{"id":"dc3a50b7-3491-4050-9b69-a735fc05e447","arxiv_id":"2505.06707","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Under flow reversal a dune inverts in about twice the time it took to form, and a barchan's central slice behaves roughly like a 2D dune.","lead":"Experiments and grain-scale simulations show that a dune takes about twice as long to fully invert after the flow reverses as it took to form from an initial pile. The results also suggest that the central slice of a 3D barchan dune behaves like a 2D dune, which would justify simpler 2D models for dune fields.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"2D/3D timescale equivalence is not established by the paper's own numbers: 2D formation/reversal are 5tc/10tc while 3D barchan times are 1–1.5tc/2–2.5tc, so the 2x rule only holds within each dataset.","rationale":"The strongest claim requires two things: (i) reversal takes twice the formation time within a given system, and (ii) the central slice of a barchan equilibrates on the same turnover-count timescale as a 2D dune. The paper's own dimensionless times violate (ii) by a factor of 3–4: 2D formation/reversal are ≈5tc/≈10tc while the 3D barchan horn metrics give ≈1–1.5tc/≈2–2.5tc. The conclusion that the central slice behaves as a 2D dune is reached by comparing ratios rather than absolute times and by using different observables for the 2D and 3D systems. I agree with the reader that the tc normalization is questionable, but the present concern stands even if tc is valid, because the reported normalized times already differ. However, the paper has real supporting features: open data and scripts, a CFD-DEM setup validated in prior work, and a qualitative morphological similarity between the 2D experiments and the simulated central slice. A same-metric, same-tc reanalysis could resolve the issue, so the conditional verdict remains appropriate rather than a rejection.","tokens_in":18339,"tokens_out":11490,"duration_ms":114123,"concrete_test":"Using the released CFD-DEM data and 2D flume profiles (Mendeley doi:10.17632/fw3bcrxknf.1), recompute central-slice formation and reversal times with the same metric used for the 2D dunes—plateau of Z/L and LStoss/L with a fixed threshold on the smoothed slope—and with a single common tc (pre-reversal developed-dune Leq and u*) for both phases. If the central-slice plateaus appear at ≈5tc and ≈10tc, the equivalence claim is supported; if they appear at ≈1–1.5tc and ≈2–2.5tc, the factor-of-two rule is an artifact of comparing different observables and the 2D-slice predictive conclusion should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.1 reports 2D dune formation at t/tc≈5 and Section 4.2 reports 2D reversal at t/tc≈10. For the 3D barchan, the same sections report horn-length plateaus at t/tc≈1–1.5 during formation (Fig. 5a) and t/tc≈2–2.5 during reversal (Fig. 5b). The 'twice' rule is therefore a ratio inside each dataset—10/5=2 and ≈2.5/1.25≈2—not equality of the characteristic times themselves. After normalizing by tc, the 3D times are 3–4 times shorter than the 2D times. The paper moves from 'same order of magnitude' to 'equivalent' without quantitative support, and it defines the 3D time from horn growth while the 2D time is defined from Z/L and LStoss/L plateaus; these observables need not saturate together. Thus the headline application—that 2D slice simulations predict realistic 3D reversal timescales—does not follow from the reported data, regardless of whether tc remains the correct normalization during reversal. This is a correctness risk in the applied claim, not an internal inconsistency: the data and code are open, so the check is feasible.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports experiments on 2D subaqueous dunes in a circular flume and CFD-DEM simulations of 3D barchans, both evolving from an initial heap to a steady dune and then through a 180° flow reversal. The central quantitative claims are: (i) the characteristic formation time of 2D dunes is about 5tc, with tc taken from Alvarez and Franklin (2017); (ii) the 3D barchan central slice attains a developed state on a comparable timescale (~1–1.5tc for horn growth); (iii) flow reversal takes about twice the formation time in both the 2D experiments (~10tc) and the 3D simulations (~2–2.5tc); and (iv) a significant fraction of grains remains static during both phases. On this basis the paper concludes that