the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 May 2023
The sensitivity of Southern Ocean atmospheric dimethyl sulfide to modelled sources and emissions
Abstract. The biogeochemical behaviour of the Southern Ocean is complex and dynamic. The processes that affect this behaviour are highly dependent on physical, chemical, and biological constraints, which are poorly constrained in Earth System Models. We assess how emissions of dimethyl sulfide (DMS), a precursor of sulfate aerosol, change over the Southern Ocean when the chlorophyll-a distribution, which influences oceanic DMS production, is altered. Using a nudged configuration of the atmosphere-only United Kingdom Earth System Model, UKESM1-AMIP, we performed nine 10-year simulations using forcings representative of the period 2009–2018. Four different seawater DMS data sets are tested as input for these simulations. Three different DMS sea-to-air flux parameterizations are also explored. Our goal is to evaluate the changes in oceanic DMS, sea-to-air fluxes of DMS, and atmospheric DMS through these different simulations during austral summer. The mean spread across all the simulations with different oceanic DMS datasets, but the same sea-to-air flux parameterizations, is 112 % (3.3 to 6.9 TgS Yr−1). The mean spread in simulations using the same oceanic DMS dataset, but differing sea-to-air flux parameterisations is 50–60 % (2.9 to 4.7 TgS Yr−1). The choice of DMS emission parameterisation has a larger influence on atmospheric DMS than the choice of oceanic DMS source. We also find that linear relationships between wind and DMS flux generally compare better to observations than quadratic relationships. Simulations that implement a quadratic emission rate show on average 35 % higher DMS mixing ratios than the linear emission rates. Simulations using seawater DMS derived from satellite chlorophyll-a data in combination with a recently-developed flux parameterisation for DMS show the closest agreement with atmospheric DMS observations and are recommended to be included in future simulations. This work recommends for Earth System Models to include a sea-to-air parameterization that is appropriate for DMS, and for oceanic DMS datasets to include inter-annual variability based on observed marine biogenic activity. Such improvements will provide a more accurate process-based representation of oceanic and atmospheric DMS, and therefore sulfate aerosol, in the Southern Ocean region.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3034 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3034 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-868', Anonymous Referee #1, 01 Jun 2023
General comments
The topic of the study is important and urgent (https://doi.org/10.1525/elementa.2022.00130) and the study is a step toward improving a component of Earth System models that is not well constrained. The study quantified the sensitivity of atmspheric DMS to oceanic DMS source and flux parameterization over the Southern Ocean. While these results are informative, there are a few issues that need to be addressed to support some of the conclusions made in the study. I provide these issues with suggestions on how they can be addressed below. I also provide specific comments below my general comments.
The choice of oceanic DMS parameterization: The authors chose Anderson et al. (2001) for deriving DMS from chlorophyll-a, while there are other algorithms to derive DMS from chlorophyll-a and other environmental variables developed since 2001.This choice needs to be explained/justified in the manuscript.
Investigation of inter-annual variability: The authors conclude that interannual variability is important (e.g. L18, L416), but this is not demonstrated well in the study. Figure A1 is the only result where the interannual variability can be discussed, bit it gives an impression that the interannual variability is not important (the year-to-year variation of the mean DMS concentration is subtle). To assess the impact of representing interannual variability in the DMS source, the authors could do an additional simulation in which MODIS-DMS does not change from year to year. By comparing this simulation with one of the MODIS simulations in Table 2, the impact can be quantified and its importance can be justified.
Investigation of spatio-temporal variability: The authors introduce the importance of spatio-temporal variability (L57) of DMS in the Southern Ocean, but the results are mostly discussed for spatially and temporally averaged values. Furthermore, some model results are averaged over the entire Southern Ocean, which are compared to the observations that are representative of specific regions/seasons (e.g. L312, L318, L320, L323, L352). While apple-to-apple comparison may not be possible, there are places for improvement. For example, the spatial maps of the simulated DMS flux and atmospheric DMS (such as done for oceanic DMS concentration as in Fig 2a-d) can be added to better compare with the observations from specific regions. These figures will also demonstrate the importance of spatial variability.
Writing style: Section 5 seems like a combination of discussion and conclusions, so I suggest to change the section title to “Discussion and conclusions”. However, I also noted in my specific comments that there are paragraphs in the Results section that I feel belong to Discussion.
