Abadie et al. implemented a mechanistic and empirical soil model of COS exchange into the ORCHIDEE land surface model and compare those with observations of soil COS fluxes at several sites, representing different soil types. Through a sensitivity study they find the most important parameters for the soil COS flux and optimize those parameters with observations at two sites to improve the COS soil flux simulations. Finally, the authors provide an updated global soil COS budget, including both oxic and anoxic (wetland) soils. This is a very complete and thorough study that I find very well readable. My comments are hence minor.

Abstract:

P1, L44: -576 Gg S yr-1 for vegetation+soil, or only the vegetation?

Introduction:

P2, L62-63: This sentence reads weird.

P2, L63-64: The numbers 700-1100 GgS yr-1 sound like a very large gap. If I’m correct, Berry et al. (2013) added an additional ocean flux of 600 to close the gap, and Kuai et al. (2015) added 559 Gg S yr-1. Do the values 700-1100 GgS yr-1 represent the total ocean emissions? So not only the emission gap? A reference to a more recent inversion could be added (Ma et al. (2021) with a total gap of 432Gg S yr^-1).

P2, L78: better say something like: “they have usually not been considered in atmospheric COS budgets”.

P3, L87: form = from

P3, L113: For clarification, consider adding something like: “at the different sites that will be used for evaluation in this study”.

P3, L119: More important to the mechanistic model than to the empirical model?

P8, anoxic soil COS production: Did you consider to use the formulations of Meredith et al., 2018? These are similar to that of Whelan et al., 2016, but then with the alfa and beta parameters specific for peatland/wetland soils. That is, fca = 3700 for boreal peatland (Meredith et al., 2019) and alfa and beta for peatland from Meredith et al., (2018,). It would be interesting to compare those COS production estimates for wetland soils.

P10, L330: “….the same training method than the one used in Spielmann et al.” should be “….the same training method as the one used in Spielmann et al.”

P10, L333: If I understand the description in Wehr et al. (2017) right, the soil COS fluxes at US-HA are not based on eddy-covariance fluxes. It can be better described as flux-profile measurements, connected to CO2 soil chamber measurements and profiles.

P10,L350-351:” The stations located in the northern Hemisphere sample air masses coming from the entire northern hemisphere domain above 30 degrees.” The stations cover mainly North-America and actually Eurasia is hardly covered, so I would not agree with this statement.

P11, L384: what does “d” stand for?

P12, L412: spell out “DA”.

Results:

P13, section 3.1.1. I think the authors could put more emphasis on the potential role of nitrogen fertilization on soil fluxes. E.g. the results of IT-CRO, an agricultural site, could be emphasized in this context. Also the overestimated COS uptake at AT-NEU and ES-LMA could be discussed in light of nitrogen fertilization.

P13, L468 (Table 3): I would consider showing Table 3 as a figure. The same also for Table 4, which could even be combined with a Fig. from Table 3.

P14, L486-488: Or the division by PFT is not sufficient, and more specific information on e.g. nitrogen content is needed.

P14, L496: globally = generally?

P14, L503-505: This seems to be a repetition of line 500-501.

Fig 4: Can you show the same plots for observations?

Fig 5: Can the numbers at the end of the parameter names be replaced with an abbreviation? It is not entirely clear to me what the numbers represent, are they a PFT ánd soil texture number?

P16, L555: Can you remind the reader what PFT 15 is?

Figure E1: Can you explain the green and blue points, which are prior and which are posterior?

P17, L600-605: It is very interesting to read that the optimized parameters not only improve the simulated soil COS flux, but also the soil hydrology! Can you give some more details on the improvement of the soil moisture, e.g. with a figure or numbers?

P17, L612-613: It may be worth discussing the resemblance of the global distribution of COS soil fluxes of oxic soils with that presented by Kooijmans et al. (2021) (see their supplementary material). It is nice to see that the implementations of both the empirical and mechanical models show very similar global distributions in ORCHIDEE and SiB4. 

P19, L683-684: So the soils do not seem to explain the biases at high latitudes, so can we conclude that the vegetation sink is underestimated at the higher latitudes?

P19, L699-702: But at the same time it is inconsistent with comparisons at AT-NEU, ES-LMA, IT-CRO and US-HA, and the marginal model-observation biases can not explain the too high atmospheric concentrations, so I find this sentence out of place and would remove it.

P19, L709-710: Instead of showing the Launois et al. results in Fig. 10 you could consider to include that of Maignan et al. (2021), which to me seems to be a more fair comparison.

Discussion:

P20, L717: It would be relevant to compare also with recent global soil COS sink estimates of Kooijmans et al. (2021) and to include those in Table 5.

P20, L738: Please, specify that this is about the lack of seasonality in the COS soil flux.

P21, L759-772: The authors here talk about under- or overestimations, but it does not read as if this is compared to actual observations. So I do not think under- or overestimations are the right term here, they are simply higher or lower than other estimates. 

P22, The authors briefly touch upon the role of nitrogen fertilization in the discussion of section 4.3, but I think the authors could (and should) put more emphasis on the potential role of nitrogen fertilization in this manuscript.

P22, L788-789: More recent references such as Kaisermann et al. (2018) would be appropriate here.

