General Comments

This manuscript presents a technically rigorous and scientifically meaningful study into how grain-scale roughness and scour morphology influence flow organization, turbulence structure, and bed stress fields around emergent cylinders in supercritical flow. The authors combine high-resolution Structure-from-Motion topography, grain-resolving computational meshes, and LES-VOF simulations with controlled flume experiments, an impressive methodological framework that allows them to examine supercritical hydraulics in gravel-bed channels at a level of detail that is rarely done.

The contribution is significant. Smooth-bed assumptions still dominate many hydraulic engineering, ecohydraulic, and sediment-transport models, yet are often inappropriate for steep, gravel-bed rivers where supercritical flows and complex roughness elements are common. This paper fills an important gap by documenting how roughness fundamentally alters velocity fields, coherent structures (e.g., horseshoe vortices), and shear stress distributions. The work is timely and well aligned with the growing recognition that grain-scale processes matter for slop e, velocity, roughness relationships in steep rivers.

The manuscript is overall well written and thoughtfully analyzed. However, the narrative is occasionally verbose and sometimes overstates interpretations using subjective wording (e.g., “remarkable,” “critical,” “severely”). The Results section in particular blends results with interpretation, and the Discussion sometimes introduces claims that should be more firmly grounded in existing literature. Organizational improvements, such as adding a table summarizing flow parameters across flume and numerical runs, would help the reader follow the experimental design.

With moderate tightening and clarification, the paper will be a strong contribution.

Specific Comments

• Line 21: “grain-dominated flow fields” is unclear. Consider revising to “flows dominated by grain-roughness effects.”
• Line 15: The phrase “rough flat beds representing post-flood recovery conditions” needs clarification.

• Line 108: Clarify that the hydraulic conditions described here refer to inflow boundary conditions.
• Sediment feed: The text should explicitly state that no sediment was fed during the mobile-bed test, so scour developed from local hydraulics alone.
• Line 158: I do not see a definition of SST. Please define upon first use.
• Line 165: The use of a 0.7 m inlet-to-obstacle computational length seems short for supercritical flow. Please justify that this distance is sufficient to establish stable inflow hydraulics and discuss any limitations.
• Lines 175–185: Please elaborate on how SfM warping was minimized or corrected. Was a local coordinate reference frame used? What was the workflow for aligning imagery and GCPs?
• Consider including a table summarizing all hydraulic parameters (Q, h, Fr, roughness condition, obstacle dimensions) for both physical and numerical runs. This would help with readability and reproducibility.

• If different Froude numbers were used across experiments, briefly justify the rationale and state limitations of interpretation.
• Line 241: Define “deformation pattern magnitude.” Is this a normalized metric, spatial derivative, or amplitude measure?
• Section 3.2: Please clarify whether the same discharge (Q) was used across flume runs and numerical simulations.
• Section 3.2.2 and elsewhere: Phrases such as “provides critical insight into how developed scour geometry interacts with grain-scale flow physics” feels interpretive. In the Results section, consider reporting observations more neutrally and reserving interpretive statements for the Discussion.
• Lines 493–501 and similar locations: These interpretive remarks would be better placed in the Discussion.

Discussion

• Some claims are stronger than the evidence presented. For example, Lines 674–679 discuss ecological implications without citing supporting literature. Either provide appropriate references or soften the interpretive strength.
• Several sections could more clearly position the results within the context of previous work on roughness effects in supercritical flow, horseshoe vortices, and obstacle-induced turbulence in gravel-bed channels.

With pleasure I have read the article entitled “Grain roughness controls on velocity and bed stress fields around a fully protruding obstacle in supercritical flow”, which addresses an important gap in fluvial hydraulics by exploring grain-scale roughness effects in supercritical flows via a combined physical experiments and high-resolution numerical modeling approach. While its methodology and motivation are strong, with caveats where needed, I still feel that a couple of points could be better emphasised by the authors to help guide the readers, around the application of this model and limitations it may have.

The study’s experimental dataset relies almost exclusively on bulk parameters (discharge, depth) due to the utilisation of extremely shallow supercritical flows with high air entrainment rates, which precluded direct velocity field measurements via standard imaging or acoustic approaches (like PIV or ADV). This means that the model validation is dependent on qualitative visual agreement and bulk metrics, rather than rigorous quantitative validation against spatially detailed velocity measurements, which is a significant limitation for assessing the accuracy of grain-scale turbulence structures. Albeit these limitations are described in the methods and revisited in the discussion, so that readers are made aware that model validation is largely qualitative, the consequences of this choice could perhaps be further considered or systematically linked to specific claims; for example, the impact of lacking turbulence measurements on confidence in stress distributions could be discussed more critically, and the restriction to a fixed-bed, single-obstacle setup could be tied more directly to what can and cannot be generalized to natural rivers or morphodynamics.

As is typical of all numerical modeling studies without direct high resolution measurements to validate the major findings, the findings regarding stress distributions and turbulent kinetic energy must be interpreted cautiously.

The paper employs a hybrid Detached Eddy Simulation (DES) in OpenFOAM coupled with the Volume-of-Fluid (VOF) method for free-surface tracking. While this choice is justified by computational tractability, DES/VOF approaches (and even LES in general) are known in the literature to potentially miss subgrid turbulence and can struggle with highly non-orthogonal, complex mesh zones—precisely the scenario for grain-dominated flows around obstacles. The authors refine the mesh around the obstacle, grain tops, and free surface, and they use non-orthogonal correctors and bounded schemes to stabilize calculations in those regions. The authors do not provide a detailed, spatially focused error or grid‑convergence analysis for stresses and velocities at grain crests or in the horseshoe‑vortex region, so the impact on local numerical accuracy is acknowledged qualitatively but not quantified. Consequently, while readers are warned that some local details may be less reliable, the paper stops short of systematically tying specific numerical issues to specific uncertainties in the key diagnostics (e.g. bed shear distributions, near‑grain turbulence statistics). As the study’s non-orthogonal mesh with aspect ratio problems at certain places, raises concerns about local numerical accuracy in regions most critical for its conclusions, the authors are asked to elaborate further on the discussion section to help the reader assess when the application of this approach is acceptable or concerning. For example, when the focus is on modeling domain-scale patterns (overall velocity field, bulk stresses, free-surface shape), local numerical accuracy in critical regions is probably sufficient to support the main qualitative conclusions. However, in detailed applications where one is focused on the quantitative use of near-bed stress distributions at grain scale (e.g. for the precise entrainment thresholds or detailed turbulence spectra), the existing mesh quality and validation strategy are not strong enough to claim high local numerical accuracy, and results should be treated as indicative rather than definitive.

Last, the research fixes grains and treats the bed as static, omitting mobile sediment dynamics and hyporheic exchange (impermeable bed surface). While readers are informed about the fixed‑bed assumption, yet its implications for erosion prediction, threshold estimation, and morphodynamic evolution, or the consequences for ecohydraulic or biogeochemical applications are not discussed remain under‑emphasized. In natural systems, grain mobility interacts with flow and stress fields, so omitting these factors limits transferability of the results, especially for predicting erosion and morphodynamic feedbacks. This can be further elaborated in the discussion section to benefit reader’s comprehension.