Specific comments
L3: Suggest to delete or rewrite this sentence. The authors did not assess how DMS emissions change when the chlorophyll-a distribution is altered. This would require an investigation of the relationship between the DMS emissions and the chlorophyll-a distribution.
L9: “The mean spread” of which quantity?
L11: “The choice of” Is this conclusion correct? The previous sentences indicate that the choice of the oceanic DMS source (112%) has a larger influence on the atmospheric DMS than that of the flux parameterization (50-60%).
L12: Suggest to rewrite this sentence. It is not that linear relationships compare better to observations, but assuming/simulation with the linear dependence of the DMS flux on wind speed results in better representation of atmospheric DMS distribution.
L15: Where is the evidence for the closest agreement? This could be clarified in "Discussion and conclusions".
L34: Suggest to replace “mechanisms, with varying focus” by “approaches that are dependent”.
L40: Suggest to replace “approximate” by “prescribe”.
L48: Suggest to repalce “and has” by “and this emission has”.
L55: Older quadratic approaches are based on other gases? If so, this could be mentioned here to emphasize the issue.
L57: Should “observations” be “concentrations”? Observations are highly variable implies that some regions/seasons are observed more than the others. I am not sure if this is what the authors meant to state here.
L59: Suggest to replace “real-world” by “remotely-sensed”.
L61: Suggest to either remove “nudged to observation” or replace it by “nudged-to-observation".
L68: “the relative importance” to what?
L73: Suggest to move the description of MEDUSA later when the oceanic DMS dataset is introdued. The model used here is the atmopsheric-only configuration, so the oceanic component does not need to be introduced here.
L85: In addition to the details provided in this section, I suggest to add information on boundary conditions such as sea surface temperature data used for these simulations that affect both the atmospheric circulation and DMS emissions (Eq 3).
L86: Suggest to rewrite “biologically productive” to something else. “Biological” is a bit ambiguous, as it is broad. It could refer to microalgae to big animals that have different seasonality. For clarity, it may be better to refer to DMS instead of biology.
L88: This paragraph belongs somewhere else, as it is not a model description.
L110: "modified” implies that the parameterization was modified, but my impression is that the authors only modified the chlorophyll dataset. If so, please rewrite this sentence.
L115: Suggest to define the acronym MODIS when it appears for the first time in the text instead of here.
L117: How are DMS sources and emissions from sea ice covered areas treated? Small gap of 1% probably excludes ice-covered grid cells. DMS emissions from marginal ice zone can be important (https://doi.org/10.1525/elementa.2020.00113).
L148: It is unclear how applying W14 and B17 parameterizations leads to testing the lower limits.
L158: Related to my earlier comment on L85, was HadISST used also for the boundary conditions for the atmospheric circulation simulations?
L172: Suggest to replace “validate” by “evaluate”.
L201: Suggest to replace “includes chlorophyll-a” by “accounts for chlorophyll-a variations”.
L209: Suggest to move “respectively” to “used, respectively)”.
Fig 3: For readers not familiar with violin plots, the caption needs a brief description of what violin plots represent in general.
L249: Suggest to either delete “and would be expected to be highly biased” or clarify what it is meant by highly biased. Also for fair comparison, when was the TAN1802 data taken (before/during/after blooms)?
L265: Suggest to add a few words to briefly explain why higher correlations are expected at higher latitude.
L266: The fact that the correlation in the Southern Ocean is high (0.75) does not necessarily support the previous statement. It needs to show the latitudinal dependence (whether the correlations increase with latitude).
L270: Half sounds many, but in fact only two (out of four), I think? It might be good to also discuss the lack of DMS models in current ESMs. Also, this paragraph sounds more like discussion than results.
L273: Suggest to replace “realistic” by “readily available”.
L285: Is Figure 3 adequate to make the claim that MODIS-DMS is adequate?
Fig 5: Suggest to utilize color for distinguishing different DMS sources
L301-302: The two sentences say the same thing. Suggest to remove one. Also, this finding (the choice of DMS source is more important than the choice of flux parameterization) is contradictory to what is written in the abstract (L11).
L303: It is unclear what this sentence is referring to.
