Review of Generation of autogenic knickpoints in laboratory landscape experiments evolving under constant forcing

In the submitted manuscript Lavaissière and colleagues present new laboratory experiments documenting the generation and retreat of autogenic knickpoints.  Overall I am supportive of this manuscript (the experiments are really cool!) and believe that it should be eventually published in ESurf. Autogenic knickpoints are seeing increasing study in our field, and experiments, like those detailed in the submitted manuscript, are important for advancing our knowledge of how these features form, what controls their retreat rate, and how we might separate autogenic dynamics from climate/tectonic forcing in natural landscapes. The submitted manuscript has potential to address all these issues.  I believe that with some moderate revisions, the authors can revise the manuscript to be both more impactful in our community, and easier for readers to digest.  I’ve outlined my comments below, and offer the comments as constructive criticism for a manuscript that I think is cool, an important contribution, and which I look forward to seeing eventually published.

1. Mechanism of knickpoint formation and retreat

I found the manuscript to be very detailed on observations, which is good, as the authors did an excellent job of relating the results of their experiments.  However, I believe the manuscript could be more impactful if the authors expanded their discussion of the mechanisms that control both autogenic knickpoint formation and knickpoint retreat.

For knickpoint formation, the abstract stresses channel narrowing as the initiation mechanism, but it seems to me after reading the manuscript that knickpoint formation may actually cause channel narrowing (i.e., the narrowing occurs after knickpoint initiation) in the experiments. Can the authors use their experimental data to separate the role of channel narrowing and channel steepening in knickpoint creation?  Based on my reading of the manuscript and looking at the data, I think that, after the initial autogenic knickpoint has been created, spatially variable erosion caused by the initial knickpoint formation leads to channel steepening, which then allows additional knickpoint formation and channel narrowing.  Is this correct?  If so, it would be helpful to make this clear in the manuscript. If not, it would be helpful to document that channel narrowing occurs before knickpoint creation if it is indeed the change in channel width that controls knickpoint formation.  I recommend the authors create a schematic diagram/cartoon to include in the manuscript that shows the steps leading to knickpoint formation.

I think I understand the authors’ mechanism for how autogenic knickpoints continue to be created once the initial one forms (spatially-variable erosion due to changing channel width result in changes in river profile concavity).  However, what creates the initial knickpoint?  Can this be determined from the experimental data?  A discussion of this would be helpful.

Sedimentation was mentioned briefly in the manuscript, and sedimentation has been shown to play an important role in knickpoint formation and retreat in previous studies (e.g., Grimaud et al 2016; Scheingross et al., 2019). It was unclear to me if sedimentation and cover of the bed contributed to the formation of knickpoints in these experiments.  Discussing this in more detail would be helpful.

For knickpoint retreat, what is the mechanism that drives this retreat?  Are knickpoints discrete steps that are undercut (e.g., Baynes et al., 2018)? Or is it simply the steeper slope of the knickpoint leads to increased erosion rate relative to the surrounding?

I also had questions about the controls on knickpoint retreat rate.  The authors argue that knickpoint retreat rates follow a ‘bell shaped curve’ where retreat rates are initially slow, rates increase up until the knickpoint has gone approximately half way to the divide, and then rates slow.  I have two questions about this:

First, I think this pattern seems clear in experiment MBV06; however, the pattern is less clear to me in experiments MBV07 and MBV09.  If the authors want to argue for this pattern in their experiments, I would encourage them to perform some statistical tests to show that the quadratic they fit to the data in Fig. 8C is indeed the most appropriate description of the data. For example, can the authors show that the observed patterns of retreat rate vs. time are statistically distinct from a constant retreat rate?  My guess (from looking at Fig. 8C) is that the retreat rate vs. time data for experiment MBV09 might be equally well described by a constant retreat rate.  If this is the case, this raises a new an important observation that base level drop rate can play an important role in setting how retreat rate varies with distance upstream.

Second, perhaps I missed it, but I didn’t see a mechanistic explanation of why the retreat rate would follow a bell shaped curve.  I would encourage the authors to provide a detailed mechanistic description of why this should occur.

Finally, for both knickpoint formation and retreat, I would encourage the authors to discuss if the material used in the experiments imparts any bias on the results.  My understanding is that the silica paste used in these experiments erodes via clear water flow, unlike natural rock which (in many cases) erodes primarily via abrasion from sediment impacts. Would you expect different results in a natural landscape where erosion occurs primarily via abrasion?  Some discussion of this would strengthen the manuscript.

