Thank you for your revised manuscript and comments. This revision has now been seen by one reviewer, who has only relatively minor comments. I'm therefore pleased to accept the manuscript for publication after those minor revisions detailed by the reviewer are comprehensively addressed.

Kind regards,
Paul

Dear Sifan Gu and Zhengyu Liu,

Thank you for your resubmitted manuscript. Following this revision, there are a number of issues which I would still like to see addressed before we can proceed to publication.

Regarding your response to the reviewer, saying 'There is no particular reason for the iterating and non-iterating grey layers. The left column is references for water column activity and the right column is for Holocene core top Pa/Th. Some references have both column activity and Pa/Th, therefore appear twice'. Please add this explanation to the caption, or remove duplicate references.

Regarding response to the reviewer ending in 'Our study is the first step trying to implementing 231Pa and 230Th into CESM. The parameters can be improved in the future with more observations available (line 489-492)'. Please can you add this point to the text. At present, you say ‘In addition, partition coefficient for different particles can be further tuned in the future, which can improve our understanding of the affinity of 231Pa and 230Th to different particles’. This leaves the reader wondering why you have not ‘further tuned’ the model. What I think you mean to say is that the parameters will hopefully be better constrained when more observational evidence becomes available?

Regarding your response to the reviewer comment which start with 'Line 230 and Fig. 9: a freshwater...' This is fine. This is a model description paper, rather than a paper trying to present new scientific results. However, please can you refer to the previous literature when describing this.

Following the reviewers comments about figure 5, is the purple the least squares regression through all of the data (i.e. all depth ranges)? Please explain in the caption. I also suggest that you change the plotting style so that points behind other points are still visible. It would also be valuable for the reader to see the slopes of the relationships in the different water depths. For example, the near 1:1 slope in fig. 5a is the result of cancelling slopes at different depths. the purple line is therefore not particularly useful.

Figures 2 and 3 captions must be improved. Explain what the filled contours are and what the circles are. It is obvious to someone who is familiar with the observational dataset, but not to others.

In your response to the reviewer ending in '“With boundary scavenging and sediment resuspensions added, dissolved 231Pa and 230Th activities in the abyssal should be reduced” in line 259-261.' This statement is true, but what the reviewer took issue with is the fact that you state that ‘Our model is unable to simulate the realistic dissolved 231Pa and 230Th activities in abyssal because boundary scavenging and sediment resuspensions are not included in our model.’ You provide a plausible hypothesis for this, but yours is a statement of fact. This requires further investigation.

Regarding the reviewers comment beginning with 'Line 281: “The sediment 231Pa...', I do not feel that you have addressed this comment adequately in your response. I would like to see this explored further, as the reviewer asks.

Regarding the reviewers comment 'Line 329: This statement should be proved statistically (like Fig. 5)', as far as I can see, adding the RMSE has not demonstrated this point. I agree with the reviewer that analysis like that used in figure 5 is the sort of thing that is required.

Regarding the reviewer's comment beginning with 'Line 360: In the following paragraph', I agree with the reviewer that this section still needs to be more carefully written. Reading this paragraph at present it is still not clear what the model can and can not reproduce. It is positive to acknowledge the limitations. Please also expand on what you mean by ‘fresh water hosing is self-consistent with the productivity pattern’.

Perhaps the most important reviewer's point that I do not feel has been addressed here is made clear by the reviewers comment 'Line 460: Yes, the parameters are somewhere in the range of the right magnitude, but not more. It would be great if this study would help to represent 231Pa/230Th in a realistic model, not only somewhere in the range of a factor of 25.'. The point being made is that there is observational data which can help constrain the parameters. The review has clearly asked in a number of places for an experiment to be done with parameters chosen to reflect this understanding. Please can you either undertake this experiment, or justify robustly why you do not feel that this is useful?

Finally, I would like to see the English being improved prior to publication. Please contact our editorial staff if you require more information about this.

Kind regards,
Paul Halloran, Topical Editor

Dear Sifan Gu and Zhengyu Liu,

Thank you for your response to the last round of comments. Almost all of the point have been addressed adequately. The one issue which I would still like you to answer is the final question in my previous correspondence:

Perhaps the most important reviewer's point that I do not feel has been addressed here is made clear by the reviewers comment 'Line 492: Yes, the parameters are somewhere in the range of the right magnitude, but not more. It would be great if this study would help to represent 231Pa/230Th in a realistic model, not only somewhere in the range of a factor of 25.'. The point being made is that there is observational data which can help constrain the parameters. The review has clearly asked in a number of places for an experiment to be done with parameters chosen to reflect this understanding. Please can you either undertake this experiment, or justify robustly why you do not feel that this is useful?

Your argument against this is (1) it is not feasible to undertake a large ensemble of sensitivity experiments, and (2) the model does an adequate job of capturing the major features as it is. However:

(1) Here the reviewer is not asking for a large sensitivity analysis, simply a single run to be done with more realistic parameters, i.e. parameters derived from observations. I would imagine that this is not too resource intensive?

(2) If the model is capturing the observed behaviour with unrealistic parameters, either the parameters are not important, or there is something wrong.

Please can you either perform this simulation and present the results, or explain why this is not a sensible approach?

Many thanks in advance,
Paul Halloran, Topical Editor.

Dear Sifan Gu and Zhengyu Liu,

Thank you very much for clarifying those points in your response. I am now happy that you have addressed the reviewers comments, and subsequently to accept your paper for publication.

Kind regards,
Paul Halloran, Topical Editor