L313: Why not compare the model results averaged over the same region as Webb et al. (2019) instead of comparing with the Southern Ocean averages for apple-to-apple comparison?
L328: “is likely positively skewed” can be checked to see if the model agrees with this finding by looking at the spatial distribution.
L332: This paragraph seems more appropriate to be a part of Discussion and conclusions, instead of Results.
Sec 3.3: How does the distribution of atmospheric DMS differ from that of oceanic DMS emission?
Figure 6: Is this model result representative of the Southern Ocean or the three observational stations?
L365 and Fig 7: Suggest to delete the results of PMOA. As the authors stated (L378), the results of PMOA require further investigation that does not seem to fit into the scope of this study. Adding this result would require additional info on intro, methods, and discussion.
L374: Suggest to rewrite this sentence. It states about CMIP6 but the paper cited is from 2011, many years before CMIP6.
L385: Suggest to cite a paper that compares nudged vs non-nudged runs to support this argument.
L402: How is this different from L399?
L420: Why recommend Lana while there is an updated climatology?
Appendix A: Figures should be labeled as A1-4 instead of A-D1
Citation: https://doi.org/10.5194/egusphere-2023-868-RC1 -
RC2: 'Comment on egusphere-2023-868', Mingxi Yang, 26 Jul 2023
Understanding and being able to simulate DMS emission in the Southern Ocean is critical for constraining aerosols, clouds, and so climate over that region. This paper tests a range of DMS emission estimates within an earth system model and examines the sensitivities in the simulated atmospheric DMS. They argue that using a satellite monthly chlorophyl to drive seawater DMS concentration is superior to using a climatological seawater DMS field, as the former includes more (interannual) variability. They further state that using a linear relationship between gas transfer velocity and wind speed is unsurprisingly better than using a quadratic relationship.
In general, the authors have explored an important topic. However I don’t feel like I’ve learned a lot from the paper because the analysis is incomplete. I think the paper may be improved by considering the following points:
- That using a satellite monthly chlorophyl to drive seawater DMS concentration is superior to using a climatological seawater DMS field seems not unreasonable in principle. However the validations using only two cruises are not very convincing to me. Can the authors include more validation datasets? For example the SO-GasEx dataset is within the 10-year window and includes seawater/atmospheric DMS as well as DMS flux.
- There have been more recent seawater DMS parametrizations than Anderson et al. 2001. The authors should include them or explain why they aren’t considering these more recent developments. They should also acknowledge that seawater DMS doesn’t just depend on Chla, but is also sensitive to a number of other biological parameters.
- Comparisons in atmospheric DMS neglects the time component entirely. Maybe the authors can further expand their 10-year window to facilitate comparisons with those earlier and very recent atmospheric DMS measurements?
- What are the impacts of these different DMS emission simulations on aerosols, CCN, AOD, etc? Can authors validate those?
Specific comments:
Please make careful distinction between flux and Kw. The authors have incorrectly interchanged their usage a number of times in the paper.
Overall the writing isn’t very concise. I suggest further proof-reading from all authors.
Line 11-12. Why, given that the spread is greater among different oceanic DMS datasets?
Line 12-15. Be more exact. It’s a linear relationship between gas transfer velocity and wind speed. More so than between flux (emission) and wind, since flux also depends on seawater DMS concentration
Line 25-26. DMS isn’t just produced by phytoplankton. There’s also bacterial production. So suggest replacing ‘phytoplankton’ with ‘marine biota’ and update references accordingly.
Line 30. This paragraph talks about how ESMs represent DMS emission. But I think it’d be better to first talk about how DMS emission is estimated in general. i.e. flux = Kw * deltaC ~= Kw * [seawaterDMS]. Then you can talk about the different parametrizations of Kw (the gas transfer velocity), as well as the different ways [seawaterDMS] is estimated.
Line 38-39. Not sure what ‘simulate biases’ means here
Line 47. Suggest adding ‘to some degree’ after ‘events’, given the complexity in biological response.
Line 49. Not between atmospheric DMS and wind speed, but between the gas transfer velocity and wind speed.
