WorksPhysics and Astronomy
The U2H map explains the effect of (sub)mesoscale currents on significant wave height statistics
Abstract
Currents modulate the energy of surface gravity waves, leading to spatial inhomogeneities in significant wave height (SWH). Previous work indicates that the overall scale of the inhomogeneities is set by the scale of the currents, that the inhomogeneities are strongly anisotropic even for isotropic currents, and that the rotational and divergent components of the currents have sharply distinct effects. We explain these and other features of current-induced SWH inhomogeneities using the U2H map, a linear relation between SWH and currents deduced from wave-action conservation by making simplifying assumptions. We obtain a linear law relating the spectrum of SWH to the spectra of rotational and divergent kinetic energy of the current. This makes it possible to relate SWH statistics (such as variance and anisotropy) to the current statistics and wave properties including directional spreading.
Introduction
Ocean surface gravity waves (SGWs) are primarily driven by wind but are influenced by ocean currents, which modify their direction, frequency, and amplitude [16, 4]. In recent years, recognition of the role of meso- and submesoscale turbulence in modulating SGWs has grown, driven by advances in numerical modeling (e.g., [18, 1, 21, 27]) and remote sensing observations (e.g., [19, 22, 17, 14]).
While currents affect all moments of the SGW spectrum, recent studies have particularly focused on their influence on significant wave height (SWH), which is routinely observed by satellite altimeters and widely used in air–sea flux parameterizations (e.g., [24, 8]). Collectively, findings from this series of recent studies on wave–current interactions suggest the following:
(i) rotational currents induce larger SWH anomalies than do divergent currents;
(ii) the spatial scales of the SWH anomaly are determined by those of the current, i.e. similar spectral slopes;
(iii) the SWH response to divergent currents is local (largest where the current is strongest) but the response to rotational currents is non-local and anisotropic.
Despite progress in our understanding of the relationship between currents and SWH anomalies, the exact mechanisms leading to points (i)–(iii) remain unclear.
Figure Figure 1 shows the SWH anomaly, denoted , induced by synthetic random currents. The mean kinetic energy (hereafter “KE”) of the currents is fixed, while the slope of the KE spectrum and the ratio between the rotational and divergent energy of the current are varied. Features (i), (ii) and (iii) are evident: the magnitude of decreases as the proportion of rotational currents decreases, the spatial scales of are larger when the spatial scale of the current is larger, and the patterns of are stretched in the primary wave direction (chosen as the positive -direction) when currents are rotational.
Figure 1. Impact of currents on SWH anomaly . The current speed, , is shown in the first column. The KE spectrum of the current is isotropic and the isotropic spectra follow the power laws (top row) and (bottom row), where is the wavenumber. The relative SWH anomaly is shown for four values of the parameter defined in (32) controlling the ratio of rotational to divergent current KE (columns two through five, with purely rotational current in column two and purely divergent current in column five). The waves have a wave action spectrum with directional-width parameter in (14) and the primary wave direction is along the -axis. See § 4 for details.
The main difference between Figure Figure 1 and figure 4 of Villas Bôas et al. [27] is the way in which SWH anomalies are computed: here we obtain using the U2H map developed in Wang et al. [29] (hereafter W25), while in Villas Bôas et al. [27] is obtained by running WAVEWATCH III (WAVE height, WATer depth and Current Hindcasting, Tolman et al. 2009 – hereafter we refer to this model as WW3). U2H is a linear map giving the SWH anomalies directly as a function of the current without resorting to solving the full wave-action conservation equation. In W25, we compare the U2H map against WW3 in a few current configurations and establish a good match between the two methods.
While Villas Bôas et al. [27] focus their attention on wave spectra with small directional spreading, hereafter “highly directional spectra”, here we do not. We examine how the directional spreading of the wave spectrum affects features (i), (ii) and (iii). This is illustrated in figure Figure 2 which is similar to figure Figure 1 but for a wave spectrum with large directional spreading. For this spectrum, features (i) and (ii) continue to hold, though less evidently for (i), but feature (iii) does not. Comparing figures Figure 1 and Figure 2 – and noting the difference in the color maps – shows that: (iv) the SWH anomaly is larger for background waves with smaller directional spreading.
