Dear Editor,

I finished reading through the manuscript by Yaminishi and Naruse. This is a very complete study of the role of sediment dynamics and lithology for the simulation of bedrock river profile evolution. Using an impressive combined dataset of remote sensing, field, and lab measurements, the authors show that Sediment-Flux-Dependent-Models (SFDM) yield better predictions than other options such as stream-power models adjusted to consider alluvium. The authors went so far as to measure the tensile strength of each bedrock unit in the lab rather than rely on reference values. This work is an important contribution to the ongoing discussion about the treatment of sediment dynamics in the modeling of river and landscape evolution. I have a few comments that I would like to see addressed, and a series of completely optional recommendations that could improve the manuscript’s ease of reading. Overall, minor to moderate revisions is all this manuscript needs to be published.

I will address the authors directly in the remaining text for simplicity.

Grain size data.

Some additional information is needed to better understand how grain size data is collected. To my eyes, unfamiliar with the rivers in question, the D50 values of most rivers, between 40 and 70 cm are very high. Is the necessary bankfull depth to transport these clasts reasonable with observations?

A back of the envelope calculation using Shields stress to calculate the critical water depth for incipient motion,

tau* = (hS)/(RD) = tau*_crit ,

gives a water depth of ca. 2m at a slope of 2%, ca. 4m at 1%. Does that seem reasonable given the field context?

What is the minimum size that the drone survey could consider (I am not very familiar with the method), does that artificially raise the D50 compared to a hand survey down to ca. 2 mm grain size? A quick glance at a few granulometry tables from the data on Zenodo shows that on ca. 1 cm seems to be the smallest grain size. How does that bottom threshold impact the D50 calculation?

More details about the manual pebble count is also needed. Which method was followed? What was the threshold to “fine”?

Timescales.

Towards the end of the introduction, l. 82–91, you go over various constraints for rates of erosion/denudation/exhumation/(rock?) uplift. While all these processes are closely related they are not always identical. And often they are measured over different integration timescales. These two paragraphs need to be revised to be more rigorous and explicit when providing this background.

The exact terminology for the various types of erosion-related rates should be double-checked. Thermochronological ages provide a rate of exhumation which is equal to the negative local erosion rate. But denudation is a little different as it regards the evacuation of material at the scale of a catchment (due in large part to the work of erosion). I think that thermochronology can constrain erosion, but not denudation. And basin-wide cosmogenic studies provide a rate of denudation (at the scale of the catchment) and not of erosion. (It is worth checking these terms against another source than myself.)

DIsplay of results

The models rely on a lot of variables that evolve along stream. Some are all strictly proportional: A, Qw, W in SFDM. Some are (partly) independent, S, H, Pc, tau_s, Shields stress. Others come from field and lab surveys D, sigma_t. When parsing through the results, I wanted to see how they all evolved to understand why a model behaves this or that way. Would it be possible to provide more information side by side on a representative profile? E.g. for the Gohyaku river, where the three models behave very differently. Figures 9 and 10 can be combined advantageously, and a figure could be dedicated to showing one river in more details.

Bankfull width and depth

Given the experience in the field, you have an opportunity to check further values of the models against measurements or at least qualitative impressions. Does the predicted width falls in the range of observed bankfull width. What about the bankfull water depth? It is not explicitly calculated but lies indirectly in Eq. 14. If you compute it, how does it compare with field observations? I don’t know if width and depth were surveyed, but even if they were not systematically measured, it would be good to provide an impression in the context of your field experience.

Lithological controls.

The SFDM model considers the tool effect, but it does not directly take the tensile strength contrast of tools and bedrock. As far as I understand, it does however indirectly acknowledge it because tool size is part of the equation and should be reasonably expected to increase with rock hardness. Is this correct? If so, you could mention the presence of this indirect role of tool/bedrock strength in the discussion.

