The manuscript presents a modeling study of permafrost temperatures and ground ice with and without (lateral) drainage of water from the model domain. The hypothesis is that well-drained blocky terrain may result in colder ground conditions and the subsurface drainage of water should therefore not be ignored in permafrost models. The model experiment is nicely designed, and the results clearly show the impact of drainage in different ground substrates. The results should be of interest for anyone interested in permafrost processes, especially in mountainous and blocky terrain. While I find the study interesting and the modeling results valuable, I have some concerns about the validation procedures and the general structure of the paper and to some extent the setup of the simulations.

Major comments:

1. The idea of the model experiment – testing the impact of drainage on the ground thermal regime – is very interesting and the approach seems sound and straightforward. However, the model validation and the links to the two sites in Norway are weak. For one of the sites there are observed temperatures available to compare with the model results but these data are not used to validate the model results in a robust way. Instead, validation is only carried out visually and no objective statistics are used to evaluate the model results. For the other site, there are no data on ground temperatures but only observed ground surface temperatures. The validation and comparison to observations from the Norwegian sites thereby adds next to nothing to the modeling study. My suggestion would be to either remove these sites from the manuscript, to add some statistical analysis based on the available ground temperature data, and/or to find a rock glacier site with available ground temperature data which could be used for a real validation of model results.
2. The methods section would benefit from restructuring and extensive editing, because it is rather confusing as currently written.

• Model description. It would be helpful to start the model description with a presentation of the mesh used: is this a vertical column (1D) mesh? What is the thickness and cell size of the domain? Some of this information is available in other parts of the methods section, but starting off by clarifying such basic facts would help the reader to envision the model setup.
• Snow model. A list of processes included in the model is provided, but there is no explanation of how these are represented in the model. If these are explained in previous publications, please clearly refer to those for the specific processes listed. For example, “the physical effect of wind drift on the snowpack” is included, according to the text, but later it is stated that the redistribution of snow by wind is phenomenologically represented using the snowfall factor. So, what effects of wind on the snowpack are actually included? Does the snowfall factor allow for variable distribution of snow on the domain, or is there also a redistribution of snow over time?
• The description of the different model runs (validation, equilibrium, and transient) is rather confusing and split up in different sub-sections of the methods description.I would suggest to clarify the purpose of the model runs and explain them in a separate sub-section. The validation runs are runs validated based on visual fit, which is not a very robust metric for model validation. It seems like the purpose of validation runs was to determine (calibrate?) the proper sediment stratigraphy for each site. But how were other parameter values determined? Table 1 presents some physical properties, but not e.g. Kh. Snowfall factor is for some reason tested in the equilibrium runs (why?). In general, the purpose of testing the snowfall factor sensitivity is rather unclear to me. The outcome of this test is not mentioned in the abstract, but the current text does not suggest that this is pure model calibration.  L230 suggests that snowfall factors were also tested in validation runs, but this is not mentioned previously.
• The forcing data used for spinup and simulation runs are also not presented in a clear manner. The spinup and initialization procedure and data vary between the sites and the runs and the motivation for the choice of spinup periods is not always clear. Perhaps a table presenting the basic details of each run would be more helpful. Also not clear in the current text: Is forcing data downloaded and downscaled for the two sites (e.i., two separate series of data)? What is the time resolution (daily? monthly?) of the data used for simulations and for validation of simulation results?

Minor comments

L155-164: Seepage face is at atmospheric pressure, but does that mean that the seepage face is located at Zwt? Could that sentence just be removed and eq. 1 explains all? Where are Kh values from? Eq. 1 mirrors Darcy’s law but values seem made up, and dz is based on the grid z. Explain the rationale behind this model and why it was chosen. (alternatively, if Kh and dlat are unknown, these two parameters could have been replaced by one single parameter.)

L282: But table 1 does not include any tested case with a porosity of 0.2!! Either the methods section does not completely describe what was done in the validation runs or there is a typo here?

L296-312: The model results suggest that the differences in model setups have very little impact on MAGST and that overall there seems to be a slight cold bias in the model. What then does this validation add to our understanding, if the purpose of the study is to investigate the impact of drainage on ground temperatures and ground ice?

Section 4.1: This section would greatly benefit from some objective measure of model fit instead of just some conclusions from visual inspection of plotted temperature curves.

