{"id":"b451a5ae-aca1-4587-b7b1-80cc2573039a","arxiv_id":"2608.04687","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Surface curvature of a turbulent water surface is predicted from near-surface velocity gradients, and its magnitude is found to scale with the square of the horizontal divergence.","lead":"An experimental and theoretical study shows that the curvature of a turbulent water surface can be predicted from the velocity field measured a few millimeters below it, using the Euler equation with gravity and surface tension. This could allow remote optical measurements of ocean surfaces to infer the subsurface turbulence that controls gas and heat transfer across the air-water interface.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sign of the modeled curvature is inconsistent between the printed Euler-Laplace equations and Eq. (4.2); because the headline instantaneous result is a signed Pearson correlation r=0.54, this inconsistency must be resolved before the 'close quantitative agreement' claim can be interpreted.","rationale":"The reader identified the inviscid/small-slope assumption as the weakest point and noted the sign error only in passing. My stress-test makes the sign issue the central concern because it directly affects the signed correlation on which the instantaneous claim rests. The inviscid assumption is a physical modeling uncertainty that could reduce accuracy; the sign inconsistency is a verifiable internal contradiction that, if confirmed, changes the interpretation of every signed comparison. The statistical scaling in Fig. 6 uses squared quantities and is not endangered by the sign issue, so the paper's broader remote-sensing motivation remains plausible. The full model (2.4) may be correct in the implementation despite the printed inconsistency; that is exactly why a re-derivation and a sign-flip recomputation settle the matter. I therefore keep the reader's CONDITIONAL verdict: the paper is promising but must fix the sign convention and re-report the correlation and the quadrant interpretation before acceptance. If the re-derivation instead confirms the printed equations and the positive correlation is robust, then no substantive objection remains on this point.","tokens_in":6,"tokens_out":16103,"duration_ms":266051,"concrete_test":"Re-derive Eq. (2.1) from the linearized Euler and free-surface boundary conditions for a single axisymmetric stagnation flow and a solid-body vortex, keeping one unambiguous sign convention for kappa. Then recompute the instantaneous correlation in Fig. 12(g) using both kappa_z as printed in Eq. (2.4) and its negative. If the positive r = 0.54 is obtained with the negative of the printed equation, or if the printed Eq. (4.2) has the wrong sign relative to the re-derivation, the sign inconsistency is confirmed and the reported local agreement must be restated after fixing the sign convention.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equations (2.1) and (2.2) state L[kappa_z] = -D beta/Dt - beta^2 + 2q, so with the unsteady term neglected the gravity-dominated curvature is kappa_gs = (beta^2 - 2q)/g. Equation (4.2), however, prints kappa_gs = -beta^2/g + 2q/g, the exact negative, and Fig. 7 uses this sign convention to place positive curvature in the q > p^2/2 sector. The same inconsistency propagates into the physical discussion: the text states that both upwellings and downwellings yield negative curvature, which follows from (4.2), while (2.5) with its leading minus gives positive curvature for the same stagnation flow. Since the central instantaneous comparison is a Pearson correlation between signed measured and modeled curvature fields, a global sign flip changes the sign of r, not its magnitude. The reported r = +0.54 therefore implies that the implementation followed one sign convention while one of the printed equations follows the opposite convention. Either Eq. (2.1)/(2.5) or Eq. (4.2) is wrong, and the interpretation of dimples/bulges in Figs. 7, 9, and 10 is inverted relative to the stated definition kappa = -nabla^2 eta. This is correctable, but it is load-bearing: the local/instantaneous agreement cannot be evaluated from the manuscript as written.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes an Euler-Laplace framework, Eq. (2.1), that relates the free-surface curvature of a turbulent water surface to the subsurface velocity field and its gradients, and tests it with simultaneous BOS surface topography and PIV velocity measurements in a zero-mean-flow turbulent water tank over a range of Reynolds and Froude numbers. The central claims are: (i) the instantaneous surface curvature can be modeled from the velocity field at z = -2 mm, with a Pearson correlation of 0.54 between measured and modeled curvature; (ii) the r.m.s. curvature scales as kappa'^2 ~ c (beta'^2)^2 / g^2 with c = 7.15; (iii) the correlation between surface and subsurface fields decreases with depth and is restricted to increasingly large spatial scales as the measurement depth increases. The authors further discuss implications for remote sensing of near-surface turbulence