Figure 2. Same as Figure 1 but with directional-width parameter in (14) corresponding to a spectrum with larger directional spreading. Note the difference in color scale compared with Figure 1.
Here we take advantage of the simple, explicit form of U2H to explain and quantify features (i)–(iv). We also clarify the relation between current and SWH spectra, refining our understanding of feature (iii), particularly where rotational and divergent currents have different spatial scales.
U2H map: the recipe
The W25 recipe for the U2H map has two ingredients:
the SGW action spectrum in the absence of currents, hereafter referred to as the background spectrum, , assumed independent of time and space;
the surface current velocity , assumed independent of time and with spatial scales much larger than SGW wavelengths.
The background significant wave height associated with is
where is the gravitational acceleration, , and
is the intrinsic frequency of deep-water SGWs. is constant in space.
The current modulates the SGWs, leading to a spatially dependent SWH
where is the current-induced SWH anomaly. The U2H map assumes that the typical current speed is small compared to the group speed of the waves:
U2H then provides an approximation to the relative SWH anomaly at order .
The map is best formulated using the Helmholtz decomposition of the current into purely divergent and purely rotational parts,
where and are the velocity potential and streamfunction. We introduce the Fourier transform
of the current velocity, with the Fourier vector (not to be confused with the SGW wavevector ). Eq. (5) becomes
where , is obtained from by rotation by about a vertical axis, and and are unit vectors. In terms of the polar representation , . The U2H map expressed in Fourier space is then
and are determined by the background action spectrum and control the impact of the divergent and rotational parts of the current on . (These functions are denoted as and in W25, respectively.) A crucial result is that and depend on the Fourier vector only through the polar angle . They are given by
and
where
The wave momentum in (9) and (10) is
Equations (8)–(12) include all the formulas required to compute the U2H map from the background action spectrum and current velocity . These formulas are linear in and , and require minimal computational resources. The computational implementation is straightforward, as described and illustrated with a Jupyter notebook in W25.
In addition to assuming that and depend only on and , that the spatial scale of is much larger than SGW wavelengths, and that is small, the U2H map neglects the vertical shear of , forcing, dissipation and wave–wave interactions. These assumptions can be violated in realistic scenarios. The assumptions of large group speed and negligible forcing/dissipation often do not apply to short wind waves; as a result, the U2H map is applicable mainly to freely propagating surface waves with small steepness distant from their generation sites, e.g., swell. The independence of background spectrum on time and space is also questionable for wave groups, and for rapidly evolving currents such as tides.
In examples shown here, we follow W25 and assume that the background spectrum takes a separable form
We take to be a truncated Gaussian distribution in intrinsic frequency concentrated around a peak value (see appendix A for details). We set for all the results of the paper except in section 7 which considers the sensitivity to .
The angular dependence follows the model of Longuet-Higgins et al. [15]:
where is the primary angle of wave propagation taken as 0, and is the directional-width parameter, with large corresponding to a small directional spreading.
The model of (13)–(14) is chosen for simplicity. We emphasize that the U2H map can be employed with arbitrary background spectra, including spectra that are not separable in and , and that our qualitative conclusions do not depend on our specific choice of the LHCS model spectrum. Different forms of the background spectra affect U2H only through the integral in (11). In the Jupyter Notebook published in W25, arbitrary wave spectra can be prescribed by users. For highly directional spectra, in particular, the directional-width parameter can be interpreted more generally for other spectral forms as , where is the directional spreading (e.g. [11]).
Figure 3 shows and for three values of . Because and are real,
Figure 3. (dashed lines) and (solid lines) computed from (11)–(12) for the LHCS spectrum (14) with and for the directional-width parameter , 10 and 40. A zoomed-in view for is shown at the upper right corner.
In figure 3 we exploit this symmetry to restrict the range of to . For highly directional spectra, i.e. , is localised around a maximum at and is much larger than . Asymptotic results reported in W25 explain these features by showing that is localised in regions around where it scales like . In contrast, is and -independent for .
