See pdf.

In this paper, the authors quantify projections of mass loss from individual glaciers around the Antarctic Ice Sheet out to 2100, with the goal of (1) identifying the glaciers most vulnerable to significant mass loss, and (2) quantifying the dominant sources of uncertainty in these projections. The authors identify Thwaites Glacier, Pine Island Glacier, Totten Glacier, and Moscow University Glacier as being the regions most likely to experience significant mass loss in the next century, and they identify the significant role that the choice of ice flow model has in contribution to uncertainty. I found the paper to be clear and the experiments themselves to be well structured and explained. The figures were easy to read and the structure of the paper highlighted the takeaways well. Below I outline some areas for potential further detail, which may allow the reader to understand the underlying assumptions of the study better.

• Uncertainty Study: while the explanation of the uncertainty results was quite clear, I found the description of the methodology to be a bit sparse, which made it difficult to understand the study itself. The authors state that they use ANOVA to partition the uncertainty into individual contributions. For those unfamiliar (or less familiar) with these statistical methods, a few more sentences about what specific test was used and what the method entails would be valuable. For example, are there any underlying assumptions about the probability distributions involved? Are the varying melt parameterizations used in the ice flow models considered part of the ice flow model variance, the climate model variance, or the interaction between the two?

• Melt Parameterizations: given that the focus of the study is the effect of melt forcing, it would be useful to describe the various melt parameterizations used in the ice flow models, as I imagine this significantly affects the response of the glaciers to climate forcing.

• Ice Flow Models: Tables 1 and 2 do a good job of outlining the experiments and the various climate and ice flow models used. However, it would be valuable to have a small description of the ways the ice flow models differ themselves, as this would provide some context to the result that most of the posterior uncertainty is due to uncertainty in the choice of ice flow model. A similar table that describes the pieces of each ice flow model, the assumptions they used (full Stokes vs. SIA, basal sliding parameterization, temperature-dependent rheology) would allow for a bit more insight into this result.

• Figures: A few of the figures could use some minor adjustments to make them clearer:
  • Fig 4: the axis labels and titles are hard to read; increase in font size would help significantly
  • Fig 3: making the lines in the legend thicker would help readability
  • Fig 2: is it valuable to have a legend so the reader can identify which simulation produces the most/least sea level contributions?