{"id":"84ae6d4f-e02a-4180-bc4c-260716defaea","arxiv_id":"2506.05436","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Near-wall velocity increments in a turbulent channel flow follow two distinct logarithmic scaling regimes, and the viscous layer maintains non-Gaussian fluctuations at all scales.","lead":"This paper analyzes how velocity fluctuations change with scale at different distances from the walls in a simulated turbulent channel flow. It reports a new two-part scaling behavior near the walls and shows that the viscous layer has strongly non-Gaussian velocity fluctuations at every scale.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-regime near-wall model and the 0.41 viscous plateau are inferred from a single 2D slice with no error bars; at scales near δ the effective sample size is O(25), so Eqs. (24)-(25) may reflect sampling noise rather than flow physics.","rationale":"The Reader correctly flags the ad hoc scale d as an unproven normalizing choice. I agree that d is not derived, but I regard the more immediate threat as statistical: the inference uses one z=3π/2 instantaneous slice, and the scales where the two-regime behavior is claimed have very few effective samples. For high-order moments, this can easily produce spurious plateaus and kinks. This does not change the overall verdict—the paper remains a useful characterization, but the central model should be treated as conditional until verified with time/ensemble averaging. The proposed check is directly feasible because the JHU database provides multiple snapshots, so the paper's central claims can be settled without new simulations.","tokens_in":10865,"tokens_out":6634,"duration_ms":90163,"concrete_test":"Recompute S2, S4, and flatness for the same y+ levels using all available JHU Channel 5200 snapshots (or at least 50 independent time fields), with block-bootstrap error bars over x-blocks; then fit Eq. (24) and Eq. (25) with d fixed and free, and test whether the kink at d and the viscous plateau at log F=0.41 survive within the bootstrap confidence intervals and beat the single-log and K41 alternatives by a likelihood ratio.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central new claim—the kink at l=d and the non-Gaussian plateau log F=0.41 in the viscous layer—is built from structure functions computed on one instantaneous 2D transect (Section II), with spatial averaging only and no time or ensemble averaging. Along the streamwise direction, 10240 pixels cover 8π, so at l≈δ only about 25 statistically independent increments exist; because the fourth-order moment is dominated by rare events, S4 and F at the scales that define the second inertial subdomain are dominated by sampling uncertainty. The gray individual-y curves in Fig. 4 visibly scatter, yet no error bars are reported, and the apparent kink and plateau are not compared against the null hypothesis of a single log-law region or a K41 region with noise. This is more load-bearing than the choice of d: even if d is universal, the two-regime model would still not be supportable if the inferred functions are statistically indistinguishable from a single-regime fit on the available sample.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript analyzes second- and fourth-order structure functions and flatness of streamwise velocity in a turbulent channel flow, using a single spanwise transect of the JHU Channel5200 DNS database. It reports standard Kolmogorov/log-normal behavior in the outer layer and, in the logarithmic region, a structure-function behavior consistent with the Davidson et al. logarithmic law. The central new claim concerns the viscous and buffer regions: the paper proposes that the inertial range is split at a scale d into two shear-dominated subdomains, with S_p(l) = (E_p + D_p log(l/d))^{p/2} for eta < l < d and S_p(l) = (K_p + L_p log(l/d))^{p/4} for d < l < delta (Eqs. 24-25). It further claims that the viscous-layer flatness does not return to the Gaussian value but plateaus at log F = 0.41, indicating non-Gaussian fluctuations at all scales. The paper concludes that near-wall turbulence does not display Kolmogorov scaling at any scale in the viscous and buffer layers.","tokens_in":11228,"tokens_out":6635,"duration_ms":83274,"significance":"If the central claims are correct, the paper would provide a useful quantitative description of near-wall structure functions and would challenge the expectation of a K41 inertial subrange close to the wall. The manuscript uses a public DNS dataset and successfully recovers known results in the outer and logarithmic regions, which serves as a useful sanity check. The proposed functional forms in Eqs. (24)-(25) are explicit and could in principle be tested against independent simulations. However, the new claims rest almost entirely on a single two-dimensional snapshot, on constants fitted to the same data used for validation, and on a transition scale d that is chosen rather than measured; the current evidence is therefore suggestive rather than conclusive.","major_comments":[{"comment":"The