In this manuscript, Berchet et al. use a five-year record of atmospheric COS concentrations measured at a Paris suburb site to examine anthropogenic and biogenic COS fluxes over Western Europe. Given the sparsity of COS observations globally, it is encouraging to see this work leveraging long-term COS observations at a single site to shed light on regional COS flux components, aided by a Lagrangian transport model. The study shows promise in enhancing the usability of COS as a photosynthetic tracer at regional scales.
While the revised manuscript has been improved in response to prior review comments, I find some remaining issues. I have a few suggestions for improving the robustness and clarity of the findings.
To alleviate potential bias from nighttime boundary layer parameterization, I suggest running additional tests to re-examine the diagnostics for different flux combinations (Table 2) using only mid-afternoon COS observations. Parameterization of the nighttime boundary layer is notoriously challenging and can lead to transport errors (e.g., Díaz-Isaac et al., 2018; Lopez-Coto et al., 2020; Monteiro et al., 2024). As a result, most Lagrangian inverse modeling studies assimilate only observations during well-mixed afternoon conditions, which both the authors and the reviewer, Dr. Whelan, have acknowledged. Lacking information on how well FLEXPART represents the nighttime boundary layer, I am not confident that bias in simulated nighttime concentrations could be attributed solely to the representation of nighttime stomatal conductance in terrestrial biosphere models (ORCHIDEE and SiB4). While I agree with some of the reasoning in the authors’ response, I do not see this issue fully addressed. I reckon that the simplest way to test this issue could be to redo a set of analyses using only mid-afternoon COS observations and see if error diagnostics still follow the patterns in Table 2. As the atmosphere integrates fluxes day and night, afternoon observations should still be sensitive to fluxes in previous nights within the footprint area.
On the evaluation of anthropogenic COS flux estimates, because point sources of viscose and coal industries are largely collocated (Fig. 2c), their influences on observed COS concentrations would be highly correlated. Therefore, it seems challenging to evaluate COS fluxes from these two sectors separately. On the other hand, because changes in the wind direction may allow the site to pick up flux signals coming from different regions (Paris/Rouen, Benelux, British Isles, etc.), it would seem more helpful to focus on contributions from different regions rather than different industries that are collocated in space.
On the evaluation of biogenic COS fluxes, given that the focus is on nighttime bias, presumably resulting from underestimated nighttime stomatal conductance, and on summertime bias caused by potential emissions from certain crops, it may help to further assess bias, RMSE, and correlation (following Table 2) for daytime and nighttime measurements and by season. Model bias in representing nighttime COS uptake has been known since Kooijmans et al. (2017), and it would be better to give quantitative insights beyond reiterating this message. I suggest some simple tests along this line, for example, treating nighttime and daytime ecosystem fluxes as two separate tracers and checking how much nighttime fluxes need to be bumped up in order to match observed COS concentrations. This would provide more detailed insights into model bias and help inform relevant process representations in terrestrial biosphere models.
Lastly, the study could give readers a high-level view of whether anthropogenic or biogenic COS fluxes dominate the variability in observed COS drawdown or enhancement (ΔCOS). Comparing Fig. 1d and Fig. 4, it appears the study domain is dominated by biogenic COS fluxes, but I’m not sure. I suspect that many in the carbon cycle research community would care about this question, but I don’t see it explicitly answered.
Specific comments
L4: "at a better scale than the global scale" –> "at the regional scale"
L13–14: "We find that the net ecosystem COS uptake simulated by both ORCHIDEE and SiB4 is underestimated in winter at night" - As I suggested above, it would be better to redo the evaluation with mid-afternoon COS observations only to confirm if the bias results from nighttime stomatal conductance representation.
L16–17: "In Summer, both models properly represents fluxes, with better agreement from ORCHIDEE in terms of magnitudes." - What about potential COS emission episodes?
L24–37: Since this study is not an atmospheric inversion, I don't see a need to introduce it. Consider focusing on the forward model application of linking flux fields to mixing ratios.
L40–80: This part feels overly detailed as an introduction. I suggest consolidating these bullet points into a coherent paragraph and emphasizing the key uncertainties this study aims to tackle. To tidy it up, it may help to list the magnitudes of different COS flux components in a table.
L100–L110: This paragraph needs to give a clear roadmap of what the study aims to achieve. Currently, it sounds like this study is picking on the bias in Zumkehr et al. (2018) inventory, but there is more to be said about biogenic COS fluxes and the underlying physiological processes.
L121–122: How often was the calibration carried out?
L134: Here, it would help to refer to Sect. 2.3 for background calculation (L232–238). Or move that paragraph on background calculation here.
L148: "DMS emissions can only be non natural ..." - Wetlands and the ocean do emit DMS. I get the point of this sentence, but the logic does not follow.
L151: "box models" - I would call them spatially resolved box models to avoid potential confusion.
L165–168: Does this mean that ORCHIDEE and SiB4 COS flux fields used here are optimized posterior estimates?
