the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Description and performance of the CARMA sectional aerosol microphysical model in CESM2
Michael J. Mills
Yunqian Zhu
Charles G. Bardeen
Francis Vitt
Pengfei Yu
David Fillmore
Xiaohong Liu
Brian Toon
Terry Deshler
Abstract. We implemented the Community Aerosol and Radiation Model for Atmospheres (CARMA) in both the high and low-top model versions of the Community Earth System Model Version 2 (CESM2). CARMA is a sectional microphysical model, which we use for aerosol in both the troposphere and stratosphere. CARMA is fully coupled to chemistry, clouds, radiation, and transport routines in CESM2. This development enables the comparison of simulations with a sectional (CARMA) and a modal (MAM4) aerosol microphysical model in the same modeling framework. The new implementation of CARMA has been adopted from previous work with some additions that align with the current CESM2 MAM4 implementation. The main updates include an interactive secondary organic aerosol description in CARMA using the volatility basis set (VBS) approach, updated wet removal, and the use of transient emissions of aerosols and trace gases. In addition, we implemented an alternative aerosol nucleation scheme in CARMA, which is also used in MAM4. Detailed comparisons of stratospheric aerosol properties after the Mt Pinatubo eruption reveal the importance of prescribing sulfur injections in a larger region rather than in a single column to better represent the observed evolution of aerosols. Both CARMA and MAM4 in CESM2 are able to represent stratospheric and tropospheric aerosol properties reasonably well compared to observations. Several differences in the performance of the two aerosol models show, in general, an improved representation of aerosols using the sectional aerosol model in CESM2. These include a better representation of the aerosol size distribution after the Mt Pinatubo volcanic eruption in CARMA compared to MAM4. MAM4 produces on average smaller aerosols and less removal than CARMA, which results in a larger total mass. Both CARMA and MAM4 reproduce stratospheric Aerosol Optical Depth (AOD) within the errorbar of the observations between 2001 and 2020, except for recent larger volcanic eruptions that are overestimated by both model configurations. The CARMA background surface area density and aerosol size distribution in the stratosphere and troposphere compare well to observations, with some underestimation of the Aitken-mode size range. MAM4 shows shortcomings in reproducing coarse-mode aerosol distributions in the stratosphere and troposphere. This work outlines additional development needs for CESM2 CARMA to improve the model compared to observations in both the troposphere and stratosphere.
- Preprint
(9620 KB) - Metadata XML
- BibTeX
- EndNote
Simone Tilmes et al.
Status: closed
-
RC1: 'Comment on gmd-2023-79', Anonymous Referee #1, 28 Jul 2023
Review of “Description and performance of the CARMA sectional aerosol microphysical model in CESM2” by Tilmes et al. for publication in Geoscientific Model Development.
The paper provides a comparative evaluation of the performance of two aerosol microphysical schemes in simulations performed with two different configurations of the Community Earth System Model (CESM2). In one the sectional CARMA module is compared to the modal MAM4 module in the high-top version of CESM2 called WACCM-MA and applied to study of the 1991 Mount Pinatubo volcanic eruption and subsequent period. In the other the two aerosol modules are run in the low-top CAMchem configuration and applied to simulations of tropospheric and lower stratospheric aerosol in the context of the 2016-2018 ATom airborne observations. It is shown that for Pinatubo the performance of both CARMA and MAM4 are improved by distributing emissions over a region instead of prescribing over the volcano column alone. CARMA results in overall larger particles following Pinatubo, with a correspondingly shorter atmospheric lifetime. Two different nucleation schemes are tested in CARMA, its default Zhao scheme and the Vehkamäki scheme used in MAM4. Overall CARMA’s performance is better with its default scheme. Tropospheric aerosols are compared to MODIS AOD and ATom profile observations. Modeled profiles are generally similar between CARMA and MAM, each with its strengths and weaknesses in terms of species and regions. CARMA somewhat overestimates the mid-tropospheric sulfate relative to ATom, and MAM has consequently better agreement in the extinction profile. CARMA is shown to have overall better agreement with observed particle size distributions for both Pinatubo and ATom cases.
The paper is well organized and relatively comprehensive. It is suitable for publication with minor revisions. I suggest a couple of points below that could be expanded on to improve the discussion, and suggest numerous minor points.
Major points
Line 183 refers to sub-stepping for stability in process rates. Some additional information would be useful here, particularly as to how it impacts performance and accuracy. Do MAM4 and CARMA take similar approaches? Do they have similar accuracy tolerances? What are those?
