The manuscript submitted by Martin and Edmonds investigates the planform differences in megafans in foreland basins as a result of abandoned channels topography on avulsion pathfinding and their healing mechanisms. The study is motivated by three megafans and their channel locations and topographic features, including abandoned channels, alluvial ridges, and internal basins. A coupled 1D diffusive channel bed elevation model and 2D cell model is used to investigate both the effects of abandoned channels as attractors or repellents for avulsion pathfinding and the effects of different topographic annealing styles on modifying abandoned channel topography. Martin and Edmonds’ results are very exciting and show that abandoned channels when acting as both attractors and repellents, create the characteristics topography of foreland basin megafans. Planform characteristics are also distinct between the proximal (near apex) and distal fan, which are directly related to avulsion potential and abandoned channel spacing. In their model, this spatial change in planform characteristics is only achieved if abandoned channel topography is transiently preserved and both high and low topography is diffused away over time. The megafan indefinitely progrades downstream if only the high or low topography of the abandoned channel is removed. These results provide a new and remarkable context to study the annealing of abandoned channel topography.

Below are major and minor comments and questions for the authors to consider.

Major Comments

The introduction is a great summary of the motivation for this study. There is an opportunity to put the subsequent modeling efforts in context to other avulsion modeling frameworks, including those mentioned in the discussion and avulsion models related to deltas.

Besides avulsion processes, floodplain processes are the other addition to the modeling. The introduction glances over floodplain processes because not as much is known. Even with the limited knowledge about aggradation rates and their spatial trends on floodplains, the choice of spatially varying floodplain aggradation rates away from the fan apex and not away from the active channel needs to be better supported.

Lines 81-82: How does this result compare to simulations?

Lines 138-139: Currently, only the avulsion location variation is plotted. What are the implications for timing?

Lines 213-214: Please clarify why only non-active channel cells have subsidence, especially since subsidence is accounted for in the equations of both the aggradation rates and 1D diffusive channel bed elevation model.

Lines 214-215: Please describe the motivation for varying floodplain aggradation away from the mountain front but not the active channel. Does this affect the planform result and avulsion patterns? What are the processes that distribute sediment across the floodplain in this system, especially with only one active channel (Line 84-85)?

Lines 331-332: See previous comments. The motivation for floodplain aggradation changing downstream but not with distance to active channel seems counter-intuitive. Please explain how results would be affected if floodplain aggradation rates varied with distance to channel.

Line 335: A_fp,f is listed in Table 2 as having two different rates for proximal and distal. Please describe this variation here and share how the boundary between proximal and distal is found. Does the boundary location change for each timestep? And how would these results differ if only one fix rate existed?

Lines 335-336: Please consider presenting the formula for A_fp,v here. Based on the description given, it's not clear if this is a positive or negative linear relationship with height difference.

337-338: Does channel depth also change downstream? What are the motivations for normalizing the floodplain aggradation based on channel depth?

Equation 12: See previous notes (and please feel free to refer to your answers there). The model presented here is a physical-based cellular model (line 13). Please describe how this formula is inspired by our communities’ understanding of floodplain aggradation processes.

Figure 4: In the physical-based model, how would the deposition only or erosion only healing modes affect the sediment availability in the surrounding fields since either sediment is needed or transported away in these modes. Would adding a component accounting for this in the floodplain aggradation affect the potential for equilibrium to be reached (lines 20-22)?

Line 357: Please be more specific of the location in the discussion, where the choice of 55,000yrs as a healing timescale is explained. How does the healing rate compare to floodplain deposition rates described previously?

Section 4.1. The planform topography and feature similarity between observations and the model are striking. A short discussion on how channels bounding some of the observational fan (Fig 1 E, F) could be affecting the along-strike comparison would be helpful for context. Additionally, a short description of the 1D model validation, including the water depth of the channel, where available, would be a powerful addition to this section, especially since mean water depth dictates the healing rates (Eq 13).

Line 411: I would encourage acknowledging that some parameters were varied between the proximal and distal zone (Table 1). Therefore, it’s unclear if the results are affected by the choices in parameter variations.