analysis is based on one instantaneous two-dimensional transect, and the structure functions are estimated by spatial averages only. At scales approaching the channel half-width delta, the streamwise domain of length 8pi contains only about 25 statistically independent increments, and the fourth-order moment is dominated by rare events; nevertheless, no error bars, confidence intervals, or bootstrap estimates are reported. The gray per-y curves in Fig. 4 visibly scatter, so the kink at l=d and the claimed viscous plateau at log F=0.41 cannot be distinguished from sampling fluctuations or from a single-regime model with noise without a statistical test. Please provide uncertainty estimates for S2, S4, and F, and compare the two-regime model against null hypotheses such as a single logarithmic region or K41 scaling with noise.","section":"Section II; Figs. 3-4"},{"comment":"All constants E_p, D_p, K_p, and L_p in Table I are fitted to the same DNS curves that are then displayed as the 'theoretical' model curves in Figs. 3-4. The agreement shown is therefore by construction and does not independently validate the model. The manuscript should specify the fitting procedure, report goodness-of-fit statistics, and include an out-of-sample test, for example by fitting on one portion of the streamwise domain and validating on another portion, or by using a different spanwise position or a different time snapshot.","section":"Table I; Eqs. (24)-(25); Figs. 3-4"},{"comment":"The transition scale d is introduced as three times the thickness of the viscous and buffer regions, following Ref. [7], but no measurement, equation, or fitting procedure establishes d as the scale at which the scaling actually changes. Since the abscissa is normalized by d and d controls the location of the kink in Eqs. (24)-(25), the claimed two-subdomain structure is not independent of the chosen normalization. Please estimate d from the data, for example by a breakpoint fit for each wall distance, report its uncertainty, and show that the two-regime model is statistically preferred over a single-regime description.","section":"Section III; definition of d"},{"comment":"The viscous plateau is reported inconsistently: the abstract states 'F(l)=0.41,' while Section IV and the caption of Fig. 4 state 'log F(l)=0.41.' With the flatness defined in Eq. (2) so that a Gaussian distribution has flatness 1, these two statements are very different (F about 1.51 versus 0.41), and the heavy-tail interpretation depends on which value is correct. Please state the plateau value consistently and give its uncertainty.","section":"Abstract; Section IV; Fig. 4"}],"minor_comments":[{"comment":"In the last row of Table I, the label in the second column reads 'L2,' but it should be 'L4' to match the K4 row above it.","section":"Table I"},{"comment":"The caption describes the dissipative-domain behavior as an 'exponential decrease l^p'; since l^p is a power law, this wording is misleading and should be corrected.","section":"Figure 3 caption"},{"comment":"The relation between the coordinate y and the wall distance \\tilde y is not defined explicitly; please state that \\tilde y is the distance to the nearest wall and clarify that the spatial average is taken at fixed \\tilde y.","section":"Eq. (3) and Section II"},{"comment":"The flatness definition in Eq. (2) includes the factor 3, so the Gaussian value is 1 rather than 3; this convention is correct but should be stated explicitly in the text and in the caption of Fig. 4 to avoid confusion.","section":"Eq. (2)"}],"recommendation":"major_revision","confidential_remarks":"The paper's new near-wall claims are interesting but currently rest on statistically fragile inference from one 2D slice, with model constants fitted to the same data used for validation and a transition scale chosen rather than independently determined. I would not recommend acceptance without independent validation or at least strong out-of-sample tests. The inconsistent statement of the plateau value in the abstract is a further source of confusion."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take: this paper is a useful first map of the second- and fourth-order structure functions and flatness in the buffer and viscous layers of a turbulent channel flow, but the central claim — a two-subdomain log model with a transition at d — is not yet supported by the evidence. I'd send it to review, but only with a request for a serious statistical upgrade and a fix to the model equations.\n\nWhat's new and good: the authors extend the structure-function analysis into the buffer and viscous layers, where prior work (Davidson, de Silva, et al.) stopped at the logarithmic layer. The outer-layer and log-layer results reproduce known scalings, which is a healthy sanity check. The observation of a viscous-layer flatness plateau at log F = 0.41 is genuinely interesting, if it holds up. The paper is clearly written and the public JHU DNS data makes the analysis reproducible.