongly enhanced than others, as demonstrated in Figure 1. For single nearly monochromatic waves this can be attributed to Stokes harmonics, but in broadband multi-wave systems this explanation fails. The selective amplification seen in real spectra therefore requires a different mechanism.

We show that the missing mechanism is space-resonance, and that even when higher-order time-resonances arise, the spectral imprint of space-resonances persists. We argue that, in the absence of time resonances, the dynamics are governed by space-resonant interactions.

The space resonance was originally introduced in a purely mathematical context, where it played a central role in proving long-time existence and scattering results for dispersive PDEs. In particular, the absence of space-resonances was used to improve long-time existence for small-amplitude water-wave solutions [20] [17, 18, 19], while their intersection with time-resonances was used to show how linear scattering is modified in the integrable 1D Schrodinger equation [21] and 3D wave equations [22]. Their presence without time-resonances, however, has not been investigated previously. Here we study precisely this regime, focusing on leading-order nonlinear interactions in which time-resonances are absent. We demonstrate that space-resonances redistribute energy across Fourier space and thereby shape the evolving solution, dominating the dynamics over intermediate nonlinear times.

To briefly illustrate the concept of space-resonances and their relation to the spatio-temporal spectrum, let us consider the interaction of two wave packets in a weakly nonlinear three-wave system. A wave packet, i.e., a localized superposition of plane waves with wavenumbers near k0k_0, has the form

ψ(x,t)=∫RdA(k−k0)ei(k⋅x−ω(k)t) dk\psi(x,t)=\int_{\mathbb{R}^{d}} A(k-k_0)e^{i(k\cdot x-\omega(k)t)}\,\mathrm{d}k

where AA is strongly localized. The envelope propagates at the group velocity

vg:=∇kω(k0),v_g:=\nabla_k\omega(k_0),

which represents the velocity at which energy or information are transported. When two such packets interact, they produce a new packet with Fourier transform

I(t,k)=∫RdB(k,k1)eitϕ(k,k1) dk1,ϕ(k,k1):=ω(k1)+ω(k−k1),I(t,k)=\int_{\mathbb{R}^{d}}B(k,k_1)e^{it\phi(k,k_1)}\,\mathrm{d}k_1,\qquad\phi(k,k_1):=\omega(k_1)+\omega(k-k_1),

where BB contains amplitude factors. Applying the stationary phase lemma gives, for t≫1t\gg1,

I(t,k)=B(k,k1∗)eitϕ(k,k1∗)(2πt)d/2eiπ4sgn⁡(∇k12ϕ(k,k1∗))∣det⁡∇k12ϕ(k,k1∗)∣+O(t−(d+1)/2),I(t,k)=B(k,k_1^*)e^{it\phi(k,k_1^*)}\left(\frac{2\pi}{t}\right)^{d/2}\frac{e^{i\frac{\pi}{4}\operatorname{sgn}(\nabla_{k_1}^{2}\phi(k,k_1^*))}}{\sqrt{\left|\det\nabla_{k_1}^{2}\phi(k,k_1^*)\right|}}+O\left(t^{-(d+1)/2}\right),

where k1∗=k1∗(k)k_1^*=k_1^*(k) satisfies

∇k1ϕ(k,k1∗)=0.(1)\nabla_{k_1}\phi(k,k_1^*)=0. \tag*{(1)}

The solutions of (1) correspond to interacting packets with identical group velocities, i.e., space-resonant wave numbers. In the (k,σ)(k,\sigma) spatio-temporal spectrum their contributions are shown to be strongly localized near

σ=ϕ(k,k1∗).(2)\sigma=\phi(k,k_1^*). \tag*{(2)}

Notice how (2) always includes the curve σ=2ω(k/2)\sigma= 2\omega(k/2); this curves is evident in numerical and experimental data, such as those provided in Section II and Figure 1. Several other features of the spatio-temporal spectrum of weakly interacting dispersive systems can also be explained by the above theoretical framework, including the appearance of branches with negative σ\sigma (for systems that break gauge invariance), and the presence of an almost straight branch near zero modes.