Section 2.4.5 could be more explicit with respect to the size distributions of emissions, specifically how they compare to each other between CARMA and MAM. Lifetime numbers reported in Table 4 are more typical of AeroCom models in MAM than they are for CARMA, which is remarkably short for both dust and sea salt. This could be a consequence of emissions being strongly weighted toward largest size bins in CARMA. Some further detail here would be helpful instead of just referring to previous work. The discussion in 5.2.1 expands a bit more on this as well. The discussion seems a bit simplistic. It would be better to present in terms of efficiencies of processes instead of references to burdens. I know there is non-linearity in the microphysics, but in terms of the sink processes they are quasi linear in terms of the burden, so it’s not very illuminating to frame it that way. Instead, what seems clear, is that CARMA prefers dry deposition, which necessarily leaves less aerosol for wet removal. MAM is more the opposite.
Minor points
Line 102: For completeness, what is the mode width for the primary carbon mode?
Line 156: “Therefore the wet radius below 190 K” is a sentence fragment of unclear intent.
Line 169: s/b McFarlane
Line 171: “two moment cloud microphysics (?)” Unclear what the (?) means, maybe a missing reference?
Line 209: s/b McFarlane
Line 255: Here and elsewhere this note about photolysis of SOA. What does this mean? Is this a destruction pathway, or something to with photolysis chemistry? A reference would help.
Line 284: Suggest “correspondingly” instead of “somewhat” smaller throughput.
Line 3232: Do you really mean QFED here? Citation is Darmenov and da Silva (2015, https://ntrs.nasa.gov/citations/20180005253)
Figure 5, second panel: Since the behavior essentially asymptotes at 30 days, suggest you compress x-axis range to allow better inspection of the first 10 days in this figure, where the models are quite different.
Line 485 and rest of paragraph: Figure 9 is stated but actually Figure 8 is meant.
Line 502: Figure 9 is meant instead of Figure 8.
Table 4: Caption states lifetime given in years. This is incorrect, it should be days.
Line 537: What MODIS products are used here?
Figure 11: Positional labeling in caption caption (top, middle, bottom…) would make sense if there are six panels, but there are eight. Suggest label a, b, c, …
Line 576: “settling” instead of “settlement”
Line 582: Here and rest of paragraph Figure 11 actually s/b Figure 13 and Figure 12 actually s/b Figure 14
Citation: https://doi.org/10.5194/gmd-2023-79-RC1 -
RC2: 'Comment on gmd-2023-79', Anonymous Referee #2, 07 Aug 2023
The manuscript “Description and performance of the CARMA sectional aerosol microphysical model in CESM2” provides a comprehensive evaluation of the CARMA sectional model for stratospheric and tropospheric conditions comparing the model to the modal model MAM4 as well as observations. The model performance for describing stratospheric aerosol properties was evaluated by simulating the massive volcanic eruption of Mt. Pinatubo. The choice of applying the model for such an extreme case is appropriate as simulating aerosol properties in such conditions will be strongly affected by the numerical methods chosen for solving aerosol microphysical processes. For example, modal and sectional models will result in very different simulated aerosol size distributions and thus radiative properties for aerosol populations. The paper is well written, extremely comprehensive, scientifically sound and I can recommend publishing it after the following minor issues have been addressed:
- As an overall comment, using the same global model host model for using both a model and a sectional model is a good way to compare the two aerosol models in the sense that the differences would originate only from the description aerosol microphyscs. However, it is unfortunate that in these simulations, MAM4 and CARMA implementations had also quite large differences in, for example, online calculated emissions of sea salt and dust which result in very different amounts of emitted aerosol. Such big changes make it difficult to get a good picture on how much MAM4 and CARMA themselves contribute to the differences between the runs.
Specific comments:
Page 4, Line 116: “For the stratosphere, sulfates are in equilibrium with the water, and a weight percent of H2SO4 is calculated based on the parameterization by Tabazadeh et al. (1997).” This is a very difficult sentence to understand. Do you want to say that for the stratosphere the water uptake is calculated using the parameterization by Tabazadeh et al. (1997)?Page 4, Line 103: The choice for such a small geometric standard of 1.2 is not well explained in the manuscript. Later in the manuscript, it is briefly mentioned that it is required to be able to simulate stratospheric aerosol from strong volcano eruptions (Niemeier et al., 2011). This should explained already here.
