I think the authors for revisions in response to the reviewers' concerns. I have a few remaining issues however before accepting the paper for publication. One there are several typos in the paper, i.e. Hollands, 2010 should be Holland et al., 2010, that need to be corrected. A careful proofread will reveal these instances (without line #s it's impossible for me to detail each one).
Second, the Vavrus et al. 2012 reference for future loss of sea ice seems a bit strange as that is based on only one model that doesn't capture the Beaufort Sea High, which certainly will impact on the evolution of the sea ice cover. Why not reference studies that show results from several or all CMIP5 models and thereby better isolate the forced component from the internal variability? Several references exist, also as a function of different emission scenarios. Stroeve and Notz, 2015 is a review paper on the models performance and future projections that may be of additional interest and deals with the interannual/forced change issue.

Third, since you are blending the NSIDC ice motion data set with CERSAT, why not show a comparison for the months you have both to assess if there will be a potential bias in blending the two data sets? As a side note a reprocessed ice motion data set will be released soon by NSIDC that hopes to solve some of the known problems, but that will be too late for this study.
Fourth, a strong caution is needed for making concrete assessments of the ice age changes. Unfortunately, the ice age product stores the highest age of an ice parcel as the age of the ice of that pixel. For example, a pixel may consist of 90% first-year ice and 10% 5 year old ice for example, yet be labeled as 5 year old ice. It's a real limitation of the current ice age product. I would mention that in your paper as there may be real changes in ice age that you are missing that could explain the thickness decreases.
5. I would be very careful about the NCEP fluxes. If you are going to use them I would justify their use. ERA-Interim is known to be more accurate than NCEP. I don't think you can state that the ocean only contributes 4 W/m2 to melt as that is highly dependent on your fluxes. You could use a few different reanalysis to estimate the error in the radiative fluxes.
6. Note also that the underlying currents have a seasonal variability, with a faster mean current in winter than in summer.
No mention is made of the uncertainties in sea ice concentration which increases in summer because of melt processes. How does that impact your area export values?
7. On page 18, seems you need a separate conclusions section as you move from summer volume fluxes directly into conclusions.
Finally, I do not agree with the last statements. These data are of limited use for model evaluation (at least for GCMs). This is because you cannot expect the models to be in phase with the natural variability that impacts on the short time-period you have. Much longer records are needed that sample both internal and forced changes. Thus, I would be very careful in stating this.