Power spectral relations
W25 validate the U2H map (8) by comparing its predictions to WW3 simulations for a few specific current configurations. In this paper we focus on the relationship between the statistics of SWH and the statistics of implied by (8). We assume that is a homogeneous random field and define the spectra and cross-spectrum of the divergent and rotational parts of the current as
and
where denotes ensemble average.
From (8) the SWH spectrum, defined as the power spectrum
of the relative change in SWH, is given by
where denotes the real part. Eq. (20) is central to this paper. It provides a simple relationship between the SWH spectrum and the spectra characterising the currents.
We assume for simplicity that the currents are isotropic, with uncorrelated divergent and rotational parts. This reduces (20) to
where and are the current divergent and rotational KE isotropic spectra, with e.g.
The assumption of uncorrelatedness is not essential for many of our results, as discussed in appendix B.
While is not isotropic because of the angular dependence of and , we can define the SWH isotropic spectrum
where we abuse notation in marking the difference between isotropic and two-dimensional spectra only by displaying the different independent variable, vs. . Integrating (21) over leads to
where the coefficients and are
We examine the consequences of (24) in the next two sections before considering the anisotropy of in § 6.
Variance
We first study the variance of the SWH anomaly. Variance reflects the overall intensity of the anomaly, and helps understand features (i) and (iv) observed in § 1.
Integrating (24) with respect to and recalling from Parseval’s identity that the variance of a homogeneous random field is the integral of its spectrum, we obtain a relation between the variance of the SWH and the variances of the divergent and rotational current speeds:
Similarly, multiplying (24) by and integrating gives
where and denote surface divergence and vorticity.
Thus, the variance of SWH is a weighted sum of the variances of the speed of the divergent and rotational parts of the current. And the variance of the gradient of SWH is the weighed sum of the variances of the divergence and vorticity of the current. In both cases, the weights are and in (25).
Figure 4 shows and for the LHCS model spectrum (14) as functions of . For , . As increases, corresponding to smaller directional spreading, increases, as for (this can be established from the large- form of mentioned at the end of § 2), while saturates at a much smaller value. These properties of and explain features (i) and (iv) noted in § 1, i.e. the dominance of rotational currents over divergent currents, and the increase in the magnitude of SWH anomalies for smaller directional spreading.
To quantity these effects, we express and in terms of in (11) by substituting (9) and (10) into (25) to find
Combining the two equations above
and therefore, as observed in Figure 4,
The large- behaviour of and can be explained by the corresponding asymptotic form of and (see W25).
We illustrate (26) in Figure 5. This shows the SWH variance computed for an ensemble of synthetic, randomly drawn currents. For all currents, the isotropic KE spectra and have the same shape, corresponding to a power law over a finite range of with isotropic spectral slope , and (see Appendix C for details). The magnitudes of and are varied in such a way that the total kinetic energy is a constant (0.01 m/s per kilogram). Figure 5 shows as a function of the parameter such
and for three values of . As expected from (26), the SWH variance is independent of the spectral slope , linear in , and increasing with (since combining (26) with (32) shows the SWH variance is proportional to ). We have checked that the SWH gradient variance behaves as indicated by (27).
Isotropic spectrum
We now examine the implications of formula (24) for the SWH isotropic spectrum . According to (24), is the weighted sum of the divergent and rotational KE spectra and , with -independent weights and . This explains feature (ii) of § 1: the length scales of are set by those of the current.
Figure 5. Variance of the SWH anomaly as a function of the parameter controlling the ratio of divergent to rotational KE of the current. Three values of the spectral slope , and are considered. Estimates from random samples (dots) are compared with the prediction (26) (lines).
When the divergent and rotational spectra share an identical spectral slope, as in Villas Bôas et al. [27] and in Figure 1 and Figure 2, it follows from (24) that has the same spectral slope. In typical oceanic flows, however, the rotational spectrum is steep, dominating the total KE at large scales, and the divergent spectrum is shallower, dominating the total KE at small scales [3, 20, 2, 5]. For such flows (24) predicts that is parallel to at large scales and to at small scales. Because , the spectral slope of is steeper than or as steep as , and the transition scales between steep and shallow spectra arises at smaller scales for than for . The shift towards small scales is particularly pronounced for waves with small directional spreading () for which .
