First, I wish to acknowledge and recognise the effort of the authors for choosing to write and submit a scientific paper to a journal in a language which is not their own.

The subject is relevant and of potential interest to the readers of Natural Hazards and Earth System Sciences. As a researcher who utilises landscape evolution models (albeit in a very different environment) I looked forward to reading this paper.

However, there are multiple major issues with this paper which means I believe this paper requires major and significant revision before it is suitable for publication. Individual points requiring attention are too numerous to individually list, but the key issues are as follows:

First and foremost, is the lack of proficiency and fluency in the use of the English language and grammar, which is consistently poor throughout this paper. This makes it very difficult to comprehend the contents of the paper – for example, it is not clear what the methods were, or how sets of parameters used in simulations were obtained / derived. Because of this,  it was also not clear how the results were obtained and what they actually represented, and consequently, whether the resulting discussion and conclusions could be substantiated or supported.  Overall, while the aims and objectives could be understood, it was not easy to determine if they had been met. Unfortunately, the authors unfamiliarity with the English language meant that too many sentences were variously incomplete,  made no sense , or utilised inappropriate or mis-spelt words.

Some comments and suggestions for improvement:

• Rather than try to describe the background to the CAESAR model and how it works themselves, I believe the authors could more clearly and succinctly acknowledge this by referencing existing publications which describe this ie Coulthard et al 2012.
• As currently written, descriptions of specific parameters and methods used in this study and the scenarios modelled are poorly described, or not described at all – for example:
  • a table of parameters lists values used in simulations, yet there is no explanation of how or why the values in the table were selected or used  – for example those representing vegetation parameters (shear stress, age to maturity, proportion of erosion) - were selected and utilised in model simulations for the scenarios in this study.
  • The authors do not explain why they selected some of the parameters ie why a specific sediment transport equation was selected. Depending on the sediment transport equation applied, very different model results may occur.  
  • It is not clear what rainfall data was used in the simulations – whether different sets of data were used for different scenarios, or one set was applied across all scenarios. The text about the downscaling rainfall data is simply confusing and does not address this.
  • As written, it is not clear how results are obtained or substantiated from the methods. The authors make assumptions about the ability of the model to erode that are not supported or substantiated by any evidence. Specifically, the authors describe how they have attempted to incorporate levees and dams into simulations by simply increasing the elevation in certain areas and not changing other parameters such as particle size. While this reviewer agrees it may temporarily reduce flow, in the longer term, this may well lead to increased erosion in other areas around the sides of the dam / levee.
  • The authors have demonstrated a poor use of figures and tables to support their results. Specifically,
    • figures variously lack scales or annotations to indicate where the areas of erosion or deposition are (eg figure 5), or where other features (such as dams) referenced in the text are located (eg figure 1).
    • Some figure captions do not make sense eg figure 2 does not clearly show any chart or process for generating the bedrock DEMs. Some figures do not appear to contain the information described in the text.
    • Tables are present in the manuscript which are not referenced in the text; different tables share the same number (eg there are 2 tables labelled as table 2); and some tables do not identify what the units in the table represent.
  • Finally, the authors use of referencing is poor and inconsistent.

Before they attempt to re-submit this paper, the authors must have the manuscript thoroughly  and comprehensively proof-read by a reviewer who is proficient in both reading and writing in the English language- and familiar with the standards of scientific publishing.  Once this has been done, it may then be possible to assess whether the authors have been able to address the aims and objectives of the paper, and if the results and conclusions can be substantiated. At the moment it is not possible to do this. In short, this paper has the potential to be an interesting and informative contribution to the science community but currently requires a significant and large amount of work (it effectively needs to be re-written) before it could be considered for publication.

This paper uses the CAESAR-Lisflood (henceforth, CL) model to simulate different management approaches to debris flow management in an earthquake-prone region of China. It looks retrospectively at a three-year period, comparing the real-world management approach to a ‘do nothing’ scenario and an enhanced management scenario. The results showed that the current management approach is successful at constraining debris and reducing downstream risk in comparison to the ‘do nothing’ scenario, and that the enhanced management scenario would have provided even greater benefit. I consider this to be a valuable contribution to the literature, demonstrating how LEM tools like CL can be used operationally and for decision making purposes, and is of interested to the readership of the journal. I recommend publication after minor edits to address the issues I outline below:

In the second Table 2 (there are two) on page 22 the sediment conservation ability in the source area increases from 0.5 to 0.55 between the PP and EP scenarios, yet I don’t think there are any additional interventions in this area (levees and vegetation are in the deposit area). I think this may be an effect of the way the authors have applied the spatially varied “m” in the model. In UP and PP, the “m” has a global value of 0.008 and in EP the vegetated areas are give a separate value of 0.02. It isn’t stated by the authors but I believe their rainfall input is catchment lumped (please could the authors confirm). My understanding of CL is that for a lumped input it will average all the “m” values and create a single lumped input from it, in this case making the input for the whole catchment less flashy. Alternatively, the authors could specify two separate rainfall input areas, one for the vegetated area and one for the rest of the catchment, in effect making two hydrological response units (HRUs) for the model, each with its own input based on the local “m” value. I don’t think this needs to be done for a revised manuscript as I doubt it would change their conclusions materially, but it should at least be acknowledged.

Further on “m”, I concur with the comments from Jorge that where possible the value should be calibrated against gauged data. If this is not available, basing the value on land cover, as the authors have done, is reasonable. However, the authors are using downscaled hourly rainfall, not observed hourly rainfall, so any calibration would need to account for this.

I also concur with Jorge’s comment on spin-up period. There are no details in the manuscript and it would be helpful to know. In this case, where much of the eroded material is fresh and loose, it could be argued a spin-up might actually be counter-productive in this instance.

The downscaling of the rainfall to hourly is really important (as shown nicely in Figure 3). Sorry to push one of my papers, but Coulthard and Skinner (2016: https://doi.org/10.5194/esurf-4-757-2016) provides some analysis of why and it would be useful to refer to this here. Unfortunately, I found the description of how this was done not clear – please could the authors revisit this description so it is easier to follow. It would be useful to also known the spatial resolution of the rainfall product that was used and the spatial resolution it was applied to the model with (I assumed it was lumped).

A verification of the model outputs for the PP scenario by comparing them to real-world observations would strengthen the analysis of the paper. For example, Figure 10 and related discussion could be included within the results as a form of verification for the model outputs.

I would recommend that the authors include in the discussion notes on how the outputs of this analysis could be used – ie, why is this work useful. Is the intention that these modelling approaches will be used in the future to design debris-flow management schemes and help to inform decision making, for example?

On the language in the manuscript, I found the vast majority of the manuscript well written and easy to follow. There were a few instances where phrasing is not quite comfortable and I think some editorial guidance would be sufficient to improve these. Some in-line references contain initials and these should be corrected.