Page 5, Line 125 it is said: “sectional nitrate and ammonium (Yu et al., 2022), are not included in the current version of the model”. Does this mean nitrate and ammonium are treated as bulk compounds of they are fully omitted?
Page 5, Line 132-133: “Currently, CARMA only allows one component of a group to be volatile. The addition of SOA in this model requires calculating SOA volatility (gas to aerosol exchange) in CAM.” I don’t understand these sentences; why is it required to calculate SOA volatility in CAM?
Page 5, Line 148-149: “The model does not currently employ nucleation influenced by ammonia or organics, which is likely important near the ground.” Please add a reference.
Page 6, Line 156: “Therefore the wet radius below 190 K.” There is something missing from this sentence.
Page 7, Line 170: Boundary layer is not a processes
Page 8, Line 200: “It is based on the number of dust and sulfate” Should this be “amount” instead of “number”?
Page 9, Line 240-241: How is cloud-produced sulfate distributed between CARMA bins?
Page 16: “Therefore, the acid molecules are rapidly lost from the gas phase and nucleate much faster.” What does this mean? If sulfuric acid molecules are lost, why do they nucleate faster? I would expect a decrease in gas phase concentration to supress nucleation.
Section 5.1. when comparing modeled AOD to satellite AOD, was the model data collocated to satellite observations?
Page 27: It is speculated that MAM4 overestimates AOD in the southern hemisphere due to too strong sea salt emissions and both models overestimate AOD over South America due to secondary organic aerosol. In addition, some of the differences between the models have been suggested to come from different dust and sea salt emissions. Could this be diagnosed from the model output by breaking the AOD to components, i.e. what fraction of AOD comes from which compound? In addition, aerosol lifetimes are very different between the models and thus dry and wet removal might be even a more significant cause for intermodel differences.
Page 31, Line 600-601: “Differences compared to the observations in the tropical Atlantic mid-troposphere may be related to how CARMA and MAM4 apply wet removal.”. Can you elaborate this?
Page 33, Lines 635-637: “Differences in burden can be a result of the details of the aerosol size distribution, which also leads to differences in stratospheric AOD (as discussed above).” This sentence is ambiguous. Do you mean that the representation of the aerosol size distributuin leads to differences in burdens?
Section 5.2.4: The comparison of size distributions seems a bit unfair to MAM4 since it uses Vehkamäki nucleation scheme which produces a negligible amount of nucleation in the troposphere. Was there a practical reason for not using Zhao et al. nucleation scheme (or any boundary layer nucleation scheme) in MAM4? With this in mind, to have a better simultaneous comparison between observations and the two models, it would make sense to switch places between Figure 19 and A6.
Page 40, Line 777: Please change “This will require a more detailed investigation of the performance of clouds” to something like “This will require a more detailed investigation of the performance of the model to simulate clouds”
Technical comments:
- Page 7, Line 171, reference missing
- Figure 9 is discussed before Figure 8Niemeier, U., Schmidt, H. and Timmreck, C. (2011), The dependency of geoengineered sulfate aerosol on the emission strategy. Atmosph. Sci. Lett., 12: 189-194. https://doi.org/10.1002/asl.304
Citation: https://doi.org/10.5194/gmd-2023-79-RC2 - AC1: 'Comment on gmd-2023-79', Simone Tilmes, 26 Aug 2023
Status: closed
-
RC1: 'Comment on gmd-2023-79', Anonymous Referee #1, 28 Jul 2023
Review of “Description and performance of the CARMA sectional aerosol microphysical model in CESM2” by Tilmes et al. for publication in Geoscientific Model Development.
The paper provides a comparative evaluation of the performance of two aerosol microphysical schemes in simulations performed with two different configurations of the Community Earth System Model (CESM2). In one the sectional CARMA module is compared to the modal MAM4 module in the high-top version of CESM2 called WACCM-MA and applied to study of the 1991 Mount Pinatubo volcanic eruption and subsequent period. In the other the two aerosol modules are run in the low-top CAMchem configuration and applied to simulations of tropospheric and lower stratospheric aerosol in the context of the 2016-2018 ATom airborne observations. It is shown that for Pinatubo the performance of both CARMA and MAM4 are improved by distributing emissions over a region instead of prescribing over the volcano column alone. CARMA results in overall larger particles following Pinatubo, with a correspondingly shorter atmospheric lifetime. Two different nucleation schemes are tested in CARMA, its default Zhao scheme and the Vehkamäki scheme used in MAM4. Overall CARMA’s performance is better with its default scheme. Tropospheric aerosols are compared to MODIS AOD and ATom profile observations. Modeled profiles are generally similar between CARMA and MAM, each with its strengths and weaknesses in terms of species and regions. CARMA somewhat overestimates the mid-tropospheric sulfate relative to ATom, and MAM has consequently better agreement in the extinction profile. CARMA is shown to have overall better agreement with observed particle size distributions for both Pinatubo and ATom cases.