I have added particular comments regarding specific lines below.

l. 44 “incision episodes”

l. 57–60, cite the authors of these models right away.

l. 61. It would be useful to read the motivation for picking this site. Surely there are many catchments in Japan with varying lithology where that work could be done. I gather it is due to the availability of studies constraining rates of uplift and erosion.

l. 70. “while”. just describe them as “inactive”?

l. 72. “[…] rocks, Cretaceous igneous rocks, Middle Miocene […], and Late Miocene” without article “the”

l. 73 “are muscovite-biotite”, no “the”

l. 74. “Formation age” I think

l. 75 “of granodiorite”

l. 80 no uppercase for formation when alone

l. 82. You are talking about rock or surface uplift here? It would be good to specify it especially since you write about exhumation in the same paragraph.

l. 84: how did Fujiwara get to that number. I am not able to access the original article in Chikyu Monthly.

l. 85–86: what is the integration time of the U-Th/He measurement?

l. 89 “most of the results”. Is that the “results” of Matsushi? To me the formulation suggests that there are multiple studies, each with their own results.

l. 90. Fukuda et al. produce thermochronological constraints for exhumation rates. It should be mentioned since the last denudation rate is from cosmogenic nuclides.

l. 99 eta denotes “surface elevation”

l. 99 and following. Make sure to specify “rock uplift”. Here d_eta/dt is surface uplift.

l. 109 and following, the equations are part of the sentences and should be followed by punctuations, comma or period, as needed. Further nitpicking but I believe that subscripts that are not variables themselves, but simply qualifiers of the main variable, should be written upright and not italics. Just as in equation 26.

l. 124 just a clarification for me, rho_s is the standard density, or the submerged density of the rock?

l. 130 I enjoy the step by step explanation of the model setup. We seem to jump quite suddenly to Ir without much context compared to all other steps.

l. 154 extra space after opening parentheses

l. 178 in this long expression, W is a function of Qw (itself already present in eq. 19) and a coefficient kw (eq. 18). Wouldn’t it be simpler to insert eq. 18 in eq. 19 and get rid of W to only keep kw a parameter you will optimize for?

l. 182 bedrock singular

l. 184 Shobe et al. 2017 (GMD) are the original authors of SPACE and should be cited here.

l. 215 Should there be a reference for Optuna?

l. 220 how are kappa_a and kappa_b set? For ks in equation 10.

l. 228 It could be clearer to directly refer to the critical drainage area: “The pixels draining more than 0.7 km2 were regarded […]”

l. 236 “linearize”

l. 241 bedrock is not countable, so singular. Also on line 244.

l. 257 “taken by drone”

l. 258 “3D point clouds” no article the

l. 259 “using Agisoft” no article the

l. 260 “Triaxial ellipsoids” no article the and plural

l. 272 “measured at two locations” no article the

l. 272–275: Can you provide details about the meanual measurement? Was a it a heel-to-toe Wolman count, did you measure all the clasts in the top x-cm of the squares shown in Fig.2? Was it on a river bar, or across bar and channel?

l. 286 concave up?

l. 291 what is “slightly”? An actual distance would be useful.

l. 292 “Check dams”, no definite article

l. 292–293 what about Sangasawa? The check dam does not correspond to a break in slope?

l. 294 I’d suggest to consider removing “typical”. I do not know what is a “typical” knickpoint in this context.

l. 307, pick one of the adverbs “closely” or “well”

l. 307. I think the reference to Fig. 6 is missing earlier

l. 308 “in the Gohyaku RIver.” “in the Sakura River”

l. 317 this is very very coarse, I’m surprised! Gravel is a size category from 2 to 64 mm. 400 and 700 mm are boulders. A D50 in boulder category is truly massive and suggests debris flow control rather than fluvial processes (from my experience). Are you sure there is not a misplaced decimal here? Alternatively, does the remote sampling of grain size introduce a bias by missing grains below a certain threshold?

