The paper presents a valuable exploration of the sources of error and potential improvements of common predictive methods for debris flow parameters.  I have a few suggestions below:

Figure 3 - isn’t this backwards for (b)?   Blue is deposition and red is erosion?

Figure 5 - I’m still confused.  It makes sense that blue, representing positive lidar differences., should be positive.

Figure 6 - why would V3 be rectangular?  Shouldn’t it be trapezoidal?

Figure 9 - Traditionally, shouldn’t the observed (known) quantify be on the x-anis and the predicted (unknown) be on the y?

Figure 11 - “Damming can be observed in the discharge record of the Colorado River on 22 July 2021. (e)”

Line 61 - consider changing to “more frequent and higher overland flow”

Line 92 - “are applicable” versus what?  “No longer than the first two years”?  Might help to clarify

Line 116 - capitalize “Quaternary”?

Section 3.3 - you might comment on the fact that you used data fro rainfall gauges close to the measured debris flow events, whereas Gartner, et al, and M1 model relied on a more scattered network.  In essence, they were forced to rely on widely spread input data while you have the advantage of comparing to local data.  I would like to hear your thoughts on this difference.  Clearly, it is better to you more local data, but what can you say about the need in other cases to rely on general data as in Gartner, et al.?

This leads to a bigger issue.  The Gartner et al. and M1 models rely on a dataset with limited accuracy, but smoothed over an area and temporally.  You are using more accurate data, which one would expect to produce better results, but in some ways compares apples to oranges.  For instance, earlier volume prediction models relied on StatsGo or other generalized soil data for things like Liquid Limit.  More accurate liquid limit measurements would not improve that model because they represent micro-data that is very different from the area-averaged data that the model used.   The area-averaged data is not very good, but simply adding focused accurate data does not improve the roughness of the model.  I want to make sure that your analysis does not fall in the same trap - providing more precision on a dataset that lacks in accuracy.  I think it would help to comment on this disconnect.

Line 195ff - even more important than the side of the canyon (north or south) and the proximity, is the elevation of the rain gauge, to account for orographic effects.  Can you include a comment on this?

Line 237 - was there notable deposition in the form of levees along the flow channel?

Line 255 - not clear to me how V1 differs from V2.  They both sound like subaerial deposition.  Perhaps distinguish between them better?  Also, I would like to see error or confidence estimates on your method of calculating the total volume.

Line 333 - I would argue that it is vegetation recovery and that sediment supply is not a limiting factor.  Perhaps include a reference or two, and maybe shift the focus more on vegetation?  Not to push my own papers, but you might check this one for an example of sediment supply independence:

Santi, P. and MacAulay, B., 2021, Water and Sediment Supply Requirements for Post-Wildfire Debris Flows in the Western United States, Environmental and Engineering Geology, vol. 27, pp. 73-85, doi.org/10.2113/EEG-D-20-00022.

Equation 6 - line 353 - wouldn’t you normally want an equation that uses the observed value on the right side ((x-axis) to predict the values on the left side?

Lines 406-413 - It is great to see this vegetation-specific analysis.  I’ve never seen this before.

Paul Santi

Excellent, comprehensive analysis of debris flow activity after a fire. Findings reveal how well USGS operational models for debris flow rainfall thresholds and debris flow volumes perform for this fire location. Writing is clear, and figures are a nice combination of quantitative data and photos. My comments are mainly minor points of clarification.

1. Equation 1: what are the values of beta, C1, C2, and C3? Are they determined for this fire specifically or are there set values used across multiple fire locations?

2. Line 158, median I15T with overbar - the overbar on I15T seems to imply a mean value? If the overbar indicates a mean, what values are included in the mean? If it doesn't indicate a mean, what does the overbar represent? The median is the median threshold value computed for all basins?

3. The first sentence in section 3.2 seems to imply that the rain gauges are mapped in Figure 1 - but they are actually shown in Figure 2. 

4. What is the precision of the rain gauge measurements (what depth per tip?)

5. line 277, "If there were multiple storms that were triggered" - do you mean multiple storms that triggered debris flows?

6. Figure 1: I would have found it helpful to see the locations of debris flows on this figure rather than on a separate figure. 

