Dear Jason and co-authors,

Thank you for revising your manuscript entitled "The high-resolution version of TM5-MP for optimised satellite retrievals: Description and Validation" following reviewer comments. I would also like to acknowledge the two reviewers for their constructive assessments of your discussion paper.

Although there are quite a few comments below, I do sincerely hope you will be able to work through them and submit a revised version for publication in GMD. It will serve as a useful benchmark for TM5-MP at high resolution on which future retrievals will be based.

Regards,
Fiona O'Connor

In relation to your responses to Reviewer 1:

1. There is too much similarity between the abstract and the conclusions. The conclusion section should benefit from including an outlook paragraph (which, e.g., should not be in the abstract). Now that a study has been done on modifying the chemistry and the resolution, and improvements in the model performance are moderate, what could be the next points of focus for further development of TM5. What would be the way forward to further improve CTMs? Another aspect which could be discussed in the conclusion : as the
differences between 1x1 and 2x3 are moderate, is it an option to still use the 3x2 version profiles as a priori for retrievals?

You have adequately addressed the similarity between the abstract and the conclusions and indeed, included a few sentences on future developments of TM5. However, I'd also be keen that you comment here on the STE findings in relation to the NH flux. How do you propose to investigate this further? What about plans to address the low bias in O3 deposition. Equally, I don't think you fully addressed the reviewer's point of "What would be the way forward to further improve CTMs?", which requires a more general response than solely focussing on TM5. Given the moderate improvement as a result of increasing resolution and given the reviewer's question above on using 3x2 profiles as a priori in retrievals, I would also suggest that you include your response in the conclusions section, should other readers have similar thoughts.

2. I have the impression that the manuscript is too strong in its argumentation that emissions are an important reason for discrepancies with observations. I would suggest that the authors make this claim more solid, e.g., by doing sensitivity experiments.

I have to agree with the reviewer here. Could I please ask that you include some sensitivity experiments, looking at the impact of reducing/increasing emissions by a fixed amount that represents the uncertainty in the emission inventories? Although you argue that including such experiments would turn the manuscript into a scientific paper rather than a description and validation paper, I believe that a key part of benchmarking a model is understanding the cause of the known biases. Such experiments would add to the current manuscript, would address a major concern by this reviewer, and are well within the scope of a GMD paper.

3. The analysis of the differences between the two resolutions should be improved. What about the difference in turbulent vertical transport between both model simulations? What about dry deposition parameterisations affected by the different resolutions? What about lightning NOx parameterisations? In the current manuscript, there is a lot of focus on convection, whereas other parameterizations might also play a role.

Although you argue here that the turbulent vertical transport scheme itself hasn't changed between the 3x2 version and the 1x1 version of TM5, its behaviour will potentially have changed due to the difference in resolution of the input fields. Therefore, could I please ask that you include a discussion on the differences, even if small, in the modelled turbulent vertical transport? In addition, you've added information on scaling the global annual lightning emissions totals between resolutions, differences in the global and vertical distributions of lightning NOx emissions are potentially important and would be worth including in the manuscript. Differences between Tiedtke and the mass flux simulations at 1x1 resolution would also be of relevance here.

4. It is also not made clear how large the differences are in the meteorological fields seen by the different resolutions. E.g., is the total precipitation equal in 3x2 and 1x1? Are the convective mass fluxes, when globallyaveraged, equal in both versions? Is the cloud cover equal when globally averaged? And the albedo? If not, it would be informative to quantify that.

I think the reviewer makes a very valid point here and some indication of how the meteorological fields themselves differ should be included in the manuscript (you've already included some of that information in your responses). It need not detract from the main focus of the paper, but would rather add to it.

5. As one of the aims of the model is to use it for generating instantaneous a-priori profiles and columns, it is not sufficient in this study to look at biases only. One should also look at the high frequency behaviour and thus, e.g., at correlations. E.g., Table 9 gives seasonal biases, but I think it is necessary to also show correlations. In addition, as the satellite retrievals will be used globally, it is not sufficient to quantify the difference between the 3° x 2° and 1° x 1° versions only for the observation locations of the manuscript. As the (tropospheric) columns of NO2, SO2 and CH2O from TM5 will be important for the retrieval, one could, e.g., estimate how well the 3° x 2° and 1° x 1° distributions are correlated spatially. This could also allow to better quantify whether using 3° x 2° instead of 1° x 1° still makes sense.

I think you have partly addressed the reviewer's comments here. However, one aspect which you haven't addressed is "it is not sufficient to quantify the difference between the 3° x 2° and 1° x 1° versions only for the observation locations of the manuscript." In order to fully address this comment, could I please ask you to put the comparisons at the observation locations into a global context?

6. Finally, it is not clear what selection criterium is used for putting some figures and tables in the main document and others in the supplementary material? If a reader decides not to read the supplementary material, he should at least have an idea of what he will miss.