and gas transfer.","tokens_in":18079,"tokens_out":9192,"duration_ms":89194,"significance":"If the central claims hold after revision, this is a valuable contribution to free-surface turbulence and remote sensing. The experimental dataset is unusually rich: simultaneous high-resolution BOS and PIV measurements at multiple depths, over a wide range of forcing conditions, provide a rare opportunity to test first-principles models of surface-sub-surface coupling. The derivation of the Euler-Laplace equation is a useful formal step, and the spectral comparisons in Figs. 11-12 are informative. The paper also makes a concrete, falsifiable statistical prediction in the form of the kappa'^2 - beta'^2 scaling law. The main limitations are the moderate correlation level supporting the instantaneous claim, the fitted prefactor in the scaling law, and a sign inconsistency in the printed equations that must be resolved before the signed comparisons can be interpreted.","major_comments":[{"comment":"The printed equations contain a global sign inconsistency for the gravity-dominated steady curvature. Starting from Eq. (2.1) and setting the unsteady and surface-tension terms to zero gives kappa_gs = (beta^2 - 2q)/g, which is also what Eq. (2.5) reduces to in the steady limit. Equation (4.2), however, prints kappa_gs = -beta^2/g + 2q/g, the exact opposite. The sign inconsistency propagates into the physical interpretation: the text following Eq. (2.5) states that vorticity leads to positive curvature (dimples), but Eq. (2.5) gives a negative contribution from omega_z^2/(2g). Since the headline instantaneous comparison is a signed Pearson correlation (Fig. 12g) and Figs. 7, 9, and 10 rely on the sign of kappa to distinguish dimples from bulges, the side of the q = beta^2/2 boundary that is labeled positive is reversed between the two equation sets. The authors must state which sign convention was actually implemented in the analysis, correct one of the equation sets, and re-check the sign labeling in the figures; as printed, the local/instantaneous agreement claim cannot be evaluated.","section":"§2, Eq. (2.5); §4.2, Eq. (4.2)"},{"comment":"The central claim of close quantitative agreement between measured and modeled instantaneous curvature rests on a Pearson correlation of r = 0.54, which leaves roughly 71% of the variance unexplained. The compared fields are also spatially Gaussian-filtered (sigma = 1 mm), temporally smoothed (30 ms), and the velocity is measured at z = -2 mm rather than at the surface, so it is not clear how much of the correlation reflects true physical agreement as opposed to filtering or interpolation effects. Please provide an uncertainty analysis for the reported correlations, such as confidence intervals across independent realizations, correlations against temporally shuffled velocity fields, or a noise-injection test on the PIV data, and temper the wording of 'close quantitative agreement' accordingly.","section":"§4.3-4.4, Figs. 8, 12(g)"},{"comment":"The scaling law kappa'^2 ~ c (beta'^2)^2/g^2 is presented as a quantitative prediction of the framework, but the prefactor c = 7.15 is fitted to the same data shown in Fig. 6, and Eq. (4.1) assumes without direct evidence that the ratios of (Dbeta/Dt)^2, beta^4, q^2, and the cross terms are universal across the range of Froude numbers. As written, Fig. 6 demonstrates the exponent and the order of magnitude, not a parameter-free prediction. Please state explicitly that c is empirical, and if possible derive or bound c from the measured single-point statistics or from a DNS of free-surface turbulence.","section":"§4.2, Eq. (4.1), Fig. 6"},{"comment":"The framework is derived from the inviscid Euler equation, but the velocity field is measured at z = -2 mm, which the authors describe as 'roughly at the edge of the viscous layer.' Viscous stresses at that depth, as well as the small-slope approximation for surface-parallel derivatives, could systematically bias the modeled pressure field and hence the modeled curvature. The present data alone do not directly validate the inviscid approximation at this depth. A quantitative check would be to compare the model error against the depth-dependent correlation in Fig. 13(a) and test whether the residual is consistent with viscous corrections, or to validate the model against a DNS of shear-free free-surface turbulence with comparable Reynolds and Froude numbers.","section":"§4.1, §2"}],"minor_comments":[{"comment":"The caption refers to 'the entire RHS of EQ (1)', but the equation number should be (2.2) or (2.3); please correct the cross-reference.","section":"Fig. 8 caption"},{"comment":"The expansion of the square of Eq. (2.5) is not complete as written: the full square contains terms such as -4 beta^2 q and -2 (Dbeta/Dt)(beta^2 - 2q), which are of the same order as the displayed terms and should either be written explicitly or shown to be negligible.","section":"Eq. (4.1)"},{"comment":"The temporal Gaussian smoothing with standard deviation 30 ms significantly modifies the unsteady term Dbeta/Dt; please state how this filter width compares with the Taylor timescale T_T and test the sensitivity of the correlation in Fig. 12(g) to the smoothing time.","section":"§3.2"},{"comment":"The statement that the boundary q = p^2/2 is 'in agreement with (2.5)' needs to be revisited after the sign inconsistency in the major comments is resolved, because the side of the boundary that corresponds to positive curvature is reversed between Eq. (2.5) and Eq. (4.2).","section":"Fig. 7"},{"comment":"The reference entry for Laxague et al. (2026) contains a duplicated author name; please correct it.