the central slice of a barchan behaves roughly as a 2D dune and that 2D slice simulations can predict reversal timescales in 3D geophysical settings.","tokens_in":18646,"tokens_out":4466,"duration_ms":40825,"significance":"If the quantitative claims held, the paper would provide a strong practical justification for using 2D slice models to estimate barchan reversal times, and the grain-scale tracking of static and mobile fractions is a valuable contribution. The manuscript is commendable for publishing the experimental images, processing scripts, and CFD-DEM setup in an open repository, and for grounding the reversal-time observation in direct measurements in both experiments and simulations. However, the central quantitative equivalence between the 2D and 3D timescales is not established by the reported numbers, so the applied claim is currently a proof-of-concept rather than a demonstrated result.","major_comments":[{"comment":"The reported numbers do not support the statement that the central slice of a barchan behaves roughly as a 2D dune in terms of timescales. In the 2D experiments, formation and reversal are read from Z/L and LStoss/L plateaus as t/tc ≈ 5 and ≈ 10. In the 3D barchan simulations, the equivalent times are read from horn-length plateaus as t/tc ≈ 1–1.5 and ≈ 2–2.5. The factor-of-two ratio holds within each dataset (10/5 = 2 and ≈2.5/1.25 ≈ 2), but the absolute times differ by a factor of approximately 3–4 after normalization by the same tc. Unless the authors demonstrate that the different observables (Z/L and LStoss/L versus horn length) are interchangeable proxies for the same morphodynamic state, and that tc remains the correct normalization during reversal, the conclusion that 2D slices predict realistic 3D timescales does not follow from the data.","section":"Sections 4.1–4.3, Figures 3 and 5"},{"comment":"The central timescale tc is defined by dropping all constants and using Leq measured from the developed dune, and both formation and reversal times are normalized by this same tc. The paper does not verify that tc, computed from the pre-reversal equilibrium dune, governs the turnover during the reversal phase, when the dune shape changes substantially and the relevant length scale might differ. Since the '2x' rule is a statement about ratios of times that share this tc, an O(1) error in tc during the reversed phase could absorb or create the reported factor of two. I ask the authors to test this explicitly, for example by computing tc from the instantaneous dune length or celerity during the inversion and replotting Figures 3b and 5b with the time-dependent normalization.","section":"Section 4.1, Eq. (5)"},{"comment":"The simulation slices the 3D barchan to a central width of 2 mm ≈ 10 grain diameters 'to avoid excessive fluctuations', but no convergence test or sensitivity analysis is provided to show that this width is representative of a 2D dune rather than an artifact of the small number of grains. The claim that the central slice behaves as a 2D dune is also based on comparing a horn-length plateau in 3D with Z/L and LStoss/L plateaus in 2D; these observables need not saturate simultaneously, and the paper itself notes that Z/L comparisons are inconclusive. A direct geometric comparison—for example, superimposing the normalized shape of the central slice and the experimental 2D profile at matching t/tc—would provide the missing evidence.","section":"Section 4.1, paragraph on central slice width and Figure S11"}],"minor_comments":[{"comment":"The statement 'the total time for achieving an inverted dune is t/tc ≈ 10' should state the exact criterion used to define 'achieving an inverted dune' (for instance, when the new avalanche face reaches the trailing edge), and should specify whether the same criterion was applied in the 3D simulations.","section":"Section 4.2"},{"comment":"Please define the dune celerity C explicitly before Equation (5), and clarify that the proportionality C ~ q/Z ~ q/Ldrag is an order-of-magnitude estimate with all prefactors dropped.","section":"Section 4.1, Eq. (5) and surrounding text"},{"comment":"For the reversal cases g–l, please state whether the listed Z and L values are measured at the onset of the reversed flow (t = 0) or at the end of the initial development phase, as this affects the normalization by Leq in Equation (5).","section":"Table 1"},{"comment":"The abstract and conclusions state