The examples shown in Figure 2 look much finer than the D50 listed in the text.

l. 322 Provide a reference to Figure 9

l. 324–325 I realize now that channel width was not surveyed in the field or remotely. Width being an important component of the calculations, can you provide a ground truth for these values? Are they reasonable?

l. 328 Is the Sangasawa knickpoint at ca. 3000 m at the transition granodiorite to sandstone? It is difficult to see in Fig. 9 d). I suggest to add a label on the figure.

l. 348 “Sediment-cover effect”

l. 357 and following. Given the importance of the cover ratio. Would it be possible to plot the ratio along the profile of Fig. 10 (maybe on a shared y-axis) so that we can see how it changes along slope and lithology? Fig. 12 provides the theoretical expectations. It would be helpful to see it applied to a river. And/or, as suggested at the beginning, provide a separate detailed display of the different variables modeled for one of the rivers.

l. 385 and following. Could you provide some references here? I imagine that the sediment-stripped channels of out-of-equilibrium systems are meant to be only for those that react to an increase in rock uplift rate, or precipitation. A river flowing across a mountain that that halved its rock uplift rate will also be out-of-equilibrium but wouldn’t it be going through a phase of alluviation, at least downstream?

l. 399 and following. In-text citation format where needed.

l. 416 Gohyaku

Figure 1. Missing the label (a) for the inset. In b) could you leave the hillshade as a semi transparent layer on top of the DEM colors? It’s a shame to mask it. c) the legend of the geological map is incomplete. I distinguish two shades of yellow/orange, one blue, and three (?) greens that are not documented.

Fig. 9–10. In my opinion Figures 9 and 10 should be combined. Seeing the three fitted models together — SFDM, A-SPM, and SPACEM — would show the quality of fit even better. And it would save you one figure in an already long article. It would be good to indicate with a symbol (arrows, vertical lines) where tributaries join the main stem to provide information that is otherwise found on Fig. 4 and hard to read by going back and forth.

Fig. 10: the symbol for A-SPM is a dashed line in the legend but a dash-dot pattern in the figure.

Good luck for the revisions,

Luca Malatesta

This paper represents an important scientific contribution, as it addresses a highly under-researched topic. I therefore strongly recommend its publication, pending some editorial revisions to the text, the inclusion of a broader body of relevant literature, and the addition of a sensitivity analysis.

In particular, the models rely on a large number of variables, most of which have been derived empirically. I wonder to what extent the model results depend on the uncertainties associated with these variables. This issue is not discussed in the current version and should be further explored.

Since the paper focuses on the influence of lithology on channel morphology and morphometry, a more comprehensive review of the literature on this topic is also needed. I have suggested several articles that could be considered; however, I have no objection if the authors prefer to reference other works. My main point is that greater acknowledgment should be given to the contributions of geologists and geomorphologists who have studied the role of bedrock lithology in controlling landscape forms.

Finally, the manuscript requires substantial revision of the text, as there are numerous sentences with grammatical errors.

Overall, I find this work highly valuable and congratulate the authors on their results. However, additional effort is required before the paper is ready for publication. Please refer to the attached document, where I have provided comments and suggestions for improvement as sticky notes. Since my comments are mainly editorial and neither technical nor really scientific (except for the sensitivity analysis), I clicked minor revisions. But some major editorial work on the text needs to be done.

Sincerely,

Fritz Schlunegger

Overview

This manuscript uses a detailed set of field data from a watershed in the Abukuma River basin, Japan, to test three models of bedrock river erosion, each with varying degrees of emphasis on bedrock strength. The new field dataset is thorough and includes measurements of rock strength and particle size, coupled with remote-sensing-based analysis of channel gradient and size. The field area includes a wide range of lithologies (particularly transitions between strong granodiorite and weaker sandstone), which makes this an excellent natural laboratory to test the control of bedrock strength on river profile and erosion. Ultimately the authors find a much better match between the model that includes sediment cover effects with the observed river profiles; models that include bedrock strength but not explicit terms for sediment cover do not produce river profiles that match the observations. Based on this, the authors conclude that bedrock strength is primarily an indirect control on river gradient, because it does influence sediment cover via sediment production. Overall I think this is a very interesting manuscript with a good coupling between field data and modeling; below I pose a few questions about knickpoint locations and formation, and note some English-language errors in the manuscript.