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"This is a strong experimental paper that fits JFM's scope well. The sign inconsistency is clearly a load-bearing issue but appears correctable within revision, and the correlation and scaling-law concerns are addressable with additional analysis rather than new experiments. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read of Ruth & Coletti. The genuinely new thing is the Euler-Laplace framework: they write the free-surface curvature as the solution of an elliptic operator gκ - (σ/ρ)∇²κ driven by the material derivative of the surface divergence and the invariants β and q, include unsteady terms that Savelsberg & van de Water neglected, and test it against simultaneous BOS/PIV measurements at z=-2 mm. That is a real step forward, and the dataset is impressive: nine forcing conditions, Re and Fr ranges, spectra, and a careful depth-decorrelation analysis. The statistical scaling κ'^2 ≈ c (β'^2)^2/g^2 with c=7.15 collapses their data well across conditions; the prefactor is fitted, but the universal-ratio argument is reasonable, and the fit is over a decade of forcing, not a single point.\n\nThe soft spot is the sign. Equation (2.5) as printed gives κ_g = -1/g(Dβ/Dt - β² + 2q), so with the unsteady term dropped the gravity-dominated curvature is (β² - 2q)/g. Equation (4.2) prints κ_gs = -β²/g + 2q/g, the exact negative. The paper also defines κ = -∇²η yet labels dimples as κ>0, which is inconsistent with that definition for a depression. Since the central instantaneous comparison is a signed Pearson correlation (r=0.54), a global sign flip changes the sign of r but not its magnitude, so the reader cannot tell from the manuscript which convention the implementation used. This is correctable, but it is load-bearing: the quantitative agreement claim can't be evaluated until it is fixed.\n\nTwo smaller issues. The correlation of 0.54 is called 'close quantitative agreement' and the snapshot in Fig. 9 is described as 'remarkably well'; that language overstates a moderate correlation, though the spectral agreement is more convincing. And the inviscid model is applied at z=-2 mm, 'roughly at the edge of the viscous layer'; viscous stresses there could bias the modeled pressure field. The authors acknowledge the surface-slope smallness but not this viscous question directly; worth addressing with a sensitivity estimate.\n\nBottom line: the framework is plausible, the data are valuable, and the depth-decorrelation results are a useful contribution for anyone working on free-surface turbulence or ocean remote sensing. The sign inconsistency must be resolved before the headline correlation can be interpreted. Send it to review, but require the authors to fix the sign convention and temper the 'excellent agreement' language.","headline":"A valuable first-principles framework and rich dataset linking surface curvature to subsurface turbulence, but the sign convention in the printed equations is internally inconsistent and must be fixed before the headline correlation can be trusted.","tokens_in":18637,"tokens_out":8799,"would_cite":true,"duration_ms":85750,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that the Euler-Laplace equation, applied to the velocity field a few millimetres below a water surface, quantitatively reconstructs the instantaneous surface curvature (Pearson correlation 0.54) and that r.m.s.","keywords":["free-surface turbulence","surface curvature","Euler-Laplace equation","horizontal divergence","background-oriented schlieren","particle image velocimetry","air-water gas transfer","remote sensing"],"falsifier":"Repeat the same PIV-BOS experiment in the weakest-forcing case with the velocity plane moved to roughly 0.5 mm below the surface, inside the viscous layer: if the modelled curvature from that plane correlates with measured curvature as strongly as 0.54 or better, the inviscid assumption at 2 mm depth was not the limiting factor; if the correlation collapses, viscous stresses at that depth are load-bearing.","tokens_in":17563,"feed_emoji":"🌊","tokens_out":5699,"duration_ms":64913,"temperature":0.7,"pith_summary":"This paper tries to show that the shape of a water surface can be computed, instant by instant, from the velocity field a few millimetres below it, using only inviscid fluid dynamics with gravity and surface tension. The central object is the Euler-Laplace equation, which ties surface curvature to the horizontal divergence and the determinant of the subsurface velocity-gradient tensor. Against simultaneous