that the characteristic time for 2D dune development 'scales with' that for 3D barchans, while the body reports specific ratios; please make the quantitative claim consistent throughout the manuscript.","section":"Abstract and Conclusions"},{"comment":"There is a typo in the Conclusions ('similations') and the text alternates between 'eolian' and 'aeolian'; please standardize the spelling.","section":"Throughout"},{"comment":"In Figures 5a and 5b, the horn length is normalized by Ldrag while time is normalized by tc; please state explicitly that Ldrag is used only for the length normalization, to avoid any impression that the time axis also uses Ldrag.","section":"Figure 5"}],"recommendation":"major_revision","confidential_remarks":"The quantitative discrepancy between the 2D and 3D timescales is the main obstacle to the paper's central claim, but the manuscript's own data and open code make a direct re-analysis feasible in revision, so I do not recommend rejection. I would also flag that the key timescale tc originates from the same group's earlier work, which makes the normalization issue delicate; the authors should address it head-on rather than treating tc as self-evidently valid during reversal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper has one genuinely new piece of evidence: a controlled, step-by-step look at how a dune inverts under flow reversal, in both a 2D flume and grain-resolving CFD-DEM, with data and scripts open. The morphodynamic picture--grains on the lee side climbing back up, a new lee face forming on the former stoss, a static interior--is well documented and worth having. The paper also shows a factor-of-two ratio between formation and reversal time that is internally consistent within each geometry.\n\nThe soft spot is exactly what the stress-test note flags. The paper's headline claim that the central slice of a barchan behaves as a 2D dune, and that 2D simulations predict realistic 3D reversal timescales, is not supported by the paper's own numbers. The 2D formation time is read as 5tc; the 3D barchan formation time, from horn-length plateau, is 1-1.5tc. For reversal, 10tc versus 2-2.5tc. That's a factor 3-4 difference, not equivalence. The 'twice' ratio survives only inside each dataset, because both times are scaled by the same tc. Calling this 'scales with' and then 'equivalence' in the abstract and conclusions overstates what is shown. I also think the normalization deserves scrutiny: tc is taken from Alvarez & Franklin 2017 with all constants dropped, and measured from the developed dune before reversal. Without error bars or an independent check of whether tc remains the right turnover timescale during inversion, the factor-of-two could be partly an artifact of the normalization.\n\nThat said, the paper is not sloppy in the ways that matter most. The experiments and simulations are described carefully, the open data and code are a real plus, and the authors explicitly note the limitations of extrapolating to the aeolian case. The comparison is also complicated by the fact that 2D and 3D times are defined from different observables (Z/L and LStoss/L plateaus versus horn length), and the authors acknowledge that. But that acknowledgment makes the strength of the conclusion harder to defend.\n\nBottom line: this is a solid, honest contribution to dune morphodynamics, but the central quantitative claim needs either better support or a serious downgrade. I would not cite the 2D/3D equivalence in its current form, though I might cite the reversal observations. Send it to peer review--a good referee can sort out the timescale question--but expect heavy revision on the 2D/3D comparison.","headline":"New dune-reversal data and open grain-scale simulations, but the 2D/3D timescale equivalence claim doesn't survive contact with the paper's own numbers.","tokens_in":19182,"tokens_out":2679,"would_cite":false,"duration_ms":24512,"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":"This paper argues that a dune under reversed flow takes about twice as long to become an inverted, steady-state dune as it took to form from an initial heap, and that the same turnover timescale used for growth describes both processes.","keywords":["dune reversal","barchan dunes","subaqueous dunes","flow reversals","CFD-DEM","turnover time","bedload transport","dune morphodynamics"],"falsifier":"Re-run the reversal experiment starting from a