L172: "In particular, biogenic fluxes exhibit a significant diurnal cycle." - I am confused here. Aren’t the posterior fluxes from Remaud et al. (2022) and Ma et al. (2021) monthly?
L175: "We assess the sensitivity of our simulations to daily varying biogenic fluxes by using 3-hourly vegetation uptake as simulated by ORCHIDEE and SiB4 for the year 2016." - Are these posterior flux estimates or unadjusted prior flux estimates?
L183–184: The components sum to 61.4 GgS per year not 62.1 GgS per year. Are there other minor components not listed here?
L186: I think you mean "effective" not "efficient".
L220: One advantage of a Lagrangian model over an Eulerian model is the flexible resolution at which transport is resolved. As a result, Lagrangian models can use meteorological input at a finer resolution than the resolution at which footprints are aggregated. Is there a reason to run FLEXPART with 1° meteorology input instead of the native resolution (0.25°) of ERA5?
L243: An RMSE of 35 ppt does not feel like a small error; it seems to be around 1/3 of the seasonal cycle amplitude of COS (Fig. 1b).
L246–248: Again, here it is crucial to clarify whether the three-hourly flux estimates are optimized posterior estimates or prior estimates coming out directly from terrestrial biosphere models.
L249: "from 0.74 to 0.72" - The decrease in correlation is a bit surprising because, all else being equal, I would expect the correlation to stay unchanged or increase slightly with the inclusion of diurnal variability. Are these correlation coefficients just for the year 2016 or the entire period?
L249: "The variability is not better reproduced when adding the natural emissions to the background" - Which component(s) of "natural emissions"?
L276–291: Unclear from this paragraph how the contributions of biogenic, anthropogenic, and oceanic COS fluxes to observed COS concentrations compare with each other. I suggest adding a table to list their respective contributions to ΔCOS.
Fig. 1: In panel a, it's quite challenging to distinguish between GIF (afternoon) from GIF (all). Consider enhancing the color contrast or use different symbols. In the caption, background calculation should refer to Sect. 2.3 not 2.1.
Fig. 2: In panel e, it would be more helpful to show the summed footprint map over the entire study period to check the influence area coverage. Readers may not know if the footprint map for May 2019 is representative of the whole period (2014–2019).
Fig. 4: Unclear what "growths" means in the figure legend. For panel c, it would be better to report COS fluxes in pmol m^-2 s^-1.
References cited
Díaz-Isaac, L. I., Lauvaux, T., & Davis, K. J. (2018). Impact of physical parameterizations and initial conditions on simulated atmospheric transport and CO2 mole fractions in the US Midwest. Atmospheric Chemistry and Physics, 18(20), 14813–14835. https://doi.org/10.5194/acp-18-14813-2018
Kooijmans, L. M. J., Maseyk, K., Seibt, U., Sun, W., Vesala, T., Mammarella, I., Kolari, P., Aalto, J., Franchin, A., Vecchi, R., Valli, G., & Chen, H. (2017). Canopy uptake dominates nighttime carbonyl sulfide fluxes in a boreal forest. Atmospheric Chemistry and Physics, 17(18), 11453–11465. https://doi.org/10.5194/acp-17-11453-2017
Lopez-Coto, I., Hicks, M., Karion, A., Sakai, R. K., Demoz, B., Prasad, K., & Whetstone, J. (2020). Assessment of Planetary Boundary Layer Parameterizations and Urban Heat Island Comparison: Impacts and Implications for Tracer Transport. Journal of Applied Meteorology and Climatology, 59(10), 1637–1653. https://doi.org/10.1175/JAMC-D-19-0168.1
Ma, J., Kooijmans, L. M. J., Cho, A., Montzka, S. A., Glatthor, N., Worden, J. R., Kuai, L., Atlas, E. L., & Krol, M. C. (2021). Inverse modelling of carbonyl sulfide: Implementation, evaluation and implications for the global budget. Atmospheric Chemistry and Physics, 21(5), 3507–3529. https://doi.org/10.5194/acp-21-3507-2021
Monteiro, V., Turnbull, J. C., Miles, N. L., Davis, K. J., Barkley, Z. R., & Deng, A. (2024). Assimilating Morning, Evening, and Nighttime Greenhouse Gas Observations in Atmospheric Inversions [preprint]. ESS Open Archive. https://doi.org/10.22541/essoar.171007090.01499748/v1
Remaud, M., Chevallier, F., Maignan, F., Belviso, S., Berchet, A., Parouffe, A., Abadie, C., Bacour, C., Lennartz, S., & Peylin, P. (2022). Plant gross primary production, plant respiration and carbonyl sulfide emissions over the globe inferred by atmospheric inverse modelling. Atmospheric Chemistry and Physics, 22(4), 2525–2552. https://doi.org/10.5194/acp-22-2525-2022
Zumkehr, A., Hilton, T. W., Whelan, M., Smith, S., Kuai, L., Worden, J., & Campbell, J. E. (2018). Global gridded anthropogenic emissions inventory of carbonyl sulfide. Atmospheric Environment, 183, 11–19. https://doi.org/10.1016/j.atmosenv.2018.03.063 |