I think this is a difficult one and personally, I don't like the use of Supplementary Material as I think it affects the flow and readability of a manuscript. However, I accept that you have made a judgement as to which figures you kept in the main manuscript and which you put in the Supplementary Material and indeed, you do reference the Supplementary Material adequately throughout the manuscript. No further action required here.


In relation to your responses to Reviewer 2:

1. A much more rigorous description of the two model setups and their differences is required. There is currently no mention of the vertical resolution although this is later reported as being important for e.g. STE fluxes. In its current format the model description in Section 2 is lacking information and is written in a way that makes it unclear which modifications are applied to the new TM5 model version and which to the higher resolution.

I think it is very clear what changes have been applied to TM5 since Huijnen et al., 2010, both in terms of science changes and MPI implementation. Perhaps, the only thing that I would ask is to be explicit in the last paragraph before Section 2.2 that you run the improved version of TM5 (TM5-MP) at both resolutions.


2. A paragraph should be added in Section 2 to clearly describe all details for the two model integrations used for
- start date of integrations and run length
- chemical initial conditions and spin up periods
- details of the analysis used (horizontal, vertical and temporal resolution).
Are these the same for both model resolutions or are they different? Are the simulated model years the same
this work. In particular, the authors should specify:
used for the emissions and observational datasets?

I think that this has been addressed sufficiently in the revised manuscript.

3. (including related points). The use of EMEP observations in it's current state is confusing. In line 242-243 the authors state that sites in "Norway, Finland, The Netherlands, Belgium, Poland, Germany, Spain, Italy and Portugal" are used for comparison. Why just use EMEP stations from the above and not the ones in other countries? Or is this a mistake (see later)?

I think you've adopted a reasonable approach, explained your rationale and included some text to explain your reasoning.

9. If the purpose of the comparison with EMEP is to evaluate the model performance in the new configuration, as well as addressing the differences in model resolution, the current analysis is not convincing. Comparison of model data with tropospheric ozone column from satellite would help better evaluate model performance on the global scale. This could also lead to better evaluate the model ozone profiles which currently show significant discrepancies with MOZAIC data. Further comparison with EMEP surface sites (and other
campaign data) would then add to the analysis, so far as the comparison is done across all suitable sites and a clear explanation is given if only a subset of sites is selected.

Again, this comment partly relates to a comment by Reviewer 1 about placing the comparisons with observations within a global context. Could I please ask you to consider including global comparisons between the two resolutions and/or using MOPITT O3 or multi-model output (e.g. ACCENT, ACCMIP) as a somewhat independent global O3 dataset?

10. In Section 5 the authors provide an analysis of budget terms for tropospheric ozone and compare these at the two different model resolutions. They state: ?the chemical tropopause calculated for 3x2 is applied for the analysis of 1x1 budget terms to ensure that a valid comparison is performed?. However, if convection and convective transport is significantly different in the two model resolutions (as the authors suggest) the position of the chemical tropopause at 1x1 should be at a higher altitude in the tropics compared to 3x2. Using the 3x2 chemical tropopause to analyse 1x1 budget terms is in my view inconsistent and the reduction in STE term with increasing resolution is likely to change If the 1x1 chemical tropopause is used.

I agree that the approach taken is valid and similar approaches have been taken in multi-model studies looking at the tropospheric ozone budget i.e. applying a tropopause definition (in some cases, derived from a single model) consistently across the range of models.

I have other minor comments, which are, in the main, grammatical:

Page 3, Line 90: Change "we examine impact on the" to "we examine the impact on the"
Page 3, Line 98: Change "pre-defined regional" to "pre-defined regional domain"
Page 3, Line 101: Explain MPI acronym on first use
Page 3, Line 114: Explain BL on first use
Page 6, Lines 208-210: Change "For our study concerned with the impact of horizontal resolution
on the performance of TM5-MP present 12 month simulations for the year 2006, which has been used
for previous benchmarking studies (Huijnen et al.,2010; Williams et al., 2012)." to "For our study
on the impact of horizontal resolution on the performance of TM5-MP, we present simulations for the
year 2006, which has been used for previous benchmarking studies (Huijnen et al., 2010; Williams et
al., 2012)."
Page 6, Line 220: "The lack of sector-specific information complicates the use of daily cycles
for e.g. the road transport component." Can you be explicit then what you do for the daily cycles?
Page 6, Line 226/227: What do you do for those emissions outside the 20S-20N band?
Page 24, Table 3: Add units for emissions of CO, HCHO, CH3OH and NOx to Table caption for completeness.
Other species in the table whose emissions are NOT in units of TgC/year are DMS, SO2 and NH3. Again,
can you add emissions units to Table caption? For example, it is currently unclear whether the SO2
emissions are in terms of S or SO2.
Page 8, Line 308: Change "Maximal" to "Maximum"
Page 9, Line 320: Change "hpa" to "hPa"
Page 9, Line 322: Correct the grammar of "In that the impact at Berlin is larger than at e.g. Barcelona
also ...."
Page 9, Lines 334-340: Significant differences exist between Rn profiles simulated with the Tiedtke
scheme and those simulated with the convective mass fluxes from ERA-Interim. Apart from saying there
are differences, can you use the observations to conclude about which simulation compares better to the
observations?
Pg 9, Line 350: Change "At global scale" to "At the global scale," - change consistently throughout manuscript,
including where you discuss the regional scale (e.g Page 16)
Pg 10, Line 364: Can you change the word "residual" to use "ratio" instead (and be explict about what ratio
it is i.e. 3x2/1x1 or vice versa) and add this explicitly to the captions for Figure S6a and S6b?
Pg 10, Line 377: Correct the grammar in "Olszyna et al., 1994) and the also the efficiency of the NOx recycling
terms"
Page 25, Table 4: Can you be explicit about the percentage difference in the caption i.e.
(ResolutionA-ResolutionB)/Resolution A,
rather than referring to it as ResA/ResB?
Pg 10, Line 388: I think the term "The stratospheric burden of O3 (BO3S)" is unclear. If it is the stratospheric
burden of O3, then I would avoid using BO3S (perhaps use BO3 (strat)?) to avoid confusion with the tagged tracer
O3S. If you are referring to the burden of stratospheric O3 in the troposphere derived from the tagged tracer, then
I would recommend changing to "The burden of stratospheric O3 in the troposphere (BO3S)". Change in Table 4 as
well for consistency.
Pg 10, Line 390: Change "At global scale" to "At the global scale," and elsewhere (e.g. Page 11, line 420)
Pg 11, Line 425: Change "have shown previous versions of TM5 to be a the low end of the model spread" to
"have shown previous versions of TM5 to be AT the low end of the model spread"
Pg 11, Line 426: Can you re-phrase "the surface deposition flux to e.g. should be increased by ~10% in TM5-MP
towards the multi-model mean value."? I found this confusing. How about "the surface deposition flux in TM5-MP
is still biased low relative to the multi-model mean value by ~10%".
Pg 11, Line 430: Change "countries chosen so to cover a range of latitudes." to "countries chosen to cover a range of
latitudes."
Pg 12, Line 448: Change "is capture rather well." "is captured rather well."
Pg 13, Line 490: Change "The maximal biases in" to "The maximum biases in" - please correct "maximal" throughout
manuscript
Pg 13, Line 494: Change "(Pearson's r in the range -0.3-0.3)" to "(Pearson's r in the range from -0.3 to 0.3)"
Pg 15, Lines 564-575: What about the role of wet deposition? It is quite a significant term in the HNO3 budget.
Pg 15, Line 580: Change " the underestimation in NO2 (Fig. 6) in not " to " the underestimation in NO2 (Fig. 6) is not "
Pg 16, Line 628: Change "term is too." - I presume you mean to add "large" or "strong" here.
Pg 16, Line 648-649: The sentence "At 1º x 1º significant improvements occur as a result of the better temporal
resolution of the emission sources as a result of increasing horizontal resolution." doesn't make sense. Do you mean
that the temporal variability of modelled SO2 has improved as a result of increasing horizontal resolution? Or do you
mean that you've altered the time profile of the so2 emissions in 1x1?

Dear Jason and co-authors,

Firstly, I would like to thank you for taking my comments on board and for submitting a revised manuscript. Secondly, with the holiday season, I would like to apologise for not making a decision on your revised manuscript sooner.

On reading your revised manuscript, I feel that you have addressed the majority of my comments following the reviews as best you can, given your funding constraints as discussed by email. Therefore, I would like to recommend that your paper be accepted for publication in Geoscientific Model Development. However, there were a few grammatical errors with the newly added text (which I detail below) and I kindly ask you to correct them before publication.

Best wishes for 2017,
Fiona O'Connor

Minor comments:
1. Abstract, lines 143/144: "Comparing monthly mean distributions in lightning NOx we show that applying ERA-interim convective mass fluxes We show that". I think this should probably read " Comparing monthly mean distributions in lightning NOx and applying ERA-interim convective mass fluxes, we show that"

2. Description of TM5, line 223: Change "Parsing" to "Passing"

3. Description of TM5,lines 273/274: Can you add units for CDNC?

4. Description of TM5, line 305: Please correct "whereas for NO3 it forms one molecule of HNO3 is formed"

5. Emissions, line 360: Change "speciating a fraction" to "specifying a fraction"

6. Effect on Atmospheric Transport, line 486: Remove "which are defined as" ... definition already included in line 483

7. Impact on tropospheric photolysis frequencies, line 504: Change "Southern" to "South"

8. Impact on tropospheric photolysis frequencies, line 514: Correct figure number

9. Impact on tropospheric photolysis frequencies, line 516/517: This is now a bit repetitive, given newly added line above

10. Short-lived reservoirs, line 835: I suggest that you change "transport merges differences between simulations" to "transport has little impact on simulations"

11. Conclusions, line 938: Remove "are" in "1x1 are over the oceans"

12. Conclusions, line 979: Remove "applying" in "using the applying 60 vertical"