The paper is well organized and relatively comprehensive. It is suitable for publication with minor revisions. I suggest a couple of points below that could be expanded on to improve the discussion, and suggest numerous minor points.
Major points
Line 183 refers to sub-stepping for stability in process rates. Some additional information would be useful here, particularly as to how it impacts performance and accuracy. Do MAM4 and CARMA take similar approaches? Do they have similar accuracy tolerances? What are those?
Section 2.4.5 could be more explicit with respect to the size distributions of emissions, specifically how they compare to each other between CARMA and MAM. Lifetime numbers reported in Table 4 are more typical of AeroCom models in MAM than they are for CARMA, which is remarkably short for both dust and sea salt. This could be a consequence of emissions being strongly weighted toward largest size bins in CARMA. Some further detail here would be helpful instead of just referring to previous work. The discussion in 5.2.1 expands a bit more on this as well. The discussion seems a bit simplistic. It would be better to present in terms of efficiencies of processes instead of references to burdens. I know there is non-linearity in the microphysics, but in terms of the sink processes they are quasi linear in terms of the burden, so it’s not very illuminating to frame it that way. Instead, what seems clear, is that CARMA prefers dry deposition, which necessarily leaves less aerosol for wet removal. MAM is more the opposite.
Minor points
Line 102: For completeness, what is the mode width for the primary carbon mode?
Line 156: “Therefore the wet radius below 190 K” is a sentence fragment of unclear intent.
Line 169: s/b McFarlane
Line 171: “two moment cloud microphysics (?)” Unclear what the (?) means, maybe a missing reference?
Line 209: s/b McFarlane
Line 255: Here and elsewhere this note about photolysis of SOA. What does this mean? Is this a destruction pathway, or something to with photolysis chemistry? A reference would help.
Line 284: Suggest “correspondingly” instead of “somewhat” smaller throughput.
Line 3232: Do you really mean QFED here? Citation is Darmenov and da Silva (2015, https://ntrs.nasa.gov/citations/20180005253)
Figure 5, second panel: Since the behavior essentially asymptotes at 30 days, suggest you compress x-axis range to allow better inspection of the first 10 days in this figure, where the models are quite different.
Line 485 and rest of paragraph: Figure 9 is stated but actually Figure 8 is meant.
Line 502: Figure 9 is meant instead of Figure 8.
Table 4: Caption states lifetime given in years. This is incorrect, it should be days.
Line 537: What MODIS products are used here?
Figure 11: Positional labeling in caption caption (top, middle, bottom…) would make sense if there are six panels, but there are eight. Suggest label a, b, c, …
Line 576: “settling” instead of “settlement”
Line 582: Here and rest of paragraph Figure 11 actually s/b Figure 13 and Figure 12 actually s/b Figure 14
Citation: https://doi.org/10.5194/gmd-2023-79-RC1 -
RC2: 'Comment on gmd-2023-79', Anonymous Referee #2, 07 Aug 2023
The manuscript “Description and performance of the CARMA sectional aerosol microphysical model in CESM2” provides a comprehensive evaluation of the CARMA sectional model for stratospheric and tropospheric conditions comparing the model to the modal model MAM4 as well as observations. The model performance for describing stratospheric aerosol properties was evaluated by simulating the massive volcanic eruption of Mt. Pinatubo. The choice of applying the model for such an extreme case is appropriate as simulating aerosol properties in such conditions will be strongly affected by the numerical methods chosen for solving aerosol microphysical processes. For example, modal and sectional models will result in very different simulated aerosol size distributions and thus radiative properties for aerosol populations. The paper is well written, extremely comprehensive, scientifically sound and I can recommend publishing it after the following minor issues have been addressed:
- As an overall comment, using the same global model host model for using both a model and a sectional model is a good way to compare the two aerosol models in the sense that the differences would originate only from the description aerosol microphyscs. However, it is unfortunate that in these simulations, MAM4 and CARMA implementations had also quite large differences in, for example, online calculated emissions of sea salt and dust which result in very different amounts of emitted aerosol. Such big changes make it difficult to get a good picture on how much MAM4 and CARMA themselves contribute to the differences between the runs.