In Figure 6 we compare the SWH spectrum to the divergent, rotational and total KE spectra of the currents for (left panel) and (right panel). The energy spectra and are taken as the power laws and , respectively, with amplitudes such that the wavenumber where is . (These parameters are chosen for pedagogical purpose and are not intended to agree with realistic currents.) The background wave spectrum is as in § 4. The spectrum is obtained by computing for 10 random current realizations using the U2H map and averaging. This spectrum is as expected, making a transition from steep to shallow at a higher wavenumber (smaller spatial scales) than the transition in the total KE spectrum. In the case with small directional spreading , the transition is shifted to even higher wavenumbers. The spectral slope of is overall steeper than , with the difference more pronounced for smaller directional spreading. The steeper spectral slopes of are also visible in results reported in Ardhuin et al. [1].
Figure 6. Isotropic spectrum and KE spectra (total, divergent and rotational KE) for the examples in § 5. The wave directional-width parameter is 1 (left panel) and 10 (right panel). The left vertical axis is for and the right vertical axis is for the KE spectra. Spectra outside are not shown.
For currents such as those of Figure 6, dominated by rotation at large scales and by divergence at small scales, the steeper spectral slope of implies that the overall scale of the SWH anomaly is larger than the scale of the current, markedly so for waves with small directional spreading. This is confirmed by Figure 7 which shows in physical space for a single realization of the currents with the same parameters as for Figure 6. The anisotropy of is striking and demands quantification.
Figure 7. SWH anomaly (middle and right panels) induced by a current sample described in § 5. The left panel shows the current speed. The background waves follow the LHCS model described in appendix § 9, with directional-width parameter taken to be 1 (middle panel) or 10 (right panel).
Anisotropy
We first examine the anisotropy of the SWH anomaly by contrasting the one-dimensional spectra
and
corresponding to observations of along or across the primary wave direction.
Figure 8 shows these spectra together with analogous KE spectra for the current and wave parameters of Figure 6 and for and 10. In the along-wave direction, peaks at the lowest wavenumbers regardless of the spectral slope of the KE spectra. The slope of for small increases with , corresponding to increasingly large scales dominating the along-wave SWH signal.
Figure 8. Along/across-wave SWH spectra and , and total, divergent and rotation KE spectra as functions of wavenumbers or . The left vertical axis is for , and the right vertical axis is for the KE spectra. The wave directional-width parameter is 1 (left panel) and 10 (right panel). Spectra outside are not shown.
In the across-wave direction, peaks at scales close to the KE peak, but has a spectral slope closer to the rotational KE (steeper than the total KE for realistic oceanic currents) and thus leading to SWH scales larger than the current scales. These properties of and reflect the streaky nature of the SWH anomaly seen in Figure 1, Figure 2 and Figure 7. They can be fully understood by combining the closed-form expression (33)–(34) with the explicit expression (21) for and the form of and shown in Figure 4. In the limit of large , the asymptotic behaviors seen in Figure 3 (discussed in § 2) amplify these properties, leading to more streaky SWH anomalies.
Figure 4. ℓψ and ℓϕ as functions of s, assuming the LHCS angular dependence. The green curve indicates the s^{3/2} dependence of ℓψ that holds when s ≫ 1.
A simple measure of the anisotropy of is the aspect ratio of typical lengthscales in the along- and across-wave directions. This ratio can be defined by
In Figure 9, we show as a function of the directional-width parameter and for different values of the parameter characterizing the relative importance of the divergent and rotational parts of the current. The other parameters are a described in § 4, with the spectral slope fixed as . Except for purely divergent currents , is greater than 1, reflecting elongation of the SWH patterns in the along-wave direction, and increasing with for .
Figure 9. Aspect ratio of the SWH anomaly as a function of the directional-width parameter and for different values of the parameter controlling the ratio of rotational to divergent KE of the current.
For , , as is established by noting that
since and is localized within an boundary layer around . This provides a scaling for the elongation of the streaks in for waves with small directional spreading.
