|The authors have clarified many aspects of their study and I like to congratulate them to this interesting article. I finally like to point out two points:|
For sea ice always temperature control during sampling and imaging is an essential issue, as one wants to avoid temperature cycling that may lead to microstructure changes. As an example for the June samples the described temperature chain was 1. Sampling at amient temperatures of ~-2°C, 2. Transported in insulated boxes without active temperature control, 3. Stored at ~-2°C sampling/"working" temperature 4. centrifuged for 5 minutes at -5°C, 5. Imaged for 18 minutes in a microCT chamber for which temperatures of +2°C were reported. Apparently during all these steps there is potential to change sample temperatures from in situ values. This could change porosity and microstructure, and as porosity is roughly inverse proportional to negative temperature, slight changes in temperature may have considerable effects on microstrcture. To ascertain that this has not happened one should check/document the salinity (and thus freezing temperature) of centrifuged brine.
2. XRT image analysis and anisotropy.
The authors have clarified how the degree of anisotropy is derived from major and minor axis lengths lmin and lmaj, that is as DA = 1- lmin/lmaj. The results, with DA between 0.14 and 0.71, thus correspond to ratios of major to minor axis length in the range of 1 to 3. I rate these values as small for a strongly anisotropic medium as sea ice. Most importantly, I rate the derived metric as problematic, as it mixes open and close pores. This is illustrated by the following calculaton: assume that the pore space consists to 90% of one large interconnected through-sample pore which would have DA~1 and a couple of small inclusions (lets say 3) that are sperical (DA = 0). Then the DA obtained by averaging over all pores would be (1+0+0+0)/4=0.25, from which one in turn would derived lmaj/lmin of 4/3. This mean value apparently tells little about the relevant characteristics and structure of the pore space. Hence, to get meaningful numbers one rather should do computations for the open and the closed pore space separately. Also a volume average for a metric may be, in terms of processes considerered, a more relevant presentation than a number average.
2. XRT spatial resolution
Another aspect is that the effective spatial resolution of the microCT in this study was 284 micrometer (reported by the authors), while most pores during May and June, observed optically under transmitted light (histogram in Fig.8) had smaller diameters. Hence these are not expected to be resolved by the microCT. This in turn points to the question how valid the CT-derived ratios of open/close porosity and the anisotropy in Fig. 10 are. To properly derive such metrics there appears to be a need for higher spatial resolution (e.g., according to Figure 8 one would need 20-30 micrometers to capture the modal values of pore widths for May and June).