developed dune that was not formed from a heap (for example, a dune equilibrated on a flat bed) and measure the inversion time divided by $t_c$; if the ratio departs from roughly 2, the factor-of-two rule is tied to the heap initial condition and not a universal reversal timescale.","tokens_in":18115,"feed_emoji":"🏜️","tokens_out":5152,"duration_ms":49262,"temperature":0.7,"pith_summary":"This paper argues that a dune under reversed flow takes about twice as long to become an inverted, steady-state dune as it took to form from an initial heap, and that the same turnover timescale used for growth describes both processes. The authors reach this conclusion by combining experiments on 2D dunes in a circular flume with grain-resolving CFD-DEM simulations of 3D barchans, tracking the barchan's central slice as a proxy for a 2D dune. They find that the central slice behaves roughly like a 2D dune, so cheaper 2D-slice computations could predict reversal timescales at geophysical scales. If true, the result gives a simple predictive rule: formation takes about five turnover times, inversion about ten.","feed_headline":"Dunes take twice as long to flip as to form","feed_subtitle":"2D simulations of a dune's middle slice can predict how fast crescent dunes invert when flow reverses.","key_machinery":"The load-bearing object is the turnover timescale $$t_c = \\frac{L_{eq}(\\rho_p/\\rho_f)(\\rho_p/\\rho_f - 1) g d}{(u_*^2 - u_{th}^2)^{3/2}} \\sim \\frac{L_{eq}}{C},$$ where $L_{eq}$ is the developed dune length, $C$ its celerity, and the transport rate follows Meyer-Peter–Muller. This supplies the common clock that makes formation and inversion comparable; the paper reads the characteristic times off plateaus in $Z/L$ and $L_{stoss}/L$ for 2D dunes and off horn length for 3D barchans. The second mechanism is the central-slice reduction: transverse dispersion on the stoss slope is balanced by inward avalanching at the lee, so the middle slice retains the morphodynamic memory of the barchan and can stand in for the whole dune.","core_discovery":"On the paper's own terms, the central discovery is a factor of two in characteristic timescales: dunes formed from a heap reach steady state in about $t/t_c \\approx 5$, while reversing an already developed dune into its inverted form takes about $t/t_c \\approx 10$, and the same $t_c$ computed from the developed dune's length and transport rate works for both. The reversal proceeds by grains on the lee side climbing back up while the internal part and toe remain static, forming a new lee face of varying slope on the former stoss side. In the 3D simulations, the barchan's central slice follows the same morphological evolution as the 2D dune, and the horns shorten, vanish around $t/t_c \\approx 1$, and regrow. The paper concludes that the central slice of a barchan behaves roughly as a 2D dune, that the Meyer-Peter–Muller-type scaling behind $t_c$ remains valid during reversals, and that about one fifth of the grains in the central slice stay static through the whole process.","pith_inferences":["If the factor-of-two rule survives changing the initial condition (for example, a dune developed on a flat bed rather than from a heap), it could serve as a field diagnostic: the ratio of reversal to formation timescales may fingerprint the history of wind or current reversals.","The roughly 20% static-grain fraction suggests the dune interior is not remobilized by reversal; a testable consequence is that grain-age distributions in an inverted dune should show a core of old grains, which could be checked with colored grains in experiments.","The central-slice equivalence was demonstrated for subaqueous rolling and sliding transport; extending the same comparison to aeolian saltation, where grain inertia and transport rates are much larger, would test whether the factor of two is universal or specific to bedload.","Because the authors note that reversal timescales could differ for dunes formed from a flat bed, a direct numerical test of that alternative initial condition would clarify whether the reported ratio is intrinsic to the reversal process or inherited from the heap start."],"forward_implications":["Barchan reversal timescales in geophysical settings can be estimated from 2D slice computations rather than full grain-resolving 3D simulations.","The same $t_c$ scaling law used for dune growth remains valid during flow reversals, giving one