2. Expand the discussion on the implications of the work

The authors motivate their study with arguments that knickpoints are often used to read tectonic and climate history recorded in river profiles. While I think the authors do a nice job of showing how knickpoints can form autogenically, there’s no discussion about what the implications of this are for inverting river profiles or for landscape evolution. In order to make their manuscript more impactful, I encourage the authors to expand their discussion of the implications of their work.  For example, how can we use the experimental results to help separate autogenic versus tectonically (or climatically or lithologically) created knickpoints in the field?  Do the autogenic knickpoints created in the experiment have any characteristic scales or morphology that one could look for in the field to identify them?  Are there certain landscapes or lithologies that will be more prone to autogenic knickpoint formation than others?

3. Motivation of the problem and knowledge gap in the introduction

 The introduction did not seem to state a clear problem, question or knowledge gap which the authors were addressing in this study.  Instead, this comes in the methods, when the authors describe how previous experiments used a fixed outlet while in this experiment the rivers are free to adjust their width at the outlet.  This is an important point that shows how the contribution here is unique relative to previous work.  I suggest the authors revise the introduction to add this information and further expand upon a statement of the overall question and knowledge gap would (in my opinion) strength the paper.

4. Writing style

Throughout the results and discussion session, I found the authors spent significant time describing the figures with detail in the manuscript, instead of placing this information in figure captions. Especially when reading the results, it felt to me as if the authors were describing the figures, rather than trying to describe the results of the experiments and the processes that were observed. I think the length of the paper could be reduced, and the paper would be easier to read, if the authors tried to re-write these sections to focus more on the observations and results themselves, and saved the details of the figures for the figure caption. In practice, this could be accomplished by a switch in writing style.  In place of text such as “In Fig. X, we show this and that, and these observations indicate this process” the authors could instead write “We observed this process (Fig. X)” and then include additional details of what’s plotted in the figure caption.  For example, L255-260 is all text that I think belongs in the figure caption and not in the main manuscript. This is just one example, but this occurs throughout the manuscript.  I think that streamlining this text will free additional space for the authors to discuss some of the issues I’ve raised above as well as comments that may come from the other reviewers. 

Related, there were a handful of English language errors throughout the manuscript. I’ve pointed out a few of them below, but not all.  These errors did not affect my ability to read and comprehend the manuscript, and were rather minor, but should be fixed.

Minor and Line-by-line comments

Somewhere in the introduction or methods, it may be useful to distinguish a knickpoint from a step in the profile. Tectonic geomorphologists are often interested in knickpoints that extend hundreds of meters to kilometers across long profiles, whereas steps can form at much smaller spatial scales in channels.  Both steps and these longer length steepened channel reaches have been called knickpoints in the literature.  Which features are the authors referring to in their use of the term ‘knickpoint’?  And are the knickpoints generated in the experiment better described as individual steps or steepened channel reaches that extend over a significant portion of the channel length?  I think it’s the latter, but this should be made clear for readers.

L16: Change these to this.

L20: Change rivers to river (singular).

L27: Remove their

L39-41: This is vague. It would be more insightful to the reader to explain what the limitation is and what the role of sediment supply is.

L46: Unclear what ‘their’ refers to in this sentence.

L48: Change has to have

L52: The abbreviation BL should be defined as base level before first use. Furthermore, I suggest spelling out base level throughout the manuscript, as it’s not a commonly abbreviated term.

L73: Replace ‘to create’ with ‘creation of’

L87: Define the term DEM on first usage.

L94: Change ‘most of channels being straight’ to ‘because most of channels are straight’

L98: Change ‘verified manually’ to ‘manually verified’

L98: What is meant by ‘define knickpoints correctly’? Do you mean against a geometric definition?  This section is unclear because it seems there are two ways to define knickpoints.  One is based on the erosion rate relative to base level fall, and the other is a manual way (presumably based on geometry and channel slope?) that is not well defined.

L127-128: I’m confused on what the threshold is.  Is this the minimum water depth for an area to be classified as a channel?  Can this be made more clear?

L168: When using the term ‘non-linear relationship’ please indicate between what variables this non-linear relationship is expected.

L176-177: Change “whatever regardless” to “independent of”

L231: See Mackey et al. (2014) for a field case showing constant retreat (https://doi.org/10.1130/B30930.1).

L240: The authors write “they do not show a clear tendency of increasing” – I think the data actually looks pretty good. I would encourage the authors not to sell themselves short.  It could be worth quantifying this statistically to show that there is a statistically significant increase.