For purely divergent currents , the aspect ratio is a constant (because ), computed to be . This is why, in the rightmost panels of Figure 1 and Figure 2 (with ), is somewhat elongated in the across-wave -direction.
Sensitivity to mean frequency
In preceding numerical examples we use mean radian frequency , corresponding to the mean period . The background action spectrum depends on (see appendix C) and the U2H functions and inherit this dependence through the integral
in (11). In this section, we examine the sensitivity of and to .
In ocean-relevant situations, the dependence of and on is comparable to the dependence on the directional spreading parameter . This is inferred from figure 10 which shows and as functions of both and for wave spectra with fixed total energy, equivalently fixed . For oceanographic context, the leftmost panel of figure 10 shows an estimated probability density of and from 13 years (2010–2023) of wave buoy observations from Ocean Station Papa in the North Pacific Ocean. We choose Ocean Station Papa as a reference location due to its exposure to a broad range of wave conditions and its long history of meteorological and oceanographic observations [e.g. 13, 25 and references therein]. Directional wave buoy observations routinely provide the first four Fourier coefficients of the directional energy distribution, from which the mean frequency and mean directional spreading can be computed [11]. There are 195580 samples included. To aid visualization, extreme observations with , , or , are deleted (0.5% of the data is discarded). The probability density plot indicates the range of typical ocean values of and . As expected, short waves (large ) have larger directional spreading (larger ) than long waves.
The second and third panels of figure 10 show and computed by assuming the LHCS spectra for background waves, with directional spreading parameter . In summing the Fourier series (10), we cut at , after verifying that does not significantly change the results. For all relevant parameter values, . and in (25) increase with or, equivalently, with mean wavenumber , i.e. short high-frequency waves are more affected by currents and produce larger SWH anomalies than long low-frequency waves. The changes in produced by varying over the data range are comparable to those resulting from variations in . For , in particular, we can show that . The was noted in section 4; the stems from the approximation
and the quadratic dependence of on the left-hand side (see (9)–(10) and (25)).
Earlier discussion emphasized the sensitivity of to changes in . Here we emphasize that is equally sensitive to changes in mean frequency . The same conclusion applies to the function and hence to the SWH anomaly spectra.
Discussion
Key statistical features of current-induced SWH are explained by the U2H map which provides a linear relation between the anomalies of SWH and currents. U2H is deduced from the wave-action conservation equation by making a number of assumptions, chiefly the assumption (4) of current speeds small compared with wave group speeds. The physical effects captured by U2H identify the main mechanisms by which currents induce SWH anomalies. In U2H, advection of the action spectrum by currents is of secondary importance and is entirely neglected on the basis of . (See Ardhuin et al. [1] for simulation results with and without advection.) Instead, currents affect the distribution of wave action, and hence of SWH, by inducing changes in both the direction and wavelength of SGWs through refraction and the “concertina effect”, respectively).
In U2H, refraction and concertina effects superpose linearly. It is straightforward to trace their contribution through the derivation of : refraction is associated with directional variations of the background action and concertina effect with wavenumber variations . Refraction can then be shown to contribute to the series in expression (10) for . The remaining term in and the entirety of are proportional to , which can be attributed to a combination of refraction and concertina effect, with respective weights and . For highly directional spectra, and is dominated by the series in (10), so we conclude that refraction, driven by vorticity, is the primary mechanism of current-induced SWH anomalies. This is consistent with the classical result showing that under assumption (4) ray curvature is determined by vorticity [12, 10, 7, 9].
While this paper focuses on statistical predictions, we emphasize that U2H provides an approximation to the SWH anomaly induced by any specific current, as Figure 1 and Figure 2 illustrate. We also emphasize that the statistical assumptions we make (isotropy of the currents and, in some cases, independence of divergent and rotational currents) are not required by U2H. Rather, we make them to obtain simple, explicit predictions that explain generic features of wave–current interactions. The regime of waves with small directional spreading is particularly robust to details of the current spectrum because the SWH anomaly is then controlled by the spectrum of the rotational current in a narrow wedge in -space and insensitive to both and .