Figure 6: Accounting for surrounding topography in T_a (Eq 1 and 2) shows a striking increase in avulsion location, especially in the proximal part of the fan. How different are the corresponding T_a distribution (Eq 1 and 2) between the red and blue line model runs?

Figure 7: Are the striking differences between distal and proximal related to slower healing rates and different water depth (h_avuls etc) for avulsion pathfinding?

Before describing the main discussion points, a short overview of the sensitivity of the results to slight variations in floodplain aggradation and subsidence rate (Table 1) and the ratio between them would be beneficial in a section.

Line 648-649: One set of studies that have been conducted related to channel beds and levees during and after avulsions has been led by Dr. Brandee Carlson on the Huanghe River, China. Please consider including the findings of the studies here.

Minor

Line 10: Previous modeling work is portrayed as if they are vastly different from this study. I am not sure that’s representative.

Figure 3 is missing its caption.

Section 3.2: Does the choice of the 1D long profile elevation model affect the profile, water depths, and sediment transport rates?

It’s unclear how the 1D model is used to initiate the 2D model. I would suggest including a plot of the initial 2D domain.

Line 148-150: This information would be extremely helpful in the introduction.

Table 1: Please add notes to the caption that describe the motivation for the choice of parameters value. Apex elevation is missing in table 3.

Line 201: Will subsidence affect surface slope in the model?

Table 2: Please add notes to the caption for the motivation of the parameters. Especially sediment discharge, incoming sediment supply, and basin width.

Line 232: It’s not clear how Table 2 relates to solving h_chan. It would be interesting to include a description of how h_chan is found.

Line 262: Please include where the formula for healing rate is found in the manuscript here.

Lines 268-270: I am confused by this statement. Could floodplain aggradation allow for the elevations of these cells to increase?

Lines 289-290: Please clarify to which cells this applies Are floodplain and abandoned channel cells equally likely to be a site for avulsion triggers?

298-299: Please specify why 30 years was selected here. How does this compare to results from equation 2 for model runs?

Line 310: Please describe how is h_avul is calculated.

Line 341: I think this should read equation 12.

Figure 5, 6, 10, etc. A more complete description in the caption of the planform elevation model results would be useful, including a color bar scale and a description of what is plotted – I am assuming it’s all cells that have been channels at one point in the modeling.

Lines 433-434: Please clarify this statement. Currently, (b) doesn’t appear to have a vertical scale.

Lines 507-510: This downstream shift in sediment conveyance because of internally drained floodplains is an exciting result and makes a lot of sense physically.

Lines 535-537: I wonder if one of the implications for this result is that more sediment will exit the domain at the downstream extent. Does this make sense with the physical-based framework set up here?

Lines 581-592: This paragraph would be a great context for the introduction.

Lines 596-597: This is super interesting, especially since it’s in agreement with previous findings.

Lines 611-625: This paragraph would be a great context to motivate the study and results.

Lines 649-650: The sentence is currently not complete.

Review of “The push and pull of abandoned channels: How floodplain processes and healing affect avulsion dynamics and alluvial landscape evolution in foreland basins” by Martin and Edmonds

General Comments

              The manuscript presents novel and interesting results from a new cellular model of river avulsions in a foreland basin setting that explicitly parameterizes the influence of abandoned channels in channel pathfinding. Although several models of river avulsion exist, there has been a lack of attention to floodplain topographic controls on flow routing, and this manuscript presents an interesting and logical new parameterization of the ways in which abandoned channels might control avulsion pathways. They treat abandoned channels in three ways: 1) their repulsion caused by the elevated topography of alluvial ridges; 2) their attraction because of channelized flow paths; 3) different modes of abandoned channel healing, depending on abandoned channel deposition and erosion. Model results show that all tested aspects of abandoned channels control avulsion location, especially whether avulsions are occurring in the proximal or distal reaches of the fan. They also show, most interestingly, that active lobes and lobe switching – that is, when avulsions occur in a clustered region of the fan before switching to a new, clustered region – only occurs under certain abandoned channel healing conditions. This suggests that abandoned channel topography exerts a strong control on the locus of avulsions and also the tendency for some fluvial fans to have active lobes. This is a valuable modeling contribution that sets the stage for field and remote sensing work to gather more real-world data about the role of abandoned channels to test some clearly defined model predictions. At the beginning of the manuscript, they also characterize the amount of abandoned channels on three fluvial megafans, to show that these are indeed prevalent features in modern environments. Overall, this is a well-written manuscript with exciting findings; the comments below are mostly minor questions or suggested revisions to improve the clarity of the manuscript before publication, although there are a few more significant comments, especially about the default channel healing mode used in most of the model runs.