L260: Replace ‘channel is in average’ to ‘channel is on average’

L249-299: This paragraph is quite long and hard to follow.  Can this be simplified?  See my comments about writing style above.  Also a discussion of the mechanism of knickpoint formation could be useful here.

L265: Change ‘before to subsequently widens’ to ‘before subsequently widening’

L272: Change ‘erosion rates value’ to ‘erosion rate values’

L382: I think defeat should be decrease instead?

L383: Change ‘downstream the knickpoints’ to ‘downstream of…’

L384: Change ‘downstream retreating knickpoints’ to ‘downstream of retreating…’

L390-410: This feels more like results than discussion to me, and I would incorporate it in the results section.

L425: Change ‘downstream the’ to ‘downstream of the’

L436: Change ‘before to decelerated’ to ‘before decelerating’

Figures

Figure 3: In panel B, the solid lines are model predictions I think?  This should be made clear.  It would also be helpful to plot the measured value in the experiments.

The purple line is hard to see in a panel A, why not plot both lines as red?  Or choose a different high contrast color?

Figure 6: This looks like the jet colorbar?  I suggest using color schemes that are colorblind friendly (see these for example: https://www.fabiocrameri.ch/colourmaps/).

Figure 7: Is the colorbar the same scale in all three panels?  If so, please make the labels on the colorbar for knickpoint retreat rate the same (experiment 9 has a different label than experiments 6 and 7). Also, what’s the difference between the dashed blue, red and yellow lines around ndd of ~0.5 relative to the gray lines?

Figure 8: Is panel A needed?  Panel A & C show essentially the same information, but panel A is very difficult to read.  What about color-coding the data in panel C and eliminating panel A?  

Figure 10: The caption says blue and orange colors, but I think purple and green is meant instead? 

Figure 11: Typo in caption, it says “shear stress (FD) for all pixels”, but I think this should be ‘shear stress (D) for all pixels”

Figure 12: Is distance on the x-axis flipped relative to fig. 6?  I think it is.  It would be helpful to change the x-axis label to “distance from the divide” or something else to indicate which direction the divide is in, and make sure the orientation of the x-axis is constant between figures.

Also in Figure 12, it would help to shade or add arrows specifying exactly where the knickpoint is.  Simply putting the label K1, K2, etc on the panel is less useful than explicitly indicating where the knickpoint is.

Supplementary Information

• Please note that throughout the supplement most of the units have powers listed as subscripts when these should be superscripts.
• At the end of the first paragraph, please spell out “Table 1” instead of “Tab. 1”

Overall, the text of supplementary information has a lot of repetition from the main text (some of it is also repeated word for word).  Please eliminate this redundancy.

The paper by de Lavaissière et al. presents lab experiments designed to explore the dynamics of knickpoint initiation and retreat under constant rate of base-level fall. The novelty of their experimental is the unconstrained width of rivers and this leads to new and very promising results. Based on their observations, the authors propose a conceptual model of autogenic knickpoint formation related to changes in river width. Autogenic processes are more and more recognized as major components of landscape dynamics and this work is a very nice contribution to this topic. 

Yet, although I think this study has a great potential, it is a bit blurred in the current version. I identified several reasons for that which I present below together with general and specific comments. I do not see any major issue as I feel that this is mainly related to problems of syntax and general structure. Therefore, I’m confident the authors could address my comments or any from other reviews and I look forward to seeing this work published in ESurf.

All the best, 

Laure Guerit

Abstract

The use of the numerical model (Floodos) should be mentioned. I really like the use of numerical models, it is a great way to extract quantitative informations from lab experiments so you should advertise for it. 

Introduction

The introduction is clear and simple, which is nice. However, I think that the manuscript would gain in strength and quality with a better contextualization of the experimental approach (see also my comment below about autogenic processes). In fact, from the current manuscript, it is bit difficult to get a clear idea of the real addition of this work. This is done later in the Method but I think it should be in the introduction. 

In addition, the paper is mostly focused on knickpoints and I missed something about autogenic processes (beyond the conclusions of the study itself). I think the paper would be enriched by addressing this topic too in the introduction and/or in the discussion. It would broaden the impact and the potential audience. You could for example have a look at the review by Hajek and Straub in 2017 (https://doi.org/10.1146/annurev-earth-063016-015935).

