I thank the authors for their efforts to revise the manuscript. Before publishing I have a few remaining edits I suggest the authors make:

Line 12, page 2, you mention ice salinity twice. After that sentence I find it confusing to first state the RT model Kaleschke et al. 2010 used and then say a second RT model was utilized, when in fact this model of Maas et al. (2013) was not used for thickness but rather for snow depth. I suggest to better streamline this discussion.

Line 25, page 2. I don't see the value of this sentence and I would remove it. You are just mentioning how age has been used as a proxy for thickness in PIOMAS, it's not really relevant how it compares to observations as that is not the point of this paper. In addition, I think you could also streamline this discussion to simply refer to the fact that SIC has been assimilated, but SIT has mostly not been. I don't quite follow the reasoning why inconsistency between SIC of the model and SIC based on SMOS is a limitation for SIT assimilation. This entire paragraph needs to be streamlined and made more clear to get the main message across. I don't feel a strong case has been shown that in order to improve SIT assimilation you need to improve SIC and surface temperature. It seems to me you are arguing that's the reason why SIT is not assimilated, when in reality it could be because the data are not accurate enough yet for assimilation. The introduction seems to continually switch back in forth between discussing assimilating Tbs or SIT. For example on page 3, paragraph beginning on line 10 you state in preparing for a TB assimilation you are trying to figure out which RT model to use for assimilation SIT.
I agree that assimilation of Tbs is likely preferable to the derived fields but this case needs to be better stated and supported in the introduction. It remains a bit confusing the way the introduction is currently written.

Methods: I think it would be better to lead off with the SMOS data since the aim is to assimilate SMOS Tbs. You should also provide a clear reason why the two RT models were chosen. One seems suited for relating SMOS Tbs to thin ice thickness and the other for snow depth over thicker ice, but what is the main objective then here? The conclusion that the KA model is better for thickness seems already known. If the objective is for SIT assimilation then I'm curious why the Maas model was evaluated in the first place despite being of interest to show how the two RT models are sensitive to different input parameters. My question is, has this already been done in previous studies, or is this the first study to do that? I think in this section you could benefit from a schematic that shows all the processing steps and the data input/output. This could help greatly in understanding exactly why this study is being conducted.

What is the reason for the 3K bias over the open ocean? If part of it is from land contamination, why not expand the land mask out a few more pixels to remove any land influence? Do you expect the offset to remain linear (i.e. a constant 3.1K bias?). Figure 1 does not show a great fit even with the bias added, so some comment on the large range of the observed Tbs is needed.

Page 11, last paragraph seems to have something missing with the sentence starting with "In is area..."

I thank the authors for attention to the additional revisions. However the first paragraph of the manuscript is still not quite right. The first sentence is incomplete, and it is not true that ice age from NSIDC is used as a proxy for ice thickness in PIOMAS. Can you say which forecasting models assimilate ice age?

In general, I still feel the introduction could be better written and streamlined. Many sentences do not flow well together. This is likely a reflection of writing style, but several grammatical errors remain. I suggest the authors more carefully proofread the manuscript.

Finally, I still find the discussion somewhat confusing, you state in the introduction: A more
elegant way is to assimilate brightness temperatures rather than the retrieved sea ice parameters into the model, and then later state you are evaluating which radiative transfer models to use for assimilating sea ice thickness into ORAP5. So which should be done, assimilation of brightness temperatures, or SIT?