Organization of the paper. The paper is organized as follows. Section II presents spatio-temporal spectra from experiments on surface gravity waves and from numerical models with quadratic nonlinearities that generate non-resonant three-wave interactions. In addition to the known second- and higher-order branches, we identify a branch near the zero mode and observe negative-frequency components [23, 24]. Section III develops a simplified theoretical model for space resonances and Section IV applies it to the weakly nonlinear water-wave equations to obtain the leading-order spatio-temporal spectrum.

The spatiotemporal spectrum from experimental and numerical data

Before tackling the problem from a theoretical perspective, we find it useful for the clarity of the paper to present first some experimental and numerical approaches in which we compute the spatio-temporal spectrum.

Stereoscopic measurements of surface gravity waves from the Acqua Alta Tower

Historically, surface gravity wave measurements have been carried out using single-point instruments, such as buoys or wave gauges, which record time series data from which classical frequency spectra can be extracted. More recently, stereoscopic techniques have made it possible to capture both spatial and temporal information over areas spanning about thousand square meters [25, 26]. The dataset examined in this study was obtained from the Acqua Alta oceanographic research tower, located in the northern Adriatic Sea (Italy) at 15 km offshore from the Venetian coast, in 17 meters of water. Data were acquired using a pair of cameras simultaneously capturing the same area. Sea surface elevation, η(x,y,t)\eta(x,y,t), was measured using a stereographic method [25, 27]. The resulting dataset consists of approximately 28 minutes of recordings over a square area with side length ℓ=32.6\ell= 32.6 m, sampled at a frequency of 12 Hz. More information on the data set can be found in [28]. The dataset represents a suitable framework for studying the spatio-temporal spectrum. In this study, we derive the so-called normal variable a(k,t)a(k,t) from the surface elevation field η(x,y,t)\eta(x,y,t). It is defined as

a(k,t)=g2ω(k)η(k,t)−iω(k)2gψ(k,t),(3)a(k,t) = \sqrt{\frac{g}{2\omega(k)}}\eta(k,t) - i\sqrt{\frac{\omega(k)}{2g}}\psi(k,t), \tag*{(3)}

where k=(kx,ky)k = (k_x,k_y) is a two-dimensional wavevector, ∣k∣|k| denotes its magnitude and ω(k)=g∣k∣\omega(k) = \sqrt{g|k|}, with gg the gravity acceleration. ψ(k,t)\psi(k,t) is the velocity potential evaluated on the free surface, [14, 15]. Since our measurements only provide the surface elevation and not the surface velocity, we estimate ψ(k,t)\psi(k,t) using a leading-order approximation:

ψ(k,t)=gω(k)2dη(k,t)dt.\psi(k,t) = \frac{g}{\omega(k)^2}\frac{d\eta(k,t)}{dt}.

The rationale behind the introduction of the normal variable will be discussed in detail in Section IV. Once the normal variable has been computed from the experimental data, we perform an additional Fourier transform in time to obtain a(k,σ)a(k,\sigma). The wave number-frequency spectrum is then given by ∣a(k,σ)∣2|a(k,\sigma)|^2. To facilitate visualization, we convert from Cartesian coordinates (kx,ky)(k_x,k_y) to polar coordinates (∣k∣,θ)(|k|,\theta) and integrate over the angular variable. The resulting spectrum is shown in Figure 2. The plot reveals the presence of multiple branches. The dominant one appears at positive frequencies and corresponds to free waves, closely following the linear dispersion relation σ=ω(k)=g∣k∣\sigma= \omega(k) = \sqrt{g|k|}. As expected, a higher-order branch is also visible, which is well described by σ=2ω(k/2)=2g∣k∣\sigma= 2\omega(k/2) = \sqrt{2g|k|}. This latter relation will be examined in detail in Section IV. Interestingly, a negative-frequency branch is also observed, with an intensity that is not symmetric with respect to its positive counterpart. An interpretation of these negative frequencies will be provided in the theoretical section of the paper.

