Review of "Sensitivity of iceberg drift and deterioration simulations to input data from different ocean, sea ice and atmosphere models in the Barents Sea (Part II)" by Herrmannsdörfer et al. 

This is the second of a two-part manuscript. I have reviewed Part I and recommended it to be merged into Part II. 

The Part II paper is the main body of results from the study, presenting Lagrangian iceberg simulations. Similarly to Part I, the focus of the paper is to exhibit the differences between iceberg simulations when switching one source of ice-ocean or atmospheric input data, but without an indication of the respective accuracy of each data sources. 

Part II simulates synthetic icebergs from their calving location until they are mostly melted, not observed icebergs. One example of simulated trajectory is selected and discussed in more details. This is an important information that should already be given upfront in the abstract to set the scene of the study. 

The dependency of Part II upon Part I is less strong than I originally thought. I have only counted three facts that could be included in this manuscript in replacement of some overly long technical parts (see below and see the Part I review). 

The model description is given in Annexes, although it is not clear whether the model is identical to the previous Monteban et al. study or if it has been modified. Either way, the authors should reconsider if the annexes need to repeat the previous paper or simply refer to it. 

The results are novel and interesting, and all the more important that the related literature is now quite outdated. However the Part II paper often lack a reflection on previously published results. This aspect should be strengthened. 

The discussion of the results could be reorganised to set the higher priorities first: if the location of the sea ice edge is the largest cause of divergence between the iceberg trajectories, then the authors could argue that the efficient assimilation of satellite sea ice concentrations is the priority issue when simulating icebergs and down-prioritise the technical issues on the search radius. 

The conclusion lack practical recommendations for other practitioners of iceberg modeling. Seeing that the statistics are "surprisingly similar" for the Topaz and the Barents-2.5 models, but knowing that fewer years of reanalysis are available from the Barents-2.5 model, I would be tempted to use the Topaz reanalysis all along and use the 30 years of data for the sake of statistical significance. Further, I would expect the conclusion to bring up the following topics: how important is the duration of the reanalysis, the consistency of the input data (with the example of the modifications of the Barents-2.5 model), the differences from / the commonalities with previously published iceberg simulation studies. 

The language should be improved with a better choice of vocabulary, logical transitions, and punctuation. 

Part II overall represents an interesting and valuable contribution to the field, although some improvements would be necessary, and I can recommend it publications after major revisions. 

Detailed comments:

- l6: "are sensitive". I would expect the abstract to indicate a degree of sensitivity, it is very or moderately sensitive? 

- l12: "surprising similarities". A word of explanation is necessary in an abstract: why are they so similar? Otherwise the reader is left on an apparent contradiction with the sentence on the sensitivity of the model. 

- l19: "ice features" -> "icebergs"

- l20: Grounding is missing here, and only mentioned far down in the manuscript. 

- l41: the multiplication 4.2603.7 is mysterious, but the exact number is not necessary in an introduction. 

- Table 1 could contain more information in additional columns and remove Column 1 (objective) which is not followed up in the text: for example, the duration, horizontal resolution, frequency and presence of tides. 

- l62: Explain why 2603*7 or remove if the exact number is not important yet. 

- l66: the acronyms ERA5 and CARRA have already been introduced. 

- l71: Why do you need geostrophic currents when the models provide surface currents? And do Slagstad et al 1990 provide data relevant for your period or only the general formula for geostrophy? 

- l72: Indicate the version of the IBCAO bathymetry used. 

- l79: "Gaps of [...] few days", how many days? Winds are unlikely to persist for more than two days in the Barents Sea. 

- l82: Many readers may be unfamiliar with the area, move the general map A1 from the annex to this location.  Hopen, Bear Island and Storfjorden should be indicated as well and a few isolines of the ocean bathymetry. 

- l85: "Empirical relations". Indicate the sources of these relations. 

- l92: How the pressure gradient force relates to geostrophic currents is not clear. Do the authors mean to account for the sea surface slope, with a force downslope rather than turning around the positive/negative anomalies of sea level? In this case, which data is used for the sea surface heights? The reader should not have to read Slagstad et al. 1990 to understand this sentence. 

- l94: "Coefficient set to zero". Indicate "no added mass" because the coefficient has not been defined here. 

- Grounding is missing from Section 2.4 and first mentioned on l.412. 

- l107: "for one time step": Time should be irrelevant for spatial interpolation. 

- Section 2.5: Distinguish assimilation from forcing. 

- Section 2.5 has too much details, and Figure 1 is not necessary since the ocean currents are notably inaccurate near the coast, as noted by the authors much later. The takeaway from that section is that inputs near the coast are fetched from a nearest neighbour, which can be several grid cells aways. As a reader I am ready to accept that there are too many uncertainties near the coast and that as long as icebergs are seeded to the ocean, the following simulation is valid. 