clock for both formation and inversion.","A dune under reversing flow loses roughly 10–15% more grains than during its formation from a heap, so inversion is not a symmetric replay of growth.","Continuum slice models that exchange mass between vertical slices are adequate for simulating barchan fields, including subaqueous cases.","The measured inversion time of about $10\\,t_c$ can be used directly to estimate how long a reversing wind or current must act to fully reshape a dune."],"supporting_citations":[{"why":"Supplies the turnover timescale $t_c$ for subaqueous barchan growth and the horn-growth method used to define characteristic times.","marker":"Alvarez and Franklin (2017)"},{"why":"Provides the mean residence time of about 10 turnover times in the barchan central slice and the dispersion-concentration mechanism that justifies treating the central slice as morphodynamically representative.","marker":"Zhang et al. (2014)"},{"why":"Validates the CFD-DEM/LES numerical setup used for the 3D barchan simulations.","marker":"Lima et al. (2022)"},{"why":"Establishes the 2D circular-flume dune experiments and the wake-repulsion behavior on which the present experimental setup is based.","marker":"Bacik et al. (2020)"},{"why":"Supplies the inertial drag length $L_{drag}$ and the relevant length-scale comparison for barchan dunes across environments.","marker":"Hersen et al. (2002)"},{"why":"Provides field evidence of reversing dunes and the speed-up mechanism that motivates studying flow reversal.","marker":"Gao et al. (2021)"},{"why":"Gives the bedload transport correlation used inside $t_c$, with the same $∼ \\theta^{3/2}$ form as Bagnold's law.","marker":"Meyer-Peter & Müller (1948)"}],"fun_headline_variants":["Dune flip time is twice the build time","Reversing dunes: 2x slower than forming","Barchan reversal takes two formations","Sand dune inversion: double the creation","Dune turnaround: double the time to form"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's central ratio rests on using the same turnover time $t_c$, fixed by the dune's pre-reversal size and flow, to measure both formation and reversal; if the relevant length or transport rate changes during inversion, the factor of two could be an artifact of that normalization.","fun_headline_variants_meta":{"raw":{"variants":["Dune flip time is twice the build time","Reversing dunes: 2x slower than forming","Barchan reversal takes two formations","Sand dune inversion: double the creation","Dune turnaround: double the time to form"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00045,"raw_usage":{"total_tokens":2334,"prompt_tokens":1076,"completion_tokens":1258,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":692,"completion_tokens_details":{"reasoning_tokens":1188}},"tokens_in":692,"tokens_out":1258,"duration_ms":13379,"temperature":1.0,"reasoning_tokens":1188,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:35:02.170242+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the reversal experiment starting from a developed dune that was not formed from a heap (for example, a dune equilibrated on a flat bed) and measure the inversion time divided by $t_c$; if the ratio departs from roughly 2, the factor-of-two rule is tied to the heap initial condition and not a universal reversal timescale.","supporting_citations":[{"cited_title":"\\ Franklin, E M","cited_arxiv_id":null,"evidence_quote":"Supplies the turnover timescale $t_c$ for subaqueous barchan growth and the horn-growth method used to define characteristic times."},{"cited_title":", Yang, X","cited_arxiv_id":null,"evidence_quote":"Provides the mean residence time of about 10 turnover times in the barchan central slice and the dispersion-concentration mechanism that justifies treating the central slice as morphodynamically representative."},{"cited_title":", Lovett, S","cited_arxiv_id":null,"evidence_quote":"Establishes the 2D circular-flume dune experiments and the wake-repulsion behavior on which the present experimental setup is based."},{"cited_title":", Narteau, C","cited_arxiv_id":null,"evidence_quote":"Provides field evidence of reversing dunes and the speed-up mechanism that motivates studying flow reversal."},{"cited_title":"\\ M\\\" u ller, R","cited_arxiv_id":null,"evidence_quote":"Gives the bedload transport correlation used inside $t_c$, with the same $∼ \\theta^{3/2}$ form as Bagnold's law."}],"review_version":1}