Some of the references aren’t the most appropriate:
Vlahos and Monahan 2009 interpreted other groups’ measurements
Wanninkhof 1992, 2014 focused on CO2, not DMS. Nightingale et al. 2000 and Ho et al. 2006 focused on 3He/SF6, not DMS. Liss & Merlivat 1986 synthesized a range of lab/field works, but not specifically of DMS. Those parametrizations are often used in ESMs for estimating DMS flux, inappropriately. So it’s useful to introduce them here, but with a clarification.
Also, suggest adding some of the original literature on DMS gas exchange:
doi:10.1029/2004GL021567
doi:10.1029/2009GL041203
doi:10.1029/2010JC006526
line 54. Again, the linear relationship is between the gas transfer velocity K and wind speed, not between flux and wind speed. Also, please add doi:10.1029/2009GL041203
doi:10.1029/2010JC006526 here
line 63. There are a few other schemes in addition to Anderson et al. 2001 that makes use of Chla. Why only Anderson et al. 2001? Why not adding e.g. Gali et al. 2018?
Line 85. That simulation is 10 years long has been said a couple of times now, but without explanation. Presumably it’s to coincide with DMS observations.
Line 106. So if I’m understanding correctly, this is a climatology because Chla and nutrients in UKESM1 are climatological?
Line 119-123. This motivation should’ve been laid out already in intro, and doesn’t warrant repeating here.
Line 126. Can you add/cite an accuracy number here? i.e. x%?
Line 132. A lot of this paragraph could’ve been laid out in intro.
Line 141. some of this has been talked about already in intro. I also find the rest of this paragraph wordy. Table 2 summarizes everything well. The main message here is just to offer a range of emissions.
Line 153-154. Sc is the ratio between kinematic viscosity and molecular diffusivity. It’s more accurately described as how well a gas diffuses relative to the diffusivity of momentum.
Eq. 3. Specify unit for T.
Line 159. Repetitive
Eq 4, 5 etc. specify units for Kw
Figure 1 y-axis label is incorrect. It’s gas transfer velocity, not flux
Line 209. Does this imply the mean Chla in MEDUSA is very different from the mean Chla from MODIS? Ok this is explained in the following paragraph..
Figure 3. grey is derived from observations. Colors are the model simulations?
Line 262. Chla from MODIS is monthly, while the comparison with in situ data is I think on hourly timescales. In reality Chla is not constant over a month. How does this mismatch in timescale affect the comparison?
Comparison in atmospheric DMS: measurements were taken from near the sea surface (e.g. 20 m height). What height is the model atmospheric DMS taken from?
Line 284. Looking at Figs. 3 and 4, I’m not sure if ‘good’ is the right word here. Perhaps better to say something like ‘agree within x% in the mean’
Line 299, 302. Again, Kw (not flux) parametrization
Line 303, 304. Sentence unclear
Figure 5. why does MEDUSALM86 have quite a large stdev in flux, when the simulated seawater DMS from MEDUSA has the smallest stdev?
Line 324-325. Is this statement about nucleation up to date? Does it consider tertiary nucleation involving amines/ammonia, for example? Also, the background aerosol surface area and thus condensation sink must be key here.
Line 335-337. I’m not sure that this is due to the linear relationship between Kw and wind speed. Because the monthly Chla field doesn’t capture the true variability in Chla, it seems likely that some peak biological events will be missed. Furthermore Chla is far from a perfect descriptor of seawater DMS, as different species of marine biota, bloom senescence, grazing, etc can all affect seawater DMS concentrations.
Figure 6. The authors have been arguing about the benefits in including temporal variability via using the MODIS Chla. Yet the comparison here is vs. mean observed atmospheric DMS, much of which lies outside of the simulation window. Why not also show the equivalent of Figures 3, 4, B1, C1, but for atmospheric DMS?
Figure 7. I thought MEDUSA has higher Chla than MODIS-DMS, so how come the latter produces higher primary organics? In general I find the discussion about primary organics off-topic. I suggest focusing the discussion on how using MODIS Chla may improve the representation of DMS emission over climatology, and what the remaining shortcomings/uncertainties are in this approach. i.e. how well can DMS be simulated when Chla is known, and at what temporal/spatial scales? What about other factors including phytoplankton speciation, bacterial production, zooplankton grazing, viral lysis?