- Section 3.1 is similarly too detailed and Figure 2 can be removed without affecting the following results. 

- Table 2: Why not indicate total mass loss? The relative contributions can be misleading when one model melts icebergs much faster than the other. 

- l201: What are "seeding characteristics"? The number or the size of icebergs seeded? An explanation why icebergs from some glaciers last longer than others would be interesting. 

- l223: In which projection is defined the output grid? 

- l230: Are there other results in the literature to support these results? 

- Figure 4 is too small for the paper version, but by blowing up on screen, the Barents2.5 shows more fine features than Topaz, which may be related to topographically steered currents. A few isolines of the topography (50 m or 200 m) may help reading this map. If Figure 4 is made bigger and more readable, Figure 5 can be removed as it seems redundant. 

- Section 3.5: The notion of iceberg extent was introduced before (Keghouche et al. 2010, possibly earlier), is the same definition applied here? 

- l251: Contradiction between the two sentences. Replace "all directions" by "most directions"? 

- Figure 6 is not colourblind-friendly. I cannot tell the two CARRA simulations apart. Perhaps make them dashed lines. 

- Section 3.6: add "simulated" in the section title for clarity. 

- Table 4: Sea water surface velocity is the norm or the v-component of the velocity? 

- Table 5: Why are there two numbers in the last three columns? 

- Figure 9 panel a) add the horizontal line at 0 degrees to indicate the melting temperature of glacial ice. Panels d-e-f are not very informative as they show the velocity modulus but not the direction. Feather plots could be more intuitive. 

- Figure 9 g-j continues on a different page, so the related part of the caption should follow as well. 

- Same figure: the "delta" and "abs" on the vertical axes are mysterious, I think they can be removed. 

- l330 sounds rather dramatic about the data availability near the coast, but pragmatically the detailed conditions near the calving front are very uncertain anyway. Iceberg simulations may be stuck near the calving front (and need an initial push as you do with data interpolation) but as soon as they have escaped, the iceberg trajectory is valid. 

- l333-340: Similarly, these considerations seem out of place, unless there is an implicit aspect that I am missing. From the results presented above, the resolution is of lesser concern than the sea ice edge disagreements between the models. The discussion could thus be shortened if it was restructured to highlight the mot important topics. 

- l365: Although I expect this sentence to be correct, I fail to see this increased deterioration in May-June in Figure 9h. 

- l372: The difference of temperature between Barents-2.5 and Topaz only comes here. The differences in sea ice variables come further on. This means that the relevant elements of the Part I paper can be inserted before section 4.2. 

- l375: The Barents-2.5 may well have too cold SST, but a comparison to satellite data would hammer the facts. 

- l381: The tidal loops should be mentioned earlier when the datasets are introduced. I cannot see tidal loops in Figure 7, are they too small for the figure or are the tides small in that area?  

- l385: The tidal component is essential for iceberg density and extent. This statement is not obvious from the figures, nor followed up in the rest of the paper. Could it be elaborated? 

- l391-394: I miss the whole idea of this paragraph. Please rephrase to clarify. 

- l397: "may indicate an impact". Please introduce the next discussion in a more direct way. 

- l408: This sentence is inconclusive about the mobility of icebergs near the calving glaciers, their grounding and data interpolation near the coast. Keghouche et al. 2010 computed the incidences of grounding, could the authors compute a similar map or a blowout of the averaged current vectors near a calving front to discriminate which effect is most important? 

- l425: "might increase": this sentence is blue sky to me. More sea ice means less melting or wave erosion, and additional sea ice stress, so the authors could be more assertive about the difference between the models. Also mention which model is more realistic.  

- l430-431: This sentence is too complex. Do you mean that higher resolution winds follow better the orography? 

- l432: The coarse resolution currents are extrapolated near the coast, irrespective if the currents are on-shore or off-shore, so I don't understand why the icebergs cannot drift close to the coast in Topaz.  

- l445: "iceberg drifted": it is singular because there was only one iceberg? 

- l448: About iceberg extension and environment forcing, Keghouche et al. 2010 did show the relationship between iceberg extent and the wind patterns. Does that relationship still hold in your study? 

- l451: I don't understand what is meant by varied forcing reproducing the variability. Rephrase. 

- l466: This idea is repeated. 

- l470: Has the influence of the seeding mechanism been accounted for in the presented results or only mentioned as a warning to the readers? 