measurements in a turbulence tank, the model reproduces the measured curvature with a Pearson correlation of 0.54 and captures the r.m.s. scaling $\\kappa'^2 = 7.15\\,(\\beta'^2)^2/g^2$ across a range of Reynolds and Froude numbers. The claimed payoff is that surface topography can act as a remote sensor for near-surface turbulence, including the horizontal divergence that governs air-water gas and heat transfer.","feed_headline":"Surface curvature reveals hidden turbulence at 0.54 correlation","feed_subtitle":"A first-principles model ties water-surface shape to velocity millimeters below, opening optical sensing of gas transfer.","key_machinery":"The Euler-Laplace equation (2.1) is the load-bearing object: it converts pressure fluctuations in an inviscid, small-slope free-surface flow into surface curvature, with gravity and surface tension opposing the deformation. Its right-hand side is built from two invariants of the surface-parallel velocity-gradient tensor, the horizontal divergence $\\beta$ and the determinant $q$, so the curvature field is reconstructed by inverting the operator $L = g + (\\sigma/\\rho)\\nabla^2$ on the combination $-D\\beta/Dt - \\beta^2 + 2q$. Re-expressing that combination in terms of vorticity and strain separates dimples (positive curvature) from bulges (negative curvature), and the gravity-dominated limit yields the scaling $\\kappa'^2 \\propto (\\beta'^2)^2/g^2$ used throughout.","core_discovery":"The authors establish that, in unbroken free-surface turbulence with small slopes, the instantaneous surface curvature $\\kappa = -\\nabla^2\\eta$ is governed by the linear operator $L[\\kappa] = g\\kappa - (\\sigma/\\rho)\\nabla^2\\kappa$ acting on subsurface quantities: $L[\\kappa_z] = -D\\beta/Dt - \\beta^2 + 2q$, with $\\beta$ the horizontal divergence and $q$ the determinant of the surface-parallel velocity-gradient tensor. Measured and modelled curvature fields agree in snapshots, time series, and space-time spectra; the modelled curvature correlates with measurements at 0.54 when the velocity is taken at $z=-2$ mm. Statistically, the r.m.s. curvature follows $\\kappa'^2 = c\\,(\\beta'^2)^2/g^2$ with $c=7.15$, as predicted when gravity dominates surface tension. The authors further show that the correlation decays with depth, dropping by nearly an order of magnitude once the depth approaches the Taylor microscale, and that only flow structures larger than the measurement depth imprint the surface, bounding the resolution of any inversion from surface shape to subsurface flow.","pith_inferences":["A natural extension the authors leave implicit is that curvature variance measured from airborne or shipborne stereo or polarimetric imagery could yield estimates of gas transfer velocity without in-water instrumentation, provided the scaling survives mean shear and swell.","The observed depth-decorrelation bound suggests that multi-scale surface measurements could be used to estimate the depth of turbulent structures beneath the surface, acting as a form of optical tomography of the upper water column.","The framework could be stress-tested in direct numerical simulations by applying the same Euler-Laplace inversion to velocity fields at multiple depths; if simulated correlations exceed 0.54, the laboratory ceiling is set by measurement noise rather than by physics.","The curvature-divergence relation could serve as a physics-based regularizer for neural-network reconstructions of near-surface flow from surface images, reducing the data needed for such data-driven inversions."],"forward_implications":["If the central claim holds, optical measurements of surface curvature can serve as a proxy for the near-surface horizontal divergence, which is the quantity that models of interfacial gas transfer depend on.","Instantaneous surface topography can be reconstructed from velocity measurements just below the surface, at least down to scales set by the Taylor microscale and the measurement depth.","The r.m.s. curvature scaling provides a calibration-free relation between curvature variance and divergence variance in gravity-dominated, unbroken free-surface turbulence.","Because the surface-subsurface correlation decays with depth, any inversion from surface shape can only recover flow structures larger than the depth of interest; smaller scales are unrecoverable.","The framework gives a first-principles basis for interpreting space-time spectra of surface curvature in field measurements, including the gravity-capillary wave band."],"supporting_citations":[{"why":"Supplies the earlier experimental attempt to relate surface slope to subsurface gradients, whose low correlation this paper improves upon by an order of magnitude.","marker":"Savelsberg & van de Water (2009)"},{"why":"Provides DNS evidence of deformable free-surface turbulence and supports the small-slope, horizontal-derivative approximation used in the theory.","marker":"Guo & Shen (2010)"},{"why":"Establishes the link between surface features and the integrated divergence of the near-surface velocity, which the present depth-decorrelation results