7. Figure 2b: I am not seeing deposition at the end of this debris flow track. For both b and c, consider adding an arrow to show flow direction

8. Figure 5d: this doesn't look like a fan - just in-channel deposition?

9. Figure 7: this is a nice figure but kind of a lot to take in. I am curious about how the erosion and deposition volumes compare at sites where both of those measurements were collected, but it's hard to evaluate that in this figure - could erosion-deposition comparison be pulled out as a separate figure or subplot? Then just show erosion in this plot? Do each of the debris flow volume points correspond with a rain intensity point in the bottom graph? Could be useful to have a volume vs. intensity plot, as an alternate way to visualize the data. 

10. Figure 8. As I understand the model, the threshold intensity to produce a debris flow will vary by basin. Yet here the symbols for "above threshold intensity" seem to be based on the median threshold across basins? Why not show the symbols based on the threshold intensity computed for each basin individually? I am also confused by the caption text "from the 11 rain gauges" - are there intensity values given for each gauge individually or are the values averaged across all gauges? 

11. Figure 9. The value of (b) is not clear to me, other than as a means to add the south canyon data. Visually the power law lines do not seem to fit the data well. 

This study uses a unique debris flow response dataset from a region with a paucity of post-fire debris flow information to test the USGS PFDF susceptibility and volume models commonly used in the western US for emergency management. The authors find preliminary support for the use of regional correction factors for the volume model and reveal potential drivers for reduced susceptibility 2 years following fire. Overall, the manuscript is clear in its objectives, well-written, well-supported, and presents important findings. I recommend some minor revisions to improve clarity and make some suggestions for presentation of results that the authors could consider.

Lines 116-117: Authors refer to Fig 1 when describing Quaternary-aged (“quaternary” should be capitalized, too) landslides but it is very difficult to see them in the hillshade as it is currently presented. Perhaps another figure in supplements if authors would like to show them, annotations directly on the plot highlighting the slides, or just no reference to Fig 1. Also, I don’t know which watershed is Devil’s Hole based on Fig 1.

Line 170: What are the models of tipping bucket rain gauges installed and their corresponding measurement resolutions?

Line 197: Shouldn’t 4 km² be 4 km (distance not area)?

Line 206: “see section 0” – there is no section 0?

Line 224: Were channel polygons hand-drawn or automatically extracted using a buffer around a flowline? Could be a nice detail to include.

Section 3.6: Recalculated dNBR post-recovery is a great idea and fits nicely with recent literature on quantifying vegetation recovery and its influence on debris flow susceptibility (such as Graber et al., 2023, link here: https://doi.org/10.1029/2023GL105101)

Line 334: Was there evidence to support sediment exhaustion of the channels such as downcutting to bedrock? Could be good to include.

Line 339: Not sure about using the term “nucleate” here and elsewhere (e.g. line 412) when referring to erosion/deposition in this context. I usually think of nucleation as a process that begins at one point and propagates outwardly, which I don’t think describes what’s happening here quite correctly. Could rephrase this as “initiate” or similar.  

Line 354-355: “changes between the debris flows and the lidar flight.” Clarify. Do authors mean to say: "changes to the debris fans occurring between initial deposition and subsequent lidar flights" or something like this?

Line 360: How were Coal Seam and South Canyon debris flow volumes estimated? Just curious if this could exert some uncertainty in a comparison of these earlier datastes to the lidar/fan based estimates for the Grizzly Creek PFDFs. It is promising that they roughly show similar area-volume scaling as the authors point out.

Fig 8B: There were no debris flows produced in 2022, correct? Maybe add in this language to figure caption since it is a bit ambiguous as is (no red stars = none correct?).

Fig 9A: The 1e5 scientific notation next to axis labels is too small. Consider blowing up this text or adding it alongside the units (e.g., 1e5 m³).

Fig 9B: I think the power law fits could be better coordinated with their respective point grouping colors – why have a Vp blue power law fit that does not match corresponding points with open red x’s and Vo black dashed fit that does not match red circles. Also, these power law fits do not visually seem to fit their respective datasets very well. Additionally, it would be good to provide an estimate of goodness of fit metric (R^2) as well as p-values (or confidence intervals) for regression parameters (prefactor and exponent) to provide some degree of confidence of these fits.

Additionally, for Fig 9B and the comparison between Coal Seam observed vs predicted (Line 360 earlier), do you see a similar ~4-fold overprediction from the Gartner et al. (2014) model? If it was close to this value, it further supports using this as a regional correction factor.