Specific comments:
Page 4, Line 116: “For the stratosphere, sulfates are in equilibrium with the water, and a weight percent of H2SO4 is calculated based on the parameterization by Tabazadeh et al. (1997).” This is a very difficult sentence to understand. Do you want to say that for the stratosphere the water uptake is calculated using the parameterization by Tabazadeh et al. (1997)?Page 4, Line 103: The choice for such a small geometric standard of 1.2 is not well explained in the manuscript. Later in the manuscript, it is briefly mentioned that it is required to be able to simulate stratospheric aerosol from strong volcano eruptions (Niemeier et al., 2011). This should explained already here.
Page 5, Line 125 it is said: “sectional nitrate and ammonium (Yu et al., 2022), are not included in the current version of the model”. Does this mean nitrate and ammonium are treated as bulk compounds of they are fully omitted?
Page 5, Line 132-133: “Currently, CARMA only allows one component of a group to be volatile. The addition of SOA in this model requires calculating SOA volatility (gas to aerosol exchange) in CAM.” I don’t understand these sentences; why is it required to calculate SOA volatility in CAM?
Page 5, Line 148-149: “The model does not currently employ nucleation influenced by ammonia or organics, which is likely important near the ground.” Please add a reference.
Page 6, Line 156: “Therefore the wet radius below 190 K.” There is something missing from this sentence.
Page 7, Line 170: Boundary layer is not a processes
Page 8, Line 200: “It is based on the number of dust and sulfate” Should this be “amount” instead of “number”?
Page 9, Line 240-241: How is cloud-produced sulfate distributed between CARMA bins?
Page 16: “Therefore, the acid molecules are rapidly lost from the gas phase and nucleate much faster.” What does this mean? If sulfuric acid molecules are lost, why do they nucleate faster? I would expect a decrease in gas phase concentration to supress nucleation.
Section 5.1. when comparing modeled AOD to satellite AOD, was the model data collocated to satellite observations?
Page 27: It is speculated that MAM4 overestimates AOD in the southern hemisphere due to too strong sea salt emissions and both models overestimate AOD over South America due to secondary organic aerosol. In addition, some of the differences between the models have been suggested to come from different dust and sea salt emissions. Could this be diagnosed from the model output by breaking the AOD to components, i.e. what fraction of AOD comes from which compound? In addition, aerosol lifetimes are very different between the models and thus dry and wet removal might be even a more significant cause for intermodel differences.
Page 31, Line 600-601: “Differences compared to the observations in the tropical Atlantic mid-troposphere may be related to how CARMA and MAM4 apply wet removal.”. Can you elaborate this?
Page 33, Lines 635-637: “Differences in burden can be a result of the details of the aerosol size distribution, which also leads to differences in stratospheric AOD (as discussed above).” This sentence is ambiguous. Do you mean that the representation of the aerosol size distributuin leads to differences in burdens?
Section 5.2.4: The comparison of size distributions seems a bit unfair to MAM4 since it uses Vehkamäki nucleation scheme which produces a negligible amount of nucleation in the troposphere. Was there a practical reason for not using Zhao et al. nucleation scheme (or any boundary layer nucleation scheme) in MAM4? With this in mind, to have a better simultaneous comparison between observations and the two models, it would make sense to switch places between Figure 19 and A6.
Page 40, Line 777: Please change “This will require a more detailed investigation of the performance of clouds” to something like “This will require a more detailed investigation of the performance of the model to simulate clouds”
Technical comments:
- Page 7, Line 171, reference missing
- Figure 9 is discussed before Figure 8Niemeier, U., Schmidt, H. and Timmreck, C. (2011), The dependency of geoengineered sulfate aerosol on the emission strategy. Atmosph. Sci. Lett., 12: 189-194. https://doi.org/10.1002/asl.304
Citation: https://doi.org/10.5194/gmd-2023-79-RC2 - AC1: 'Comment on gmd-2023-79', Simone Tilmes, 26 Aug 2023
Simone Tilmes et al.
Simone Tilmes et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 397 | 230 | 12 | 639 | 4 | 6 |
- HTML: 397
- PDF: 230
- XML: 12
- Total: 639
- BibTeX: 4
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1