See also doi: 10.3389/fmars.2020.596763
Also, can the authors validate their simulations against other parameters such as aerosol non-seasalt sulfate, aerosol number concentration, CCN, AOD?
Line 399-400. This seems correct but is in conflict with abstract
Line 402. Why is that, given the fact that the flux should be more directly linked to atmospheric DMS than is seawater
Line 407. I agree with this statement, yet the authors subsequently advocate for the use of LM86, which is neither recent nor specially suitable for DMS.
Figure D1. SOExchange isn’t the right acronym and data source isn’t cited. It’s SO-GasEx (doi:10.1029/2010JC006526). Also for (j) and (k) here it’s not obvious if the authors are showing observations or simulations.
Citation: https://doi.org/10.5194/egusphere-2023-868-RC2 - AC1: 'Authors' response to reviewers on egusphere-2023-868', Yusuf Bhatti, 28 Sep 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-868', Anonymous Referee #1, 01 Jun 2023
General comments
The topic of the study is important and urgent (https://doi.org/10.1525/elementa.2022.00130) and the study is a step toward improving a component of Earth System models that is not well constrained. The study quantified the sensitivity of atmspheric DMS to oceanic DMS source and flux parameterization over the Southern Ocean. While these results are informative, there are a few issues that need to be addressed to support some of the conclusions made in the study. I provide these issues with suggestions on how they can be addressed below. I also provide specific comments below my general comments.
The choice of oceanic DMS parameterization: The authors chose Anderson et al. (2001) for deriving DMS from chlorophyll-a, while there are other algorithms to derive DMS from chlorophyll-a and other environmental variables developed since 2001.This choice needs to be explained/justified in the manuscript.
Investigation of inter-annual variability: The authors conclude that interannual variability is important (e.g. L18, L416), but this is not demonstrated well in the study. Figure A1 is the only result where the interannual variability can be discussed, bit it gives an impression that the interannual variability is not important (the year-to-year variation of the mean DMS concentration is subtle). To assess the impact of representing interannual variability in the DMS source, the authors could do an additional simulation in which MODIS-DMS does not change from year to year. By comparing this simulation with one of the MODIS simulations in Table 2, the impact can be quantified and its importance can be justified.
Investigation of spatio-temporal variability: The authors introduce the importance of spatio-temporal variability (L57) of DMS in the Southern Ocean, but the results are mostly discussed for spatially and temporally averaged values. Furthermore, some model results are averaged over the entire Southern Ocean, which are compared to the observations that are representative of specific regions/seasons (e.g. L312, L318, L320, L323, L352). While apple-to-apple comparison may not be possible, there are places for improvement. For example, the spatial maps of the simulated DMS flux and atmospheric DMS (such as done for oceanic DMS concentration as in Fig 2a-d) can be added to better compare with the observations from specific regions. These figures will also demonstrate the importance of spatial variability.
Writing style: Section 5 seems like a combination of discussion and conclusions, so I suggest to change the section title to “Discussion and conclusions”. However, I also noted in my specific comments that there are paragraphs in the Results section that I feel belong to Discussion.
Specific comments
L3: Suggest to delete or rewrite this sentence. The authors did not assess how DMS emissions change when the chlorophyll-a distribution is altered. This would require an investigation of the relationship between the DMS emissions and the chlorophyll-a distribution.
L9: “The mean spread” of which quantity?
L11: “The choice of” Is this conclusion correct? The previous sentences indicate that the choice of the oceanic DMS source (112%) has a larger influence on the atmospheric DMS than that of the flux parameterization (50-60%).
L12: Suggest to rewrite this sentence. It is not that linear relationships compare better to observations, but assuming/simulation with the linear dependence of the DMS flux on wind speed results in better representation of atmospheric DMS distribution.
L15: Where is the evidence for the closest agreement? This could be clarified in "Discussion and conclusions".
L34: Suggest to replace “mechanisms, with varying focus” by “approaches that are dependent”.
L40: Suggest to replace “approximate” by “prescribe”.
L48: Suggest to repalce “and has” by “and this emission has”.
L55: Older quadratic approaches are based on other gases? If so, this could be mentioned here to emphasize the issue.
L57: Should “observations” be “concentrations”? Observations are highly variable implies that some regions/seasons are observed more than the others. I am not sure if this is what the authors meant to state here.