The framework of U2H can be generalized to map currents to current-induced anomalies of other wave quantities computable from wave action spectra, including directional spreading, mean periods, and the Stokes drift.
U2H depends on a linear approximation and requires that . To ensure that is small relative to , some restrictions are needed. In particular, small directional spreading leads to strong SWH anomalies, calling for an upper bound on . W25 discuss this and show that condition (4) needs tightening to
We show the curve or equivalently, in terms of the directional spreading and mean frequency , for m s in the probability density plot in Figure 10 (left panel). All data points are well above this green dashed line, indicating that the inequality (39) is satisfied. This suggests that (39) is only violated in rare circumstances, e.g. when current speeds are well in excess of 1 m s.
Figure 10. Joint probability density of wave frequency and directional spreading from Ocean Station Papa observations (left panel) and U2H functions (middle panel), and (right panel) as functions of , based on the LHCS model and the relation . The probability density is constructed from 3030 bins equally spaced in and with the color indicating the fraction of samples in each bin. The solid contour encloses bins with probability densities greater than 0.001 s; these bins enclose 92% of the data. Some values of corresponding to are marked on the right vertical axes. The green dashed line in the left panel indicates the boundary of the linear regime (see (39) in Discussion).
U2H is not applicable to length scales larger than km. Over large propagation distances, the cumulative scattering by currents leads to substantial broadening of the SGW spectrum, challenging the assumption that is independent of space. Smit and Janssen [23] and Villas Boas and Young [28] show that this broadening amounts to a directional diffusion of the action, ultimately leading to isotropic spectra (diffusion in frequency is weak as it depends on the time-dependence of the currents, see Cox et al. [6]). The comparison between U2H and WW3 simulations in Wang et al. [29] suggests this directional diffusion is negligible over the scales of hundred of kilometers considered in this paper. The neglect of wave–wave interactions also fails over large propagation distances. For quadruplet wave interactions, assuming a steepness parameter of and background wave period of s, the nonlinear interactions are negligible up to a spatial scale at order km.
Even with non-negligible source terms in the wave action equation, U2H can still provide a useful estimation of the SWH anomalies induced exclusively by currents. Accurately resolving the corresponding transport terms in the wave action equation introduces additional computational costs, as they call for small time steps and fine angular grids (see Figure 7 in Ardhuin et al. [1]). When such additional computational costs are prohibitive, the U2H framework can provide a fast estimate of the modulations induced by currents.
W25 note that, because the dependence of the SWH on both the divergent and rotational parts of the current, it is not in general possible to invert the U2H map and infer the current from SWH alone. It should however be possible to infer some useful statistical properties of the currents. For example, we show in § 6 that for waves with small directional spreading, the across-wave KE spectra tend to have similar spectral slopes to the along/across-wave rotational KE within their spectral supports. Thus, if the assumptions involved are met, one can infer the slopes of rotational KE spectra from SWH spectra.
Acknowledgments. This paper is a contribution to the projects M2, T2 and W2 of the Collaborative Research Centre TRR 181 “Energy Transfers in Atmosphere and Ocean” funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - Projektnummer 274762653, which supports HW. ABB is supported by the ONR MURI award N00014-24-1-2554, and NASA awards 80NSSC23K0979 through the International Ocean Vector Winds Science Team and 80NSSC24K1640 through the SWOT Science Team. JV is supported by the UK Natural Environment Research Council (grant NE/W002876/1). WRY is supported by the National Science Foundation award 2048583. We thank Jean Bidlot for interesting discussions. Comments from the two anonymous reviewers have improved the manuscript.
Data availability statement. Codes used for creating synthetic currents, computing the U2H map and plotting are accessible at https://github.com/hannnwang/U2H_examples.
Background wavenumber spectrum
The wavenumber dependence described by in (13) is as in Wang et al. [30]. We detail its form to introduce the two parameters that appear, namely the mean frequency and the frequency spread . We also show that the U2H map is insensitive to .
We specify an SGW energy spectrum of the form such that
and choose a frequency dependence in the form of the truncated Gaussian
with , which introduces the two parameters and .