 

l. 27-28: suggestion: the definition is ok but it would work for any step along a channel. Maybe add a scale ? 

l. 42 to 44: this sentence and the references are not really needed, maybe simply start with However or something similar. This would lighten the text a bit. 

l. 44 please change "they" to "knickpoint" (the sentence is a bit unclear with "they")

l. 46 I do not understand what "they" stands for

l. 50 maybe specify which forcings. 

l. 50 I do not really understand this sentence. Do you mean that landscapes alternate between periods of steady state and periods of knickpoint propagation ? Please reformulate. 

Methods

I would appreciate (here, in the discussion or even in Supplement) a discussion about the choice of material and how it could affect your results. 

 

I would present the three experiments here and not at the beginning of the results (therefore, Table 1 should also be moved here). This would give a better sense of what you did and it would also help with some issues with text organization (see my comments on Figures 1 for example). Finally, to ease reading, I suggest to structure the Methods into 3 subsections 1) experimental setup, 2) knickpoint extraction, 3) hydraulic data extraction.

l. 70 please give values for the base level fall rate.

l. 73 the description of the initial configuration should come before the experiment itself.

l. 98 what does "correctly" means ? This sentence suggest that you define knickpoints based on erosion rate and that you consider them as "correct" based on another criteria (slope I guess) ? The procedure is not that clear, please consider rephrasing. 

l. 111 to 126: almost word by word repetition with the supplement material.  

l. 116: the outputs are repeated line 118. Consider rephrasing

l. 117 the use of "spatial distribution of precipitation" is strange as as you state on the previous page that precipitation are constant. Do you refer to heterogeneities (inherent to the experimental design) mentioned in the supplement ? This should be clarified. 

l. 111 to 136: I found this while paragraph difficult to follow. Please try to be more straightforward or to rephrase a bit.

Results

The text misses a clear organization and suffers from repetitions. This prevents me from getting a clear sense of the main results of the study, although I believe the material is already here (I think that lines 195 and 295-296 are the main results of this study). Some ideas to help rephrasing:

 - figures are sometimes described instead of being used to support a result (l. 159-161, l. 190-191, l. 251-257) and some of them are barely used (Figure 6 for example).

 - repetitions (for ex. l. 188 and l. 195, l. 197-198 and 198-199) 

 - use numbers to quantify your statements. There are only three experiments, it is really easy to be more quantitative. For example, line 159, 166: give the three values. l. 167 "increase in average": there is only three points so I assume you can be more specific, l. 176 "very comparable" is quite vague. 

 - the initiation of knickpoint section could come before the section on the dynamics (knickpoints must form before they can propagate)

 - split the text when a new idea is developed (some suggestions : l. 171, l. 277, l. 284)

The normalized distance to divide (l. 174) and the normalized longitudinal distance of maximum velocity of knickpoint (l. 237 - maybe find a simpler expression ?)  are important parameters for your study. I think they should be defined in the Methods, not in the Results. Once again, this is give a better organization to the manuscript and ease the reading. 

l. 299-233 seem to be rather discussion

l. 233 which effect ? please specify

l. 240-243 this paragraph is a bit confusing. The authors state that there is no clear tendency but that there is positive correlation (that I could indeed observe in both graphs). Please reformulate, make your point clearer and do not minimize your results. 

l. 264 and 269 "first knickpoint" should be "second knickpoint" as you mentioned that a knickpoint is already there l. 261.

l. 265 quantify "widens"

l. 267-271 syntax is a bit complex, please consider rephrasing for simplification

l. 276 and 277 third ? 

l. 283 the mean is positive but the median is negative. What does it imply, could you comment on that ? 

l. 288-299 sounds a bit like discussion. Please avoid repetition (l.288-299 and 295). 

l. 436-440 almost word by word repetition of l. 229-231.

Discussion

Here again, I would suggest to structure the discussion into subsections to ease the reading. A general comment for this last section is that the authors do not highlight their specific contribution. I really think this study has a great potential to document in a quantitative way the initiation and dynamics of autogenic knickpoints and the authors even provide a mechanism (via river width) for that. However, this is somehow hidden in the text.

The paper could do better here with the idea of autogenic knickpoints: what are the implications for stratigraphy or for tectonic studies ? How to recognize autogenic from allogenic knickpoints ? I really think the paper would gain in strength by addressing these questions in the discussion. 