Kaisermann, A., Jones, S. P., Wohl, S., Ogée, J. and Wingate, L.: Nitrogen Fertilization Reduces the Capacity of Soils to Take up Atmospheric Carbonyl Sulphide, Soil Syst., 2(4), doi:10.3390/soilsystems2040062, 2018.

Kooijmans, L. M. J., Cho, A., Ma, J., Kaushik, A., Haynes, K. D., Baker, I., Luijkx, I. T., Groenink, M., Peters, W., Miller, J. B., Berry, J. A., Ogée, J., Meredith, L. K., Sun, W., Kohonen, K.-M., Vesala, T., Mammarella, I., Chen, H., Spielmann, F. M., Wohlfahrt, G., Berkelhammer, M., Whelan, M. E., Maseyk, K., Seibt, U., Commane, R., Wehr, R., and Krol, M.: Evaluation of carbonyl sulfide biosphere exchange in the Simple Biosphere Model (SiB4), Biogeosciences Discuss. [accepted], https://doi.org/10.5194/bg-2021-192, 2021.

Meredith, L. K., Boye, K., Youngerman, C., Whelan, M., Ogée, J., Sauze, J. and Wingate, L.: Coupled Biological and Abiotic Mechanisms Driving Carbonyl Sulfide Production in Soils, Soil Syst., 2(3), doi:10.3390/soilsystems2030037, 2018.

Meredith, L. K., Ogée, J., Boye, K., Singer, E., Wingate, L., von Sperber, C., Sengupta, A., Whelan, M., Pang, E., Keiluweit, M., Brüggemann, N., Berry, J. A. and Welander, P. V: Soil exchange rates of COS and CO18O differ with the diversity of microbial communities and their carbonic anhydrase enzymes, ISME J., 13(2), 290–300, doi:10.1038/s41396-018-0270-2, 2019.

Abadie et al. simulated soil COS fluxes on a global scale using a mechanistic soil model in a land surface model ORCHIDEE, and evaluated the simulation results, from both the mechanical model and an empirical model based on scaling soil respiration, against 7 sites in the Northern Hemisphere. Furthermore, an atmospheric transport model LMDZ was used to investigate the contribution of different soil flux products to the latitudinal gradient of atmospheric COS concentrations. Moreover, sensitivity analyses were performed to reveal the importance of various parameters, which is useful to understand the control mechanisms of the soil COS fluxes. Note that the mechanistic model has been previously developed and published. Nevertheless, implementing the existing model in ORCHIDEE to study the global soil COS fluxes is desired. This is a very nice model study. The paper is well structured and very well written and is certainly suitable for the journal of Biogeosciences.

As has been noticed by the authors, the available field observations of soil COS fluxes are very limited, which is especially true when global COS fluxes are the focus of the study. In fact, all 7 sites are located in a narrow latitude range of 42 – 62 °N, and do not cover a full seasonal cycle, which makes it difficult to evaluate the simulation results on a global scale, and raises many questions around whether the presented simulation results are justified, e.g., whether smaller global COS soil flux than previous estimates is trustworthy, whether very large emissions in part of the tropics exist, not to mention the validation of the seasonal cycle and the diel cycle of the simulated results. On the other hand, these are also nice topics to be followed on. For this manuscript, I strongly feel that these points need to be better clarified in the revised version before publication.

Regarding the selection of the field sites, why were the soil flux measurements in an agricultural field in the Southern Great Plains by Maseyk et al., 2014 not used?

L25: remove “budgets” after “atmospheric COS”

L37 specify the region of the tropics, otherwise, it sounds like high emissions in all tropical regions

L170: please briefly discuss why the steady-state condition is valid? what assumption has to be made to make the steady-state valid?

L457-458: Clearly, the mechanistic model predicts nearly no seasonal cycle except for large production signals in the summertime, as is shown in Figure 2. However, this implies that relatively large net soil uptake exists in northern high latitude in winter times, when the temperature can be rather low and the land is covered by snow. This makes me wonder what the applicable range for the parameters shown in the method section, e.g., the valid temperature range of f_CA in eq. 16, the valid temperature range for 𝛼 and 𝛽 in eq. 17.

L471-473: Note that vegetation was also removed for the FI-HYY site, as is in Sun et al., 2018 “The moss layer or any other vegetation was removed to expose the humus layer inside the chambers.” This contradicts the statement. If the assumption would be true, what would be the mechanism for artificially enhanced COS production?

L478: The diel cycles of simulated COS soil fluxes by the mechanistic model shown in Figure 3 are totally not supported by the observations. Note that when relatively large uncertainties in the observations are considered, a minimum net soil COS uptake in the observations is not significant at all. Actually, the large discrepancy calls for a better understanding of the mechanistic model: what causes the diel cycles in the model but not shown in the observations.

L568: Section 3.1.5, although it is nice that the authors have made an effort to optimize soil COS flux, it may be premature. As the results are not used in the following results, I suggest leaving this section out or putting it into the Appendix.

L645: Section 3.2.2 Temporal evolution of the soil COS budget. It is expected that oxic soil COS sinks would decrease when atmospheric COS concentration decreases, and one could even expect that the decrease is, to the first order, proportional to the decrease of COS concentrations. However, the sharp decrease from 2016 is far beyond this. What are then the main reasons that can explain the sharp decrease in the mechanistic model?