The wavenumber dependence in (1) is related to via with . Therefore,
The wavenumber dependence affects the U2H map via the -integral in (11). Using (A3), this integral is computed with help of a symbolic integrator:
where
The function increases with positive . For , the value of ranges in , which corresponds to a variation of less than 2.8%, an amount we regard as negligible. For higher , the function increases faster: for , , meaning that when is as large as , the U2H map’s operators can change by 29% in magnitude.
We assume is less than 0.5 and so neglect the impact of on the U2H map. We set , which satisfies for the range of considered here. With these values, the truncated Gaussian in (A2) is indistinguishable from a full Gaussian and and are excellent approximations to the mean and standard deviations.
Cross-spectrum contribution
We show that provided that the cross-spectrum is isotropic, it plays no role in the isotropic spectrum of the SWH anomaly and hence in the variances of the SWH and of its gradient. Therefore the results of sections 4 and 5 hold also for isotropic currents with correlated divergent and rotational components.
In the power spectral relation (20), integration of the cross-spectrum term in (20) with respect to leads to the term
which we now show vanishes.
The operator in (9) can be written as
Clearly then, only terms proportional to and in can contribute to the integral (B1). Rearranging expression (10) for as
isolates these term in . We then compute
using that is real.
Numerical details of synthetic currents
For the synthetic currents in section 4, we draw current samples on square, 2500 km-wide grids with grid spacing of 2.5 km. For , the isotropic spectra and follow the power laws . For larger wavenumbers, the spectra are set to decay rapidly. For smaller wavenumbers, they are set to zero in order to alleviate artifacts from a square, non-isotropic spectral grid. The currents are drawn so that the Fourier coefficients of the potential and streamfunction follow zero-mean, random phase distributions, independent at each wavenumber except when the reality condition demands. The potential and streamfunction are independent.
The synthetic currents shown in figures 1 and 2 are drawn similarly. The only difference is that for sake of demonstration, we draw a single random function, say, and take the streamfunction and potential as and so that they are correlated. This choice is made in order that the current speeds remain unchanged in each row, despite different in Figure 1 and Figure 2.
References
- [1]Ardhuin, F., S. T. Gille, D. Menemenlis, C. B. Rocha, N. Rascle, B. Chapron, J. Gula, and J. Molemaker, 2017: Small-scale open ocean currents have large effects on wind wave heights. Journal of Geophysical Research: Oceans, 122 (6), 4500–4517.
- [2]Balwada, D., J. H. LaCasce, and K. G. Speer, 2016: Scale-dependent distribution of kinetic energy from surface drifters in the gulf of mexico. Geophysical Research Letters, 43 (20), 10–856.
- [3]Bühler, O., J. Callies, and R. Ferrari, 2014: Wave–vortex decomposition of one-dimensional ship-track data. Journal of Fluid Mechanics, 756, 1007–1026.
- [4]Cavaleri, L., B. Fox-Kemper, and M. Hemer, 2012: Wind waves in the coupled climate system. Bulletin of the American Meteorological Society, 93 (11), 1651–1661.
- [5]Chereskin, T. K., C. B. Rocha, S. T. Gille, D. Menemenlis, and M. Passaro, 2019: Characterizing the transition from balanced to unbalanced motions in the southern california current. Journal of Geophysical Research: Oceans, 124 (3), 2088–2109.
- [6]Cox, M. R., H. A. Kafiabad, and J. Vanneste, 2023: Inertia-gravity-wave diffusion by geostrophic turbulence: the impact of flow time dependence. Journal of Fluid Mechanics, 958, A21.
- [7]Dysthe, K., 2001: Refraction of gravity waves by weak current gradients. Journal of Fluid Mechanics, 442, 157–159.
- [8]Edson, J. B., and Coauthors, 2013: On the exchange of momentum over the open ocean. Journal of Physical Oceanography, 43 (8), 1589–1610.