The concluding paragraph does not give credits to your work and I can not get a sense of your results from these lines. "Our model is consistent with this proposition" is quite a limited conclusion for your study. 

l. 369 why admitting ? you just demonstrated this !

l. 393-410 to me, it is a major outcome of this study, that could almost belong to the Results. This should be highlighted, maybe with a conceptual figure ? As mentioned elsewhere, I think that figures could be simplified and some even removed so it’s ok to consider a new one. I may have miss it but why do rivers widen at all ?Also, you mention that erosion is inhibited after the passage of a of knickpoint but widening implies lateral erosion. Maybe specify vertical erosion or discuss this point a bit further. 

l. 440-445 this is already said in the results. Consider removing these sentences. 

l. 425-426: this is already said l. 421-422 and said again l. 433-434. Please simplify. 

Additional comments

I noticed some minor language mistake. It did not alter my reading or my understanding of the work but it should be corrected. The text could also be simplified with a bit of rephrasing. Syntax is sometimes complicated, I noted a lot of repetitions and you could change structures like "On the graph shown in Figure 11C (l. 257), […]" to "normalized values of […] (Fig. 11C) . This would lighten the text and give some space to develop the suggestions proposed by the reviews. 

Nomenclature: I guess the names of the experiments are related to your work flow but it does really served the manuscript. To make it easier to read, what about experiments 1, 2, 3 (or 5- 10-15 to refer to the rate of base level fall) ? 

l. 32 sel-level -> sea level

l. 40 limitation -> limitations (?)

l. 74 remove "here"

l. 116 missing capital f (floodos)

l. 171 (and elsewhere in the text): "In order to be able to compare" can be reduced to "to compare". Such a simplification would ease the reading and shorten a bit the manuscript. 

l. 174 Figure 2 should be Figure 4

l. 176 remove regardless

l. 329: missing word after "once a knickpoint has" (I guess it is "passed")

l. 413: "is" instead of "in"

l. 417: missing space (Fig.12)

Figures and table

As a general comment, the figures are nice and clearly illustrate the experiments and results, but  they are sometimes a bit dense and could benefit from simplifications and harmonization (see specific comment below). The size of the text changes from one figure to another and is sometimes very, very small. I suggest considering a fixe, readable size and to adjust the content of the figures. 

Finally, some figures are not exploited at their full potential. For example, Figure 1 is only referred to once. The individual panels of Figure 1 and 3 are not referred to and there is no reference to Figure 13. The text should take better advantage of the illustrations and/or figures could be simplified or moved to the Supplementary. 

Figure 1

Panel A: please indicate the total height of the box.

Panel C: I can see numbers of the left and right sides of the box, what are they for ? 

caption: the experiments are not introduced at the point of the text. Please adjust either the text of the caption.

Figure 2

Text is really small. 

caption: the experiments are not introduced at the point of the text. Please adjust either the text of the caption.

Figure 3

Panel A: the superposition of the DEM colorsmap together with the water depth colormap makes the figure a bit complicated to read.The elevation does not seem to be used so I suggest to remove the colors. Maybe use grey shade/ slope illumination only ? If you keep the colors, please add the color scale. 

Panel B: axis numbers are very small. 

caption: it is very long ! Most of it should be in the main text. l. 146: shallow instead of thin ? 

Table 1 

nDDmax is not a straightforward variable. Please use the full name or define it in the caption

Mean knickpoints retreat rate: do these values come from Fig. 8 ? If so, maybe use similar units to ease the reading. Also, 4 digits after the comma might be a bit excessive.

Figure 5

caption: please describe the inset figure

Figure 7

The colors scale of experiment 9 should be the same as experiments 6 and 7 to ease the comparison. 

caption: please indicate what are the blue, orange and yellow lines for. 

Figure 8

Text is quite small.

The 3 panels are more or less similar and are just different ways of presenting the data. In order to simplify and gain space, I suggest to choose one way and to remove the other ones (or to move them to the supplementary). 

Figure 9

caption is a bit unclear, please use full wording instead of symbols.

Figure 10

Panel B is referred to before Panel A, please adjust the text or the figure. 

Panel A: what do the changes in colors correspond to? 

Panel C: missing color scale for the elevation. The indication of cross section is a bit difficult to read, maybe try in white ? 

The caption mentions the propagation of two knickpoints together with blue and orange colors that I do not see. Please correct.

Figure 11

Text is small

Panel C: y-axis is "normalized erosion or sedimentation" but in the caption, it is given as "normalized elevation changes". please correct. 

caption: FD should be D. "Orange solid […] percentiles": as in Figure 8B, this could be in the legend of the panels.

Figure 13

I found no reference to this figure in the text.