Section 4.2: It is not surprising that snow insulates the ground from wither cooling, generally leading to higher ground temperatures. More relevant than a comparison of snowfall factors, would be to check if observed snow depths and ground temperatures are recreated with the forcing data used here. When reading this section, I get curious about how much snow is generally observed at these two sites and does the forcing data capture the range of observed snow, or is a snowfall factor scaling needed to “correct” the forcing data.

General comments:

The authors have used CryoGrid simulator to perform three synthetic experiments based on soil stratigraphies with/without drained soil conditions to investigate whether well-drained blocky soil conditions (referred to as blocks only, drained in the manuscript) can lead to colder soil thermal conditions in permafrost regions. One-dimensional simulations with dynamic water/ice were carried out using an empirical subsurface lateral flow representation. The results showed significant sensitivity to wind-driven snow distribution and soil moisture conditions (lateral flow). While the effects of snow, soil stratigraphy, and soil moisture on permafrost thermal regimes are well-known, I still believe the study, once revised, will make a nice manuscript for permafrost researchers, especially mountainous permafrost audience. However, I have some concerns with the manuscript in its current form.

Major comments:

• Describe in more detail the rationale for each scenario.
• The effects of drainage, snow, soil moisture, etc. have been studied for other permafrost sites, but may not be particularly for mountainous regions, the physics does not change from low- and moderate-relief regions to mountainous regions. How does this study connect with existing literature that studied the effect of snow, soil moisture, etc. on permafrost thermal regime? A better referencing is needed.
• I would strongly suggest rewriting/reorganizing the Introduction (also Methods) section. The authors have done a good job in providing detailed background; however, it needs to be organized so the reader can follow it. Especially, I found a disconnect between the driving mechanisms and how this work is going to address those. The last two paragraphs in the Introduction section provide a slight background but that needs to be expanded.
• There are lots of short (4-5 lines) paragraphs throughout the manuscript, and probably not needed and can easily be merged.
• The example of CryoGrid processes provided (lines 134-135) is highly abstract. I am not expecting to provide all the details, but at least some details for a quick reference.
• Paragraphs in the abstract? does the journal allow it, usually not seen/recommended?
• There are many places where authors need to be specific. for example, L135: "Likewise, different process representations for the seasonal snow cover can be chosen." this needs to be expanded to mention specific snow processes rather than "different processes"
• Sensitivity to computational domain depth and bottom boundary condition is needed. Provide details that why the domain depth of 5 m and the prescribed geothermal flux were chosen. Describe if the results are sensitive, they will be, to the domain depth and bottom boundary conditions. I understand this can get complicated but at least mention it in the text.
• A schematic of the model domain with boundary conditions, soil discretization, soil layers, etc. can help better follow the results.
• Figure 7 shows results for transient runs (1951-2019). What is the air temperature gradient (or increase in the mean annual air temperature) over this period? and did the authors try to run detrended data to isolate the effect of temperature increase? Otherwise, this effect is not due to soil stratigraphy and drainage only. And since the porosity in the "Blocks with sediment" case is 25% (half of the two other cases), more degradation is not unexpected. Also, what caused the patchy low ice content in some of the subplots in Figure 7 (for example, top/bottom right)?
• The authors kept referring to "at depth 5 m". Please draw/highlight surface elevation (datum) in the figures. For instance, what elevation would be “5 m depth” in Figure 7? It is not clear. 
• The study is performed on an idealized column domain with a fixed surface datum. So, it would be easy for the reader to have the vertical scale in "depth" [0,5], instead of elevation, which I don't think is needed unless I am missing something.
• The focus of this work is to study the effect of soil stratigraphy (with drained/undrained conditions) on the soil thermal regime; however, no sensitivity study is performed on “soil stratigraphy”. For instance, how the blocks only scenario with porosities of 0.6 and 0.4 will affect permafrost conditions?
• I would also suggest some of the sensitivity-related results, for example, section 4.2, to be moved to a supplemental document, and focus more on what is new here.
• Also, in section 4.1 (line 307) authors mentioned all simulations used a single snowfall factor, however, later in section 4.3 they use different snowfall factors. This needs more explanation. Providing a table (which may not be in the main manuscript) listing all scenarios (other than the three listed in Table 1) will help the reader better understand it otherwise it is hard to untangle.
• What type of soil retention curve is used in the study? Moreover, how sensitive are the results to hydraulic conductivity? I didn’t find any mention of the role of spatial gradient (steepness). All these factors significantly impact drainage. Discuss.

Minor comments:

Please see the annotated manuscript (attached as a supplement) for more comments.