Dear Drs. Shi and Wang: I have received the reviewer reports of your revised manuscript and both reviewers concluded that the manuscript was significantly improved but that there remain a number of points that need clarifying and correcting before the m/s can be accepted for publication. I also have some comments from my reading of your revised paper:

1. The last sentence of the abstract need clarifying. Suggestion: “In summary, our results show projected trends of decreasing NH SWE under three different global warming scenarios, related to a shortening of the snow accumulation period and a reduction in solid precipitation.”

2. In reference to Reviewer 1 Major Comment 1, I suspect this is related to excessive summer SWE values from a small group of models (BCC-CSM, GFDL, NorESM1). To be consistent, you should include some evaluation of the models’ seasonal cycle to go with Figures 1 and 2. This process may identify some “outlier” models. You can check if this has any impact on projected changes by plotting projected SWE change against your seasonal cycle index. My experience is that the projected changes are fairly robust across models with the exception of the summer period.

3. You don’t mention what version of GlobSnow you used (version 2.0?). If you used the standard version of GlobSnow this is masked out over mountainous terrain which means it is biased low. The bias will be most pronounced in the summer period. The affects the validity of the conclusions reached from Figure 2. I suggest you include an additional estimate of NH SWE from MERRA which gave an unbiased estimate of NH SWE in the recent evaluation of Mudryk et al. (2015) - paper attached fyi. Note that Mudryk et al. (2015) added an estimate of mountain region SWE to GlobSnow for the comparison in their Fig. 3a.

I look forward to seeing your revised m/s.

Best regards, Ross Brown

Dear Authors - apologies for the delay as I was away from the office without e-mail access. I just read through your revised m/s and the English still needs work in several places to clarify the text. I will go through it this week and send you a version with my suggested edits.

I did not see your responses to the 3 comments I included in my previous e-mail.

Ross Brown, Editor

The paper reads much better now. Note that the seasonal cycle of SWE in your Fig. 5 is affected by a small number of models with anomalously high SWE during the summer months (NorESM1, MIROC5, GFDL). However, removing these models from your analysis would probably not change the results appreciably.