Specific Comments

Base run and validation

In section 4.1, the authors validate the model results by comparing a baseline run to 2 modern megafans. However, it is unclear why the strike and dip-section profiles are compared to two different megafans. Is the dip-section of the model inconsistent with the Taquari fan? Is it important to have reasonable consistency in both dimensions? A lot of the main conclusions of this paper are about the proximal-distal location of the avulsions, so the length-to-width ratio of the fan area does seem like an important property. Also, for the along-strike comparison with the Taquari fan, the Y-axis values are comparable but the x-axis values are very different; the plots make them look the same, but the axis values are different. How important is this difference? Additional text should be added to address this, and ideally the dip and strike profiles should be compared from the same fan system or systems.

Abandoned channel healing mode in the first 3 sets of model runs

Lines 374-387 describe four series of model runs that are then analyzed in the results section. If I understand this correctly, series 1-3 were all run in far-field directed healing mode, and thus the effects of the repulsion and attraction parameters were tested only with the far-field healing. However, in the fourth series of model runs, the authors show that the depositional and erosional healing modes cause major (and interesting!) changes in avulsion behavior. Do the impacts of the attraction and repulsion parameters still occur under the depositional/erosional channel healing modes, or are they only important variables under the far-field directed healing mode? It would be good to see some model runs added (perhaps just a limited subset of the model space) that address this question. Is the channel healing mode more important than the attraction/repulsion rules? Also, what is the justification for using the far-field directed mode as the default mode? Is there evidence to suggest that this is the most common/reasonable healing mode in modern floodplains? The authors mention that this is the least computationally intensive, but it would be good to see scientific justification for this as the default mode.

Model mass balance variations based on healing mode

Because of the way this model is set up, mass balance is not constrained between model runs, and that is not an impediment to the analysis. However, the different healing modes shown graphically in Figure 4 have very systematic differences in mass balance that might have big impacts on the model results. For example, the depositional only mode of healing would require a much larger sediment input to the floodplain than the far-field directed or erosion-only modes. There is no mention of this in section 4.5 or in the discussion, but I think this difference in mass balance between the model types warrants some analysis. If equivalent amounts of sediment were added into the proximal floodplain in the far-field directed healing model runs, even outside of the abandoned channels, wouldn’t that also cause lobe switching? In other words, to what degree are these results caused by the abandoned channel healing versus just accumulation of more or less sediment in the floodplain around the active channel belt?

Re-organization of introduction & background

The introduction section of this manuscript is very short (3 brief paragraphs!) and does not give enough background to set up the hypotheses or the model set-up. Other sections of the paper (specifically section 2.2.2 and Lines 611-625 of the discussion) would be better suited to this introduction, so that more specific background about abandoned channels and cellular modeling of avulsions was provided to the reader before they continue on in the paper. Also, the topic of avulsion reoccupation is not mentioned in the introduction, but the observations from the modern and the ancient that channels commonly reoccupy previous channel courses is critical to the motivation for this paper. There should be at least some description of the evidence for channel reoccupation of abandoned courses in the introduction here.

Discussion section revisions

As noted with detailed line numbers in the following section, a lot of the discussion reads more like background about avulsion models and justification of the model rules used in this paper, rather than contextualizing these novel results. I think the background information and justification of model parameters should be moved earlier in the manuscript, and a more detailed analysis of the results could be added to this discussion section – especially thinking about their broader implications. For example, based on the prevalence of lobe-switching in modern fans, does this suggest that the deposition/erosion-only healing rules are most consistent with modern observations? Additionally, what are the implications for stratigraphic analysis of avulsion patterns? In systems with clearly clustered avulsions (such as the Ferris Formation; see Hajek et al. (2010)), does that suggest abandoned channels were attracting avulsing flow more than in randomly distributed systems (e.g., the Williams Fork Formation; see Chamberlin et al. (2017))?