extend.","marker":"Babiker et al. (2026)"},{"why":"Documents the turbulence facility, flow quality, and near-surface statistics on which the present experiments rely.","marker":"Ruth & Coletti (2024)"},{"why":"Provides the background-oriented schlieren method used to reconstruct the surface topography with the required resolution.","marker":"Moisy et al. (2009)"},{"why":"Supplies the invariant decomposition of the velocity-gradient tensor into $\\beta$ and $q$ used to express the curvature forcing.","marker":"Perry & Chong (1987)"},{"why":"Supports the near-surface statistics of divergence and vorticity, including the weak forcing dependence of the divergence kurtosis used in the scaling argument.","marker":"Qi et al. (2025a)"},{"why":"Supplies free-surface turbulence properties and the near-surface velocity-gradient statistics used to argue universality of small-scale ratios.","marker":"Wu et al. (2026)"},{"why":"Defines the regime map that places the experiments in the weak and gravity-dominated turbulence regimes.","marker":"Brocchini & Peregrine (2001)"},{"why":"Provides the teapot-shaped joint PDF of $p$ and $q$ in homogeneous turbulence, used to interpret the curvature conditioned on the invariants.","marker":"Cardesa et al. (2013)"}],"fun_headline_variants":["Surface curvature exposes subsurface turbulence at 0.54 correlation","Turbulence's fingerprint: surface curvature at 0.54 correlation","Curvature of water surface betrays subsurface turbulence at 0.54","Optical sensing of turbulence via surface curvature: 0.54 correlation","How surface shape reads turbulence: 0.54 correlation at 2 mm depth"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The model assumes the inviscid Euler equation with a linearized, small-slope free-surface condition is valid at the measurement plane 2 mm below the surface, where viscous stresses may still matter; if viscosity or surface-normal velocity components contribute significantly there, the predicted curvature would deviate systematically from measurements.","fun_headline_variants_meta":{"raw":{"variants":["Surface curvature exposes subsurface turbulence at 0.54 correlation","Turbulence's fingerprint: surface curvature at 0.54 correlation","Curvature of water surface betrays subsurface turbulence at 0.54","Optical sensing of turbulence via surface curvature: 0.54 correlation","How surface shape reads turbulence: 0.54 correlation at 2 mm depth"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001291,"raw_usage":{"total_tokens":5300,"prompt_tokens":1005,"completion_tokens":4295,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":4199}},"tokens_in":621,"tokens_out":4295,"duration_ms":34192,"temperature":1.0,"reasoning_tokens":4199,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:53:29.602748+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the same PIV-BOS experiment in the weakest-forcing case with the velocity plane moved to roughly 0.5 mm below the surface, inside the viscous layer: if the modelled curvature from that plane correlates with measured curvature as strongly as 0.54 or better, the inviscid assumption at 2 mm depth was not the limiting factor; if the correlation collapses, viscous stresses at that depth are load-bearing.","supporting_citations":[{"cited_title":"Journal of Fluid Mechanics 619 , 95--125","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier experimental attempt to relate surface slope to subsurface gradients, whose low correlation this paper improves upon by an order of magnitude."},{"cited_title":"Journal of Fluid Mechanics 658 , 33--62","cited_arxiv_id":null,"evidence_quote":"Provides DNS evidence of deformable free-surface turbulence and supports the small-slope, horizontal-derivative approximation used in the theory."},{"cited_title":", Aarnes, Jørgen R","cited_arxiv_id":null,"evidence_quote":"Establishes the link between surface features and the integrated divergence of the near-surface velocity, which the present depth-decorrelation results extend."},{"cited_title":"& Coletti, Filippo 2024 Structure and energy transfer in homogeneous turbulence below a free surface","cited_arxiv_id":null,"evidence_quote":"Documents the turbulence facility, flow quality, and near-surface statistics on which the present experiments rely."},{"cited_title":"Experiments in Fluids 46 (6), 1021--1036","cited_arxiv_id":null,"evidence_quote":"Provides the background-oriented schlieren method used to reconstruct the surface topography with the required resolution."},{"cited_title":"Annual Review of Fluid Mechanics 19 (1), 125--155","cited_arxiv_id":null,"evidence_quote":"Supplies the invariant decomposition of the velocity-gradient tensor into $\\beta$ and $q$ used to express the curvature forcing."},{"cited_title":"& Peregrine, D","cited_arxiv_id":null,"evidence_quote":"Defines the regime map that places the experiments in the weak and gravity-dominated turbulence regimes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the teapot-shaped joint PDF of $p$ and $q$ in homogeneous turbulence, used to interpret the curvature conditioned on the invariants."}],"review_version":1}