\n\nWhere it's soft: the statistical basis is thin. Structure functions are computed from a single instantaneous 2D transect, with spatial averaging only and no time or ensemble averaging. At scales approaching δ, that leaves roughly 25 independent increments; S4 and flatness are then dominated by a few large events. The gray per-y curves show visible scatter, yet no error bars or confidence intervals are reported. The transition scale d is chosen ad hoc — three times the combined viscous/buffer thickness — and the constants in Table I are fitted to the same curves used for validation, so the agreement is partly a fit.\n\nMore troubling, the model itself has inconsistencies. Table I lists a row 'L2' where it should be 'L4' (typo), and more importantly the flatness expressions (28) and (29) as written do not give the claimed viscous plateau: plugging the tabulated constants into (29) yields a large-scale limit around 4.5, not 0.41. The authors admit Eq. (28) does not match the viscous data, so the central flatness model is unvalidated. This is a fixable problem, but it undercuts the quantitative claim of a two-regime law.\n\nProportionately: the outer/log-layer parts are fine confirmatory work. The near-wall part is a promising lead, not a finished law. The 'first time' novelty claim also deserves a broader literature check, though it passes the quick look I gave it.\n\nWho this is for: turbulence phenomenology people and wall-model developers who want a concrete target to hit. A serious referee should engage with it, but the revision needs time averaging or at least honest error bars, a derivation or independent test of d, and a corrected, consistent set of model equations.","headline":"A useful first map of near-wall structure functions, but the two-regime model rests on thin statistics and inconsistent formulas.","tokens_in":11739,"tokens_out":6218,"would_cite":false,"duration_ms":69674,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Near a wall, turbulence skips Kolmogorov scaling and follows two log-law regimes.","keywords":["turbulent channel flow","structure functions","flatness","near-wall turbulence","logarithmic scaling","shear-dominated turbulence","intermittency","Kolmogorov scaling"],"falsifier":"Recompute the second and fourth order structure functions from the same DNS while varying the normalization scale $d$ over a range around three times the combined viscous-plus-buffer thickness: if the curves no longer collapse or the crossover moves away from $l=d$, the two-regime model is an artifact of the normalization. The predicted energy-density change, from $l^{-1}$ below $d$ to $l^{-1}/\\sqrt{\\log l}$ above $d$, also gives a spectral signature that can be checked independently from the streamwise spectrum.","tokens_in":10706,"feed_emoji":"🌊","tokens_out":9916,"duration_ms":110198,"temperature":0.7,"pith_summary":"This paper tries to establish that the streamwise velocity in a turbulent channel flow behaves differently in the near-wall layers than the classical picture assumes. Using direct numerical simulation data at friction Reynolds number 5200, it studies structure functions and flatness across the viscous, buffer, logarithmic, and outer layers. The paper claims that in the viscous and buffer layers the inertial range splits at a scale $d$ equal to three times the combined thickness of those layers into two shear-dominated logarithmic regimes: $S_p(l)=(E_p+D_p\\log(l/d))^{p/2}$ below $d$ and $S_p(l)=(K_p+L_p\\log(l/d))^{p/4}$ above $d$. In the viscous layer the flatness plateaus at $\\log F = 0.41$ instead of returning to the Gaussian value 0, so velocity increments are non-Gaussian at every scale. If these scalings hold, near-wall turbulence is shear-controlled at all inertial scales and no single logarithmic model describes the structure functions.","feed_headline":"Two log-law regimes replace Kolmogorov scaling near the wall","feed_subtitle":"Viscous-layer velocity jumps stay non-Gaussian at every scale, even the largest.","key_machinery":"The organizing object is the flatness $F(l)=S_4(l)/(3S_2(l)^2)$ built from the second and fourth order structure functions of the streamwise velocity at fixed wall distance. The load-bearing device is a piecewise logarithmic ansatz for the structure functions in the near-wall layers: below $d$, $S_p(l)=(E_p+D_p\\log(l/d))^{p/2}$; above $d$ up to $\\delta$, $S_p(l)=(K_p+L_p\\log(l/d))^{p/4}$. These forms correspond to an energy density scaling as $l^{-1}$ below $d$ and as $l^{-1}/\\sqrt{\\log l}$ above $d$. The crossover scale $d$ carries the argument: the same $d$ must collapse curves for all wall distances in the viscous and buffer regions and mark where the fourth-order structure function becomes linear in $\\log(l/d)$.","core_discovery":"The central claim is an empirical scaling law for streamwise velocity increments in the viscous and buffer regions of a turbulent channel flow. For wall distances $\\tilde{y}$ in these layers, the second and fourth order structure functions obey two consecutive logarithmic forms separated by one scale $d$: $S_p(l)=(E_p+D_p\\log(l/d))^{p/2}$ for $\\eta