- l472-477: It is well known that Lagrangian trajectories diverge over time (there is a vast body of literature about dispersion in the ocean surface since Okubo 1971, see also Koszalka et al. 2009 for a geographically closer example), the first part of the paragraph is correct, but the end is unrelated: one single example of iceberg trajectory does not say much over averaged statistics (See Figure 4 for example). 

- l475: "Similar initial conditions". Are they similar or strictly identical? 

- l479: Use a more precise vocabulary: The iceberg trajectories diverge. 

- l483: Here again, the end of the paragraph seems unconnected from the preceding argument.  

- l496: "can be seen", are you referring to Figure 9? 

- l489: Sentence unclear. It seems to be rephrasing the same idea, but with unclear words ("derive"). 

- l491-495: This paragraph sets the context of the results and should be moved to the beginning of Subsection 4.4 

- l499: You found that the results were not so sensitive to the change of atmosphere reanalysis, do not state the opposite in the conclusions. 

- l500: Surprising similarities: Please elaborate on how the differences in the forcings cancel out in the final statistics. 

- l502: This statement is contrary to previous statements that average sea ice thickness and sea surface temperatures cause most of the differences, not so much the resolution of the data. The resolution only seems to be an issue near the coast. 

- l503: the enumeration lacks logic. This paragraph is restating previous results instead of concluding on the take-home message. 

- l515: Did we see that the iceberg looping increases their density? 

- l570: I was hoping to understand what "pressure gradient" is meant in the Annex, but to no avail. The confusion between ocean currents and geostrophic currents is still puzzling and the notation Fc,p rather indicates that the pressure gradient is the same as the Coriolis force. 

- l572: Drag coefficients are different from those in Table 2. 

- Eq. A15. Indicate the reference for this equation. 

- Section A5. It is not necessary to have a section header and one line there since the Tables are referenced in the text. 

English corrections: 

Systematic errors:

"Following," -> "Thus"

"Note, ": No comma. 

"Thereby, " 

"For more," -> Furthermore, 

"Even though," ->  "Even then, "

"the differently forced simulations" -> using different forcings. 

"Relevant sea ice": the word "relevant" can often be removed. 

- l2: "bridge the gap" sounds odd for a lack of data. "Filling" is more adequate. 

- l9: "Deviations" caused "differences". The authors probably wished to avoid a repetition here, but making the sentence puzzling because the inputs fields do "differ" more than "deviate" (from what?), causing different (diverging) iceberg trajectories. 

- l10: the impact "of" sea ice variables is especially large. 

- l15: "comparably" ... to pixels in satellite data? 

- l16: "inhabited" -> trafficked, navigated. 

- l21: "Forced convection" -> heat convection? 

- l25: "uncertainties resolves from" -> result from 

- l26: "Diverse bathymetry": Shallow? 

- l29: "physical description": physics. 

- l32: "results shall improve" -> are intended to improve. 

- l42: How "exceptional" is that iceberg trajectory, exceptionally long?  

- l44: Forcing a model is not data assimilation (systematic confusion, also commented in Part I). 

- l46: "difference of" -> between (twice). 

- l78: "translated" -> rotated. A mere translation would be very bad.

- l201: "from source" ... to source. 

- l210: that the drift distance is longer with Topaz is surprising, I would have expected that the tidal loops would have made them longer with Barents2.5. 

- L212: The sentence on atmospheric forcings is repeated. 

- l240: "a larger proximity" -> further off 

- l261: "similar scale" -> by the same amplitude? 

- l350: "bridge de gap that is the lack of [...] observations". Bridging is not the right verb here. 

- l355: "contrasting". I don't know what is meant here. 

- l368: "occurring sea ice" -> remove "occurring" 

- l383: No comma after "indicate".

- l402: "life" -> "live". 

- l420: lager -> larger

- l433: "decreased" -> low 

- l446: "atmospheric-forcing". No hyphen, the forcing is also ocean and sea ice. 

- l488: less days -> fewer days 

- l549: "It's" -> its. 

- Eq. A6-8: Is the melting M expressed in meters? 

- l559-560: These lines are repeated in the main text. 

Typos: 

- l.7 add a comma between resolution and bathymetry. 

- l38: all four citation should have - human - authors and years. 

- l90: "Suite" -> "suit" 

- l179: Is M_fb the same as M_fw? 

- Table 4, the column named v_ai does not correspond to the caption. It seems the water and sea ice velocity columns have the wrong name. 

- Figure 9: use subscript consistently for sea ice thickness h_si. 

References: 

Koszalka I., J. H. LaCasce, and K. A. Orvik. Relative dispersion in

the Nordic Seas. J. Mar. Res., 2009.

Okubo, A. (1971), Oceanic diffusion diagrams, Deep Sea Res., 18, 789–802.