- [9]Gallet, B., and W. R. Young, 2014: Refraction of swell by surface currents. Journal of Marine Research, 72 (2), 105–126.arxiv.org/abs/1410.1676
- [10]Kenyon, K. E., 1971: Wave refraction in ocean currents. Deep-Sea Research, 18, 1023–1034.DOI
- [11]Kuik, A., G. P. Van Vledder, and L. Holthuijsen, 1988: A method for the routine analysis of pitch-and-roll buoy wave data. Journal of Physical Oceanography, 18 (7), 1020–1034.
- [12]Landau, L. D., and E. M. Lifshitz, 2013: Fluid Mechanics: Course of Theoretical Physics, Volume 6. Elsevier.
- [13]Large, W., and S. Pond, 1981: Open ocean momentum flux measurements in moderate to strong winds. Journal of physical oceanography, 11 (3), 324–336.
- [14]Lenain, L., and N. Pizzo, 2021: Modulation of surface gravity waves by internal waves. Journal of Physical Oceanography, 51 (9), 2735–2748.
- [15]Longuet-Higgins, M. S., D. E. Cartwright, and N. Smith, 1963: Observations of the directional spectrum of sea waves using the motions of a floating buoy. Ocean Wave Spectra, S. EC, Ed., Prentice-Hall, Englewood Cliffs, New Jersey, 111–136.
- [16]Phillips, O. M., 1977: The Dynamics of the Upper Ocean. Cambridge University Press.
- [17]Quilfen, Y., and B. Chapron, 2019: Ocean surface wave-current signatures from satellite altimeter measurements. Geophys. Res. Lett., 46 (1), 253–261.
- [18]Rascle, N., B. Chapron, A. Ponte, F. Ardhuin, and P. Klein, 2014: Surface roughness imaging of currents shows divergence and strain in the wind direction. Journal of Physical Oceanography, 44 (8), 2153–2163.
- [19]Rascle, N., F. Nouguier, B. Chapron, A. Mouche, and A. Ponte, 2016: Surface roughness changes by finescale current gradients: Properties at multiple azimuth view angles. Journal of Physical Oceanography, 46 (12), 3681–3694.
- [20]Rocha, C., T. Chereskin, S. Gille, and D. Menemenlis, 2016: Mesoscale to submesoscale wavenumber spectra in Drake Passage. Journal of Physical Oceanography, 46 (2), 601–620.
- [21]Romero, L., D. Hypolite, and J. C. McWilliams, 2020: Submesoscale current effects on surface waves. Ocean Modelling, 101662.DOI
- [22]Romero, L., L. Lenain, and W. K. Melville, 2017: Observations of surface wave–current interaction. Journal of Physical Oceanography, 47 (3), 615–632.
- [23]Smit, P. B., and T. T. Janssen, 2019: Swell propagation through submesoscale turbulence. Journal of Physical Oceanography, 49 (10), 2615–2630.
- [24]Taylor, P. K., and M. J. Yelland, 2001: The dependence of sea surface roughness on the height and steepness of the waves. Journal of physical oceanography, 31 (2), 572–590.
- [25]Thomson, J., E. D’Asaro, M. Cronin, W. Rogers, R. Harcourt, and A. Sherberina, 2013: Waves and the equilibrium range at ocean weather station p. Journal of Geophysical Research: Oceans, 118 (11), 5951–5962.
- [27]Villas Bôas, A. B., B. D. Cornuelle, M. R. Mazloff, S. T. Gille, and F. Ardhuin, 2020: Wave–current interactions at meso-and submesoscales: Insights from idealized numerical simulations. Journal of Physical Oceanography, 50 (12), 3483–3500.
- [28]Villas Bôas, A. B., and W. R. Young, 2020: Directional diffusion of surface gravity wave action by ocean macroturbulence. Journal of Fluid Mechanics, 890, R3.
- [29]Wang, H., A. B. Villas Bôas, J. Vanneste, and W. R. Young, 2025: Scattering of surface waves by ocean currents: the U2H map. J. Fluid Mech., 1005, A12.
- [30]Wang, H., A. B. Villas Bôas, W. R. Young, and J. Vanneste, 2023: Scattering of swell by currents. J. Fluid. Mech., 975, A1.DOI