L59: Suggest to replace “real-world” by “remotely-sensed”.
L61: Suggest to either remove “nudged to observation” or replace it by “nudged-to-observation".
L68: “the relative importance” to what?
L73: Suggest to move the description of MEDUSA later when the oceanic DMS dataset is introdued. The model used here is the atmopsheric-only configuration, so the oceanic component does not need to be introduced here.
L85: In addition to the details provided in this section, I suggest to add information on boundary conditions such as sea surface temperature data used for these simulations that affect both the atmospheric circulation and DMS emissions (Eq 3).
L86: Suggest to rewrite “biologically productive” to something else. “Biological” is a bit ambiguous, as it is broad. It could refer to microalgae to big animals that have different seasonality. For clarity, it may be better to refer to DMS instead of biology.
L88: This paragraph belongs somewhere else, as it is not a model description.
L110: "modified” implies that the parameterization was modified, but my impression is that the authors only modified the chlorophyll dataset. If so, please rewrite this sentence.
L115: Suggest to define the acronym MODIS when it appears for the first time in the text instead of here.
L117: How are DMS sources and emissions from sea ice covered areas treated? Small gap of 1% probably excludes ice-covered grid cells. DMS emissions from marginal ice zone can be important (https://doi.org/10.1525/elementa.2020.00113).
L148: It is unclear how applying W14 and B17 parameterizations leads to testing the lower limits.
L158: Related to my earlier comment on L85, was HadISST used also for the boundary conditions for the atmospheric circulation simulations?
L172: Suggest to replace “validate” by “evaluate”.
L201: Suggest to replace “includes chlorophyll-a” by “accounts for chlorophyll-a variations”.
L209: Suggest to move “respectively” to “used, respectively)”.
Fig 3: For readers not familiar with violin plots, the caption needs a brief description of what violin plots represent in general.
L249: Suggest to either delete “and would be expected to be highly biased” or clarify what it is meant by highly biased. Also for fair comparison, when was the TAN1802 data taken (before/during/after blooms)?
L265: Suggest to add a few words to briefly explain why higher correlations are expected at higher latitude.
L266: The fact that the correlation in the Southern Ocean is high (0.75) does not necessarily support the previous statement. It needs to show the latitudinal dependence (whether the correlations increase with latitude).
L270: Half sounds many, but in fact only two (out of four), I think? It might be good to also discuss the lack of DMS models in current ESMs. Also, this paragraph sounds more like discussion than results.
L273: Suggest to replace “realistic” by “readily available”.
L285: Is Figure 3 adequate to make the claim that MODIS-DMS is adequate?
Fig 5: Suggest to utilize color for distinguishing different DMS sources
L301-302: The two sentences say the same thing. Suggest to remove one. Also, this finding (the choice of DMS source is more important than the choice of flux parameterization) is contradictory to what is written in the abstract (L11).
L303: It is unclear what this sentence is referring to.
L313: Why not compare the model results averaged over the same region as Webb et al. (2019) instead of comparing with the Southern Ocean averages for apple-to-apple comparison?
L328: “is likely positively skewed” can be checked to see if the model agrees with this finding by looking at the spatial distribution.
L332: This paragraph seems more appropriate to be a part of Discussion and conclusions, instead of Results.
Sec 3.3: How does the distribution of atmospheric DMS differ from that of oceanic DMS emission?
Figure 6: Is this model result representative of the Southern Ocean or the three observational stations?
L365 and Fig 7: Suggest to delete the results of PMOA. As the authors stated (L378), the results of PMOA require further investigation that does not seem to fit into the scope of this study. Adding this result would require additional info on intro, methods, and discussion.
L374: Suggest to rewrite this sentence. It states about CMIP6 but the paper cited is from 2011, many years before CMIP6.
L385: Suggest to cite a paper that compares nudged vs non-nudged runs to support this argument.
L402: How is this different from L399?
L420: Why recommend Lana while there is an updated climatology?
