To the authors,

I thank you for replying to the two Reviewer’s comments. In reviewing your responses, I generated a number of requests myself, that I would like to see addressed prior to publication in GMD. Given the number and high-level nature of some of my comments, I classify this as further 'major revisions' that I would like to see robustly addressed.

Also, when replying to these comments (or Reviewer comments in general) if you could specifically note the specific changes you made to all comments (including specific or technical comments) in the Reply to Reviewers, that would be great.

####General comments####

First, and most importantly, I am concerned that despite wide use of the words ‘coupling’ and ’semi-coupling’, it appears to me that VAMPER and iLOVECLIM are not in fact coupled in the physical sense, at all. Suggesting that software coupling (in the sense that VAMPER runs in the same executable as iLOVECLIM) is equivalent to physical coupling or semi-coupling (where there is at least SOME two-way flow of information between VAMPER and iLOVECLIM) could be very misleading to general readers. Ultimately, semantics are very important, because software and physical coupling are extremely different things, and a software-coupled model shouldn’t be regarded in the same family as truly physically-coupled models. I have expanded on this concern in the Specific comments.

Furthermore, generally, I request a very detailed and fine-grained wordsmithing to remove or clarify a number of:

1) vague or grammatically incorrect statements, for example the following direct quotes:
“In fact, Koven et al. (2013) recently reported on the Coupled Model Intercomparison Project phase 5,”
“The modeling run not including the effects of snow cover overestimate the present permafrost extent.”
“since,as just mentioned,”
2) subjective or judgement-based words, for example:
“fortunately”
“overall”
“reasonable”
“suitable”
“cumbersome”
“well-known”

####Specific comments####

P1L1: Title: I have some issue with this title, given this manuscript does not actually present a coupled permafrost-climate model (by the author's own admission). I request a title change to accurately reflect that this manuscript describes steps towards full coupling, but not full coupling itself. In addition, the title change should reflect that other perhaps critical processes (namely hydrology) are not included in the model. For example: "Progress towards coupling of the VAMPER permafrost model within the earth system model iLOVECLIM (version 1.0)”. This concern is reflected in some of my following comments as well.

P1L8: “has been enhanced for coupling”: following from the above comment, I suggest rewording to “has been enhanced in preparation for full coupling”.

P1L11: I feel that the term semi-coupled is misleading in the context of this manuscript. The term semi-coupled refers to at least SOME coupling being present (as an arbitrary example, communicating albedo changes but not surface heat fluxes from land to atmosphere). I request a re-terming of ‘semi-coupled’ in this manuscript, to accurately reflect the state of the software being presented. Actually, it appears that practically VAMPERS is completely uncoupled in a physical sense. Rather, it is ‘software-coupled’, in that it simply runs in the same model executable as iLOVECLIM. For example, with some infrastructure changes, I suspect the authors could in theory first run ECBILT-1, then run VAMPERS with ECBILT-1 output, and obtain the exact same results as shown here.

P1L35: “In other words, the effects of (changing) permafrost are not fed back to the climate”: please specify for the non-permafrost expert reader here what these effects actually would be (presumably, ground heat flux, which is noted later in the manuscript).

P2L29: “It should be noted that additional coupling mechanisms are possible between iLOVECLIM components and VAMPER, which include hydrology and the carbon cycle, but are not yet implemented at this time.” The effect of these omissions should be explicitly noted as caveats, so that users of the model are aware of these, and also, so that reviewers of future scientific publications based on the model described here can clearly take these caveats into account.

P3L7: Ensure that the reader doesn't leave with impression that Lawrence and Slater, Dankers, and Schaefer are only global-scale permafrost models (for just one example of additional work, see paper by reviewer Chris Avis: 10.1038/ngeo1160). Generally I think significantly more than 2 lines describing historical efforts in global permafrost-climate modeling are necessary, to put the present work in a historical context.

P3L9: Please provide references for previous works that applied one or the other approach.

P3L37: By definition, parameterization schemes are not cumbersome, but are meant to be lightweight representations of unresolved physical processes. In fact, models like iLOVECLIM are by their very nature parameterization-heavy (in lieu of explicit simulation of small-scale processes). Thus, this statement is somewhat confusing.

P4L6: “Since we are also interested in this timescale, we originally employed the same reasoning: relying on large-signal paleoclimatic changes”. This statement is somewhat unclear.

P4L13-P4L18: Suggest these few sentences are reworded to increase clarity and flow.

P5L3: It is not immediately clear to naive readers that decreasing the time step is the model development that causes appearance of an active layer. It seems the authors sometimes to refer to the same model development as 1) decreasing time step, and 2) simulation of an active layer. It could be beneficial to clearly specify that implementing 1) causes 2) (if this is indeed correct).

P5L23-P5L26 This paragraph seems to belong in the Results section.

P5L29: “the surface offset” is unclear to the average reader - suggest a better description of this.