Principle Concerns & Questions

1. Lithologic boundaries and river profiles

The authors state in lines 284-285 and the figure 4 caption that there is not a relationship between river profile and lithologic boundaries, but there are several places in figure 4 where it appears to me that changes in lithology coincide with knickpoint locations. Specifically the following locations: a) Takinosawa – between siltstone and sandstone, and between sandstone and conglomerate with an intervening tuff; c) Fukazawa – between sandstone and gneiss, and between breccia and tuff; d) Sangasawa – between granodiorite and sandstone (explanation is given about drainage area increasing, but this also coincides with the lithologic boundary quite closely…).

Certainly there are changes in slope in each profile that do not coincide with lithologic change, but these numerous examples make me seriously question the assertion that there is no relationship between river profile and lithology. I think a more detailed evaluation of this claim is warranted, and ideally the connection between knickpoint location and lithologic boundary change should be evaluated statistically.

1. Knickpoint genesis

Although this paper isn’t about the geomorphic history of the Abukuma basin (and that is fine!), some explanation is needed about the knickpoint origins in these profiles, especially since the authors are asserting that they are not controlled by differences in bedrock strength. The chi-plot analysis does not show corresponding knickpoints in different tributaries (lines 294-296), so the knickpoints are inconsistent with a watershed-wide base level control. Are there upstream controls? Or is it indeed lithologic variation? This point needs to be explained in the discussion section.

Other Comments

Overall this is a very well-written, clear manuscript so I have only a few line comments. There are some small wording/grammatical errors throughout the text, so I recommend careful proof-reading. Those errors are not denoted here, but below are a few comments about line-specific items.

Line 21-23: unclear what you mean here

Line 42: wording

Line 49-50: citations needed – you say “relatively few studies”, implying there are some which should be cited here

Line 85: clarify whether there is any evidence for variation in uplift rate within the study area specifically

Figure 1: show the mapped stream network and location of major faults

Figure 5: it would be useful to show the lithology along the channel profile on this chi plot as well, especially considering the discussion point that the transition from sandstone to granodiorite in 2 of the rivers might explain their different lines in this figure.

Review of: Limited influence of bedrock strength on river profiles: the dominant role of sediment dynamics, by Yamanishi and NaruseReview by Gary Parker

This study is timely and interesting. The most important conclusions are as follows:

1. SFDM (Sediment flux dependent models) best capture predictions of the long profiles of bedrock rivers.

2. Local lithology does not directly influence channel slope due to a negative feedback between sediment cover and river gradient.

The paper is generally well-written. It fills an important gap, in that “relatively few studies have quantitatively examined the combined influence of bedrock strength and sediment cover on actual river profiles using direct measurements of rock strength. The methodologies appear to be justified well. There is a point that needs to be clarified before the paper is accepted for publication.

This point concerns the SFDM model. This model depends on the value of cover factor P_c, which must be between the values of 0 and 1 for incision (Eq. 10). Cover factor P_c is related to volume bedload transport rate per unit width q_s through Eq. 11, and q_s is in turn related to erosion rate E(x) via Eq. 16. The predictor for slope at steady state (E = U) is then Eq. 16. I can’t see through this analysis to determine if indeed P_c is between 0 and 1, and if it shows some consistent pattern of variation downstream. I think that material needs to be added to the manuscript in this regard. The issue could be clarified by plotting predicted P_c versus streamwise distance along with lithology type on diagrams corresponding to Figure 4.

ASPM is a variant of the stream power model. While this model has proved quite useful as a tool for studying bedrock channels and landscapes, it is reaching the end of its usefulness. This is because the coefficient k_a in Eq. 21 has weird dimensions, which in turn indicates insufficient physics. It is good to see other alternatives which include cover explored here.

Line 348. “…these smooth river profiles…” could be confusing, as it might be thought that the statement refers to the smoothness of the bed. I have suggested alternative wording” “…the smoothness of the longitudinal profiles of these rivers, in which clear breaks corresponding to change in lithology are not seen,…”.

Line 349. “Through the cover ratio being higher in soft rocks and lower in hard rocks, erosion is suppressed in soft rock areas…” This needs to be shown explicitly in terms of plots of P_c, as noted above.

I have added notes to the pdf form of the manuscript where some modification of wording would be helpful.