Appendix A: Figures should be labeled as A1-4 instead of A-D1
Citation: https://doi.org/10.5194/egusphere-2023-868-RC1 -
RC2: 'Comment on egusphere-2023-868', Mingxi Yang, 26 Jul 2023
Understanding and being able to simulate DMS emission in the Southern Ocean is critical for constraining aerosols, clouds, and so climate over that region. This paper tests a range of DMS emission estimates within an earth system model and examines the sensitivities in the simulated atmospheric DMS. They argue that using a satellite monthly chlorophyl to drive seawater DMS concentration is superior to using a climatological seawater DMS field, as the former includes more (interannual) variability. They further state that using a linear relationship between gas transfer velocity and wind speed is unsurprisingly better than using a quadratic relationship.
In general, the authors have explored an important topic. However I don’t feel like I’ve learned a lot from the paper because the analysis is incomplete. I think the paper may be improved by considering the following points:
- That using a satellite monthly chlorophyl to drive seawater DMS concentration is superior to using a climatological seawater DMS field seems not unreasonable in principle. However the validations using only two cruises are not very convincing to me. Can the authors include more validation datasets? For example the SO-GasEx dataset is within the 10-year window and includes seawater/atmospheric DMS as well as DMS flux.
- There have been more recent seawater DMS parametrizations than Anderson et al. 2001. The authors should include them or explain why they aren’t considering these more recent developments. They should also acknowledge that seawater DMS doesn’t just depend on Chla, but is also sensitive to a number of other biological parameters.
- Comparisons in atmospheric DMS neglects the time component entirely. Maybe the authors can further expand their 10-year window to facilitate comparisons with those earlier and very recent atmospheric DMS measurements?
- What are the impacts of these different DMS emission simulations on aerosols, CCN, AOD, etc? Can authors validate those?
Specific comments:
Please make careful distinction between flux and Kw. The authors have incorrectly interchanged their usage a number of times in the paper.
Overall the writing isn’t very concise. I suggest further proof-reading from all authors.
Line 11-12. Why, given that the spread is greater among different oceanic DMS datasets?
Line 12-15. Be more exact. It’s a linear relationship between gas transfer velocity and wind speed. More so than between flux (emission) and wind, since flux also depends on seawater DMS concentration
Line 25-26. DMS isn’t just produced by phytoplankton. There’s also bacterial production. So suggest replacing ‘phytoplankton’ with ‘marine biota’ and update references accordingly.
Line 30. This paragraph talks about how ESMs represent DMS emission. But I think it’d be better to first talk about how DMS emission is estimated in general. i.e. flux = Kw * deltaC ~= Kw * [seawaterDMS]. Then you can talk about the different parametrizations of Kw (the gas transfer velocity), as well as the different ways [seawaterDMS] is estimated.
Line 38-39. Not sure what ‘simulate biases’ means here
Line 47. Suggest adding ‘to some degree’ after ‘events’, given the complexity in biological response.
Line 49. Not between atmospheric DMS and wind speed, but between the gas transfer velocity and wind speed.
Some of the references aren’t the most appropriate:
Vlahos and Monahan 2009 interpreted other groups’ measurements
Wanninkhof 1992, 2014 focused on CO2, not DMS. Nightingale et al. 2000 and Ho et al. 2006 focused on 3He/SF6, not DMS. Liss & Merlivat 1986 synthesized a range of lab/field works, but not specifically of DMS. Those parametrizations are often used in ESMs for estimating DMS flux, inappropriately. So it’s useful to introduce them here, but with a clarification.
Also, suggest adding some of the original literature on DMS gas exchange:
doi:10.1029/2004GL021567
doi:10.1029/2009GL041203
doi:10.1029/2010JC006526
line 54. Again, the linear relationship is between the gas transfer velocity K and wind speed, not between flux and wind speed. Also, please add doi:10.1029/2009GL041203
doi:10.1029/2010JC006526 here
line 63. There are a few other schemes in addition to Anderson et al. 2001 that makes use of Chla. Why only Anderson et al. 2001? Why not adding e.g. Gali et al. 2018?
Line 85. That simulation is 10 years long has been said a couple of times now, but without explanation. Presumably it’s to coincide with DMS observations.
Line 106. So if I’m understanding correctly, this is a climatology because Chla and nutrients in UKESM1 are climatological?
Line 119-123. This motivation should’ve been laid out already in intro, and doesn’t warrant repeating here.
