On the whole this is a well thought out and thorough revision to the paper.

The concern that has not really been completely addressed is the shortness of the run used for evaluation. I am confused as to why it is impossible to run the model for longer, as I thought that this model is to be part of the operational system, which means there must be resources to run it for longer. Please explain this.

Related to this, it is not clearly stated in the manuscript (at least I could not find it) why such a short period has been chosen. In the case where a longer run really is a computational impossibility due to lack of resource, this must be stated in the manuscript, as it is not reasonable to expect readers to go trawling through the response to the referees to find this out. Also, please explain in the manuscript why was this particular snapshot of history was chosen. It might also be helpful to admit that this is a limitation of the work and explain what evaluation experiments are expected to be performed in future.

In relation to the model-data differences, I agree that it is useful to split the RMSE into mean and STD, but do not agree with you calling the STD the "random" component of the error. The STD is the variation about the mean. There is nothing random about it.

It would be too cruel to say that the language is tortured, but it is still someway from the elegant simplicity that enables easy reading. Copy editing will improve it further, but may not be able to do much with long and twisted sentences. Please try and make some further improvements to this in the next version.

The geoscience community is very large, and it is not possible for us all to know everything. It is therefore perfectly fine to say things like, "to the best of our knowledge" rather than to make a statement of which you are not confident.
In particular,
"To our knowledge, no other observations of ice properties are used for assimilation in short range NWP systems"
has become,
"Other observations of ice properties are rarely used for assimilation in short range NWP systems".
This change suggests that you now have knowledge that ice properties are sometimes used, in which case you should give references!
Instead, I suggest that this is perfectly acceptable:
"The authors of this paper know of no other cases where observations of ice properties are used for assimilation in short range NWP systems".

Thanks for your patience. I think this version is much improved in readability for a somewhat wider audience.