Figure 2. Spatio-temporal spectrum of sea surface elevation measured at the Acqua Alta oceanographic tower, situated 15 km offshore from Venice. The red dashed lines correspond, starting from the top one, to σ=2g∣k∣\sigma= \sqrt{2g|k|}, σ=g∣k∣\sigma= \sqrt{g|k|}, σ=−g∣k∣\sigma= -\sqrt{g|k|}, σ=−2g∣k∣\sigma= -\sqrt{2g|k|}

Numerical simulations of the water wave equations

We supplement the experimental evidence in Section II A with numerical simulations of water waves. In this section, we focus on the setup with a one-dimensional (1D) free surface, since the results are sufficient to reveal all key signatures on the spatio-temporal spectrum we study theoretically in this paper. The simulations with two-dimensional free surfaces have been conducted before, with the results of spatio-temporal spectra available in [29] [3]. For gravity waves on a 1D free surface of an incompressible, inviscid and irrotational fluid, the flow can be described by a velocity potential ϕ(x,z,t)\phi(x,z,t) satisfying the Laplace’s equation, with xx and zz the horizontal and vertical coordinates and tt the time. The surface velocity potential is defined as ψ(x,t)=ϕ(x,z,t)∣z=η\psi(x,t)=\left.\phi(x,z,t)\right|_{z=\eta}, where η(x,t)\eta(x,t) is the surface elevation. The evolutions of η\eta and ψ\psi satisfy the Euler equations in Zakharov form [30]:

ηt+ηxψx−(1+ηx2)ϕz=0,(4)\eta_t+\eta_x\psi_x-(1+\eta_x^2)\phi_z=0, \tag*{(4)}
ψt+gη+12ϕx2−12(1+ηx2)ϕz2=0,(5)\psi_t+g\eta+\frac{1}{2}\phi_x^2-\frac{1}{2}(1+\eta_x^2)\phi_z^2=0, \tag*{(5)}

where ϕz(x,t)=∂ϕ/∂z∣z=η\phi_z(x,t)=\left.\partial\phi/\partial z\right|_{z=\eta} is the surface vertical velocity.