Technical Comments

Section 2.2 - Megafan floodplain topography discussion: the organization/title of this section is odd, because this is really the background needed for the model set-up, not a discussion of the field results. I think almost all of the content in this section would be better in a background section about avulsion set-up, initiation and pathfinding that would come before the remote sensing section.

Figure 3: missing caption

Line 140-141: There is also good evidence from the rock record that avulsion reoccupation of previous channel courses is common – e.g., see references in Chamberlin and Hajek (2015).

Line 161: There are several studies that have observations of oxbow lake sedimentation rates, which is a type of abandoned channel sedimentation. These studies should be cited and discussed here. (for example, Wren et al., 2008, “The evolution of an oxbow lake in the Mississippi alluvial plain”).

Line 163: The language “if one assumes that abandoned channels do heal” is confusing – what would another option be? Over geologic time, they must heal, right?

Line 168: The details in the discussion should be moved up front into a background section.

Table 1: Maybe “numerical” or “cellular” model is more clear than “non-experimental” in the table title?

Line 263: This is an interesting way to code channel healing! Can you add some more explanation – maybe here, maybe in the justification for the model set-up – about the mechanistic justification for these different healing modes? In other words, for the depositional only healing, how would that actually work? Via overbank sedimentation? Via temporary reoccupation of the abandoned channels during floods? I know that this model is not attempting to resolve those processes, but some general outline of how each of these healing mechanisms could be possible would be helpful for thinking about the implications.

Line 333: I’m surprised by this decision to have floodplain aggradation independent of active channel position. Why are they decoupled? Additional justification of this choice would be helpful, because this would have a big influence on superelevation dynamics. This might be a point that would be good to add into the discussion and a direction for future modeling work.

Line 341: I think you mean equation 12, not 17

Line 347: Put the detailed explanation of other models in the background, not discussion, and then this line can be removed from this paragraph.

Figure 4: this is a very helpful figure that I referred to many times when reading this!

Table 3: I think presenting model run parameters based on figure # is confusing. It would be more straightforward to include a table that shows the parameter ranges for each of the four series of model runs.

Lines 472-474: Interesting! Is there any evidence for this in modern systems, that avulsion nodes cluster immediately downstream of previous avulsion nodes? This would be an interesting point to expand on in the discussion.

Figure 12: Make dashed lines have a bigger dash, they are hard to distinguish from solid lines. Also, typo in the first part of the caption – missing “on” before (a).

Line 592: Abandoned channels impact the compensation rule in the Chamberlin and Hajek models because they leave an elevated abandoned alluvial ridge, so they cause repulsion away from the abandoned channels in the compensational rules.

Lines 596-598: The clustered mode in the Chamberlin and Hajek (2015) model randomly selects a channel location within the clustering zone; it only selects the lowest location when an elevation threshold above the far-field floodplain has been reached.

Paragraph beginning at line 611: I think this would be better suited as background material.

Lines 637-647: This material would be helpful for justifying the healing rules used, and thus could be presented earlier in the paper.

Section 5.3: These are interesting predictions focusing on the proximal-distal location of avulsion nodes! I think adding predictions about channel healing & lobe-switching behavior would be very helpful too. Are there characteristic types of fans that show lobe switching? Is there something about the floodplain aggradation style on those fans that influences abandoned channel healing? That would be really cool to explore.

I found this manuscript to be clear, well-structured, and very detailed. The model design makes a lot

of sense, and I think the authors have shown a few very intuitive outcomes while also demonstrating a

few less intuitive ones that spark interest. In particular, I thought the outcome where avulsion locations

shift basinward when abandoned channels are barriers to flow was very intuitive, and makes for a satisfying

result. In contrast, I found it surprising that imposing a rule that only negative or positive relief can be

erased can drive the model to never achieve steady-state. These outcomes are presented and framed well,

the conclusions are well-supported and impactful. My constructive comments are limited to some

comments on the visual presentation of the figures, and a few clarification questions about model design

choices. Other than these, I recommend the article be published. I look forward to citing this paper when

my future work involves the stratigraphic architecture of fans.