Line 126. Can you add/cite an accuracy number here? i.e. x%?
Line 132. A lot of this paragraph could’ve been laid out in intro.
Line 141. some of this has been talked about already in intro. I also find the rest of this paragraph wordy. Table 2 summarizes everything well. The main message here is just to offer a range of emissions.
Line 153-154. Sc is the ratio between kinematic viscosity and molecular diffusivity. It’s more accurately described as how well a gas diffuses relative to the diffusivity of momentum.
Eq. 3. Specify unit for T.
Line 159. Repetitive
Eq 4, 5 etc. specify units for Kw
Figure 1 y-axis label is incorrect. It’s gas transfer velocity, not flux
Line 209. Does this imply the mean Chla in MEDUSA is very different from the mean Chla from MODIS? Ok this is explained in the following paragraph..
Figure 3. grey is derived from observations. Colors are the model simulations?
Line 262. Chla from MODIS is monthly, while the comparison with in situ data is I think on hourly timescales. In reality Chla is not constant over a month. How does this mismatch in timescale affect the comparison?
Comparison in atmospheric DMS: measurements were taken from near the sea surface (e.g. 20 m height). What height is the model atmospheric DMS taken from?
Line 284. Looking at Figs. 3 and 4, I’m not sure if ‘good’ is the right word here. Perhaps better to say something like ‘agree within x% in the mean’
Line 299, 302. Again, Kw (not flux) parametrization
Line 303, 304. Sentence unclear
Figure 5. why does MEDUSALM86 have quite a large stdev in flux, when the simulated seawater DMS from MEDUSA has the smallest stdev?
Line 324-325. Is this statement about nucleation up to date? Does it consider tertiary nucleation involving amines/ammonia, for example? Also, the background aerosol surface area and thus condensation sink must be key here.
Line 335-337. I’m not sure that this is due to the linear relationship between Kw and wind speed. Because the monthly Chla field doesn’t capture the true variability in Chla, it seems likely that some peak biological events will be missed. Furthermore Chla is far from a perfect descriptor of seawater DMS, as different species of marine biota, bloom senescence, grazing, etc can all affect seawater DMS concentrations.
Figure 6. The authors have been arguing about the benefits in including temporal variability via using the MODIS Chla. Yet the comparison here is vs. mean observed atmospheric DMS, much of which lies outside of the simulation window. Why not also show the equivalent of Figures 3, 4, B1, C1, but for atmospheric DMS?
Figure 7. I thought MEDUSA has higher Chla than MODIS-DMS, so how come the latter produces higher primary organics? In general I find the discussion about primary organics off-topic. I suggest focusing the discussion on how using MODIS Chla may improve the representation of DMS emission over climatology, and what the remaining shortcomings/uncertainties are in this approach. i.e. how well can DMS be simulated when Chla is known, and at what temporal/spatial scales? What about other factors including phytoplankton speciation, bacterial production, zooplankton grazing, viral lysis?
See also doi: 10.3389/fmars.2020.596763
Also, can the authors validate their simulations against other parameters such as aerosol non-seasalt sulfate, aerosol number concentration, CCN, AOD?
Line 399-400. This seems correct but is in conflict with abstract
Line 402. Why is that, given the fact that the flux should be more directly linked to atmospheric DMS than is seawater
Line 407. I agree with this statement, yet the authors subsequently advocate for the use of LM86, which is neither recent nor specially suitable for DMS.
Figure D1. SOExchange isn’t the right acronym and data source isn’t cited. It’s SO-GasEx (doi:10.1029/2010JC006526). Also for (j) and (k) here it’s not obvious if the authors are showing observations or simulations.
Citation: https://doi.org/10.5194/egusphere-2023-868-RC2 - AC1: 'Authors' response to reviewers on egusphere-2023-868', Yusuf Bhatti, 28 Sep 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 579 | 442 | 22 | 1,043 | 16 | 19 |
- HTML: 579
- PDF: 442
- XML: 22
- Total: 1,043
- BibTeX: 16
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cited
Yusuf Bhatti
Laura Revell
Alex Schuddeboom
Adrian McDonald
Alex Archibald
Jonny Williams
Abhijith Venugopal
Catherine Hardacre
Erik Behrens
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3034 KB) - Metadata XML