P7L1: “The following is a stepped description of the snow algorithm for the ECBilt-1 VAMPERS semi-coupling”: perhaps reword something like “The following is a stepped description of the snow algorithm to generate a VAMPERS snowpack from ECBilt-1 precipitation”

Section 2.2.1: Despite iLOVECLIM being described elsewhere, I strongly suggest a brief description of the model components other than ECBilt-1 here (basic physics, resolution, etc.), to make this manuscript more self-describing.

P7L29: Specify exactly what ‘air surface temperature’ is. Is it equivalent to 2m air temperature, or the temperature right at the boundary layer (however that is simulated in ECBilt-1).

P7L37: “In this case, the snow surface temperature is taken to be the air surface temperature .” This is somewhat confusing. For example, a reader would ask “isn’t there still an air surface temperature, even if snow is present? How can snow surface temperature replace air surface temperature?”

P8L2: “phase change capability”: echoing reviewer comments, I request that this capability be explicitly described in this particular manuscript, since it is presumably an important aspect of the model.

P8L11: “This system is often referred to as a bucket model in previous text.”: please reference these previous texts.

P8L12: “As of currently, this bucket model, which is the surface hydrology in iLOVECLIM, is not coupled to VAMPERS. It would be a sensible next step to connect the active layer with this bucket model”: so does that mean that two land models run simultaneously for a given grid cell: VAMPERS, and the bucket model? And the bucket model is the model that provides feedback to the climate? If so, this is important to note.

P8L14: “The results presented in this current work is only a function of performing semi-coupled experiments”: this is an unclear statement.

Section 2.2.3 and Section 2.2.4: These seem out of place in the ‘Coupling’ section; do they rather belong in the description of ‘VAMPER model enhancements’?

P9L30: “This configuration, therefore, allows only the examination of the iLOVECLIM model to reproduce current permafrost extent and depths as function of the currently established climate of the iLOVECLIM model.”. This statement has an unclear meaning, and is perhaps self-obvious?

P9L38: “until approximate equilibrium between ECBilt temperatures and VAMPER(S) model is reached”: the word equilibrium implies coupling. Given that the present setup does not have full coupling, perhaps a better phrase would be “until VAMPER(S) equilibrated under equilibrated ECBilt temperatures”.

P10L19: Suggest replacing “how well the” with “the extent to which the”

P10L22: Aligning with previous comments: suggest replacing “ECBilt-VAMPER semi-coupling” with “VAMPER driven with ECBilt forcing”.

P10L23: “Swing of inaccuracy” is unclear

P10L23: “This swing of inaccuracy is at least partially due to simply attempting to match results from a low resolution grid to spatial coverage of much higher resolution”. I disagree with this statement, since if it were true, I think both VAMPER simulations should show the same bias (since they are both on the same resolution grid).

P10L25: “In addition, we expect some inaccuracy since we cannot parameterize the snowpack characteristics and more importantly, the nature of the snowmelt.” But, isn’t capturing the first-order nature of snowpack characteristics and/or melt the point of including a snow model in VAMPERS? Please clarify.

P10L27: “As opposed to our generalized approach described earlier” and “recognized discrepancies in generalizing snow model details,”: please specify what exactly these statements refer to.

P10L31: “However, as long as the VAMPERS model is doing a reasonable job, we contend it is a better option over merely applying artificial offsets or assuming none at all since snow plays a critical role in the ground thermal conditions and should be represented.”: I understand this argument, but think that it could be reworded in a more scientific way. For example, what within what bounds do you define ‘reasonable’?

P10L33: “Further, with the snow optiont, changing precipitation patterns that are often the byproduct of a shifting climate would otherwise have no effect on the subsurface thermal conditions.” There is a typo, and also, don’t you mean “withOUT the snow option”?

P11L1: Again, the ‘surface offset’ needs to be clearly defined, perhaps each time the statement is used, for non-specialist readers.

P11L23: This is a great paragraph that should be combined more linearly with the paragraph starting P10L21.

P12L3: “(> 300 m) and some very underestimated (>700 m).” Should one of these ‘>’ perhaps go the other way?

P13L19: “Overall, Fig. 11 illustrates that ECBilt-VAMPERS does a reasonable job of predicting shallow subsurface temperatures since a majority of the points fall near the 1:1 line. This result, therefore, supports the notion that the preindustrial climate is well represented by iLOVECLIM.” I suggest calculating a regression coefficient, so that you can say something quantitative about the comparison, instead of the subjective statement ‘near the 1:1 line’.

P13L39: “Due to minimal parameterization”: do you perhaps mean ‘minimal complexity’?

P14L5: Suggest renaming “Next steps” to “Future development”

Figure 3: Is ground heat flux actually passed back to the climate, in the current model? If not, this figure is misleading, since it implies physical coupling. Also, what is the dashed white line?

The manuscript is much improved; I thank the authors. Please find further technical corrections inline within the attached PDF file (comments included using Acrobat Pro - please inform me if you cannot see these for some reason). These do not require a re-review by myself or reviewers prior to publication.