References

  1. [20]A. D. Ionescu and F. Pusateri, Global solutions for the gravity water waves system in 2d, Inventiones mathematicae 199, 653 (2015).
  2. [21]J. Kato and F. Pusateri, A new proof of long-range scattering for critical nonlinear schrödinger equations, (2011).
  3. [22]Y. Deng and F. Pusateri, On the global behavior of weak null quasilinear wave equations, Communications on Pure and Applied Mathematics 73, 1035 (2020).
  4. [23]J. Zaleski, M. Onorato, and Y. V. Lvov, Anomalous correlators in nonlinear dispersive wave systems, Physical Review X 10, 021043 (2020).
  5. [24]A. Villlois, G. Dematteis, Y. V. Lvov, M. Onorato, and J. Shatah, Anomalous correlators, negative frequencies and non-phase-invariant hamiltonians in random waves, arXiv preprint arXiv:2502.20574 (2025).arxiv.org/abs/2502.20574
  6. [25]A. Benetazzo, Measurements of short water waves using stereo matched image sequences, Coastal engineering 53, 1013 (2006).DOI
  7. [26]P.-C. Shih, Offshore stereo measurements of gravity waves, Coastal Engineering 64, 127 (2012).
  8. [27]F. Bergamasco, A. Torsello, M. Sclavo, F. Barbarol, and A. Benetazzo, Wass: An open-source pipeline for 3d stereo reconstruction of ocean waves, Computers & Geosciences 107, 28 (2017).DOI
  9. [28]P. V. Guimarães, F. Ardhuin, F. Bergamasco, F. Leckler, J.-F. Filipot, J.-S. Shim, V. Dulov, and A. Benetazzo, A data set of sea surface stereo images to resolve space-time wave fields, Scientific data 7, 145 (2020).DOI
  10. [29]Z. Zhang and Y. Pan, Role of triad interactions in spectral evolution of surface gravity waves in deep water, Journal of Fluid Mechanics 1008, A41 (2025).DOI
  11. [30]V. E. Zakharov, Stability of periodic waves of finite amplitude on the surface of a deep fluid, Journal of Applied Mechanics and Technical Physics 9, 190 (1968).DOI
  12. [31]D. G. Dommermuth and D. K. Yue, A high-order spectral method for the study of nonlinear gravity waves, Journal of Fluid Mechanics 184, 267 (1987).
  13. [32]B. J. West, K. A. Brueckner, R. S. Jand, D. M. Milder, and R. L. Milton, A new method for surface hydrodynamics, J. Geophys. Res. 92, 11803 (1987).
  14. [33]Y. Pan, High-order spectral method for the simulation of capillary waves with complete order consistency, Journal of Computational Physics 408, 109299 (2020).DOI
  15. [34]K. Hasselmann, T. P. Barnett, E. Bouws, H. Carlson, D. E. Cartwright, K. Enke, J. Ewing, A. Gienapp, D. Hasselmann, P. Kruseman, et al., Measurements of wind-wave growth and swell decay during the joint north sea wave project (jonswap), Ergaenzungsheft zur Deutschen Hydrographischen Zeitschrift, Reihe A (1973).
  16. [35]J. Shatah, Normal forms and quadratic nonlinear klein-gordon equations, Comm. Pure and Appl. Math. 38, 685 (1985).
  17. [36]A. Ionescu and J.-M. Delort, Sobolev estimates for two dimensional gravity water waves, Astérisque 374 (2015).
  18. [37]A. Ionescu and F. Pusateri, Global regularity for 2d water waves with surface tension, Mem. Amer. Math. Soc. 256, 123 (2018).
  19. [38]M. Berti, R. Feola, and F. Pusateri, Birkhoff normal form and long time existence for periodic gravity water waves, Comm. Pure Appl. Math. 76, 1416 (2023).
  20. [39]Y. Deng, A. Ionescu, and F. Pusateri, On the wave turbulence theory of 2d gravity waves, i: deterministic energy estimates, Comm. Pure and Appl. Math. 78, 211 (2025).arxiv.org/abs/2211.10826
  21. [40]P. Germain, N. Masmoudi, and J. Shatah, Global solutions for 3d quadratic schrödinger equations, International Mathematics Research Notices 2009, 414 (2009).
  22. [41]J. Kato and F. Pusateri, A new proof of long range scattering for critical nonlinear schrödinger equations, Differential and Integral Equations 24, 923 (2011).
  23. [42]Y. Deng and F. Pusateri, On the global behavior of weak null quasilinear wave equations, Comm. Pure Appl. Math. 73, 1035 (2020).
  24. [43]K. Hasselmann, On the non–linear energy transfer in a gravity–wave spectrum. Part I: General theory, J. Fluid Mech. 12, 481 (1962).
  25. [44]K. Hasselmann, On the non–linear energy transfer in a gravity–wave spectrum. part 2, J. Fluid Mech. 15, 273 (1963).
  26. [45]K. Hasselmann, On the non–linear energy transfer in a gravity–wave spectrum. {P}art {3}, J. Fluid Mech. 15, 385 (1963).
  27. [46]P. A. Janssen and M. Onorato, The intermediate water depth limit of the zakharov equation and consequences for wave prediction, Journal of Physical Oceanography 37, 2389 (2007).DOI
  28. [47]L. J. Tick, A non-linear random model of gravity waves, Journal of Mathematics and Mechanics , 643 (1959).DOI
  29. [48]G. Komen, Nonlinear contributions to the frequency spectrum of wind-generated water waves, Journal of Physical Oceanography 10, 779 (1980).doi.org/10.1175/1520-0485(1980)010<0779:ncttfs>2.0.co;2

Paper details

vlnc:W_wtkktyv3
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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