Evaluation of atmospheric sulfur dioxide simulated with the EMAC (version 2.55) Chemistry-Climate Model using satellite and ground-based observations

Makroum, Ismail; Jöckel, Patrick; Dameris, Martin; Theys, Nicolas; De Leeuw, Johannes

doi:10.5194/egusphere-2025-3915

Preprints

https://doi.org/10.5194/egusphere-2025-3915

Preprints

05 Sep 2025

| 05 Sep 2025

Evaluation of atmospheric sulfur dioxide simulated with the EMAC (version 2.55) Chemistry-Climate Model using satellite and ground-based observations

Ismail Makroum, Patrick Jöckel, Martin Dameris, Nicolas Theys, and Johannes De Leeuw

Abstract. Sulfur dioxide (SO₂) is a key atmospheric pollutant, primarily emitted through human activities such as fossil fuel combustion. In atmospheric models, accurate representation of SO₂ emission sources, transport, and removal processes are essential for evaluating air quality and radiative forcing.

In this study, we present, for the first time, a comprehensive examination of atmospheric SO₂ simulated by the ECHAM/MESSy Atmospheric Chemistry (EMAC) model. First, the tropospheric sulfur budget simulated by EMAC is verified to be close, that is, all sulfur sources and sinks are balanced, ensuring no artificial gain or loss occurs over time due to numerical or conceptual errors. This budget closure is a prerequisite for any further analysis. Second, the results of EMAC simulations are compared with observations from three ground-based networks (the Clean Air Status and Trends Network (CASTnet), the European Monitoring and Evaluation Program (EMEP), and the Acid Deposition Monitoring Network in East Asia (EANET)), mainly over polluted regions, and with vertical column densities retrieved from a TROPOspheric Monitoring Instrument (TROPOMI) on board the Copernicus Sentinel-5 Precursor mission (Sentinel-5P) satellite. The EMAC simulated SO₂ concentrations near the Earth’s surface for the year 2019 are, depending on the region, between 1.4 and 1.8 times larger than observed. This discrepancy aligns well with the differences between simulated and retrieved satellite-based measurements of SO₂ vertical column densities over the same regions. It indicates that the prescribed SO₂ emissions used for the EMAC simulations might be overestimated. Over a longer time period (2000–2019), the EMAC simulation reproduces the measured declining trends of SO₂ concentrations and deposited sulfur fluxes in the USA and Europe, but fails to simulate the observed trends in East Asia. This is most likely attributable to the prescribed SO₂ emission inventories. Furthermore, sensitivity simulations are performed to assess the emitted amount of SO₂ following the Raikoke and Ulawun volcanic eruptions in 2019. The results show a very good agreement of the simulated temporal evolution of the amount of atmospheric SO₂ after the eruptions with that retrieved from satellite-based observations.

Received: 11 Aug 2025 – Discussion started: 05 Sep 2025

Competing interests: At least one of the (co-)authors is a member of the editorial board of Geoscientific Model Development.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Ismail Makroum, Patrick Jöckel, Martin Dameris, Nicolas Theys, and Johannes De Leeuw

Status: closed

RC1:
'Comment on egusphere-2025-3915', Anonymous Referee #1, 06 Oct 2025
General comments
This manuscript is a thorough evaluation of the EMAC v2.55 sulfur simulations and makes a useful contribution by (i) closing a model‑internal sulfur budget (ii) documenting how the model compares with satellite data in 2019 to evaluate how it responds to volcanic emissions and (iii) evaluating against long‑term measurements, 2010–2019
The paper is well organized and generally clear. However, it is unnecessarily long, and the presentation of results is at times too detailed, making it difficult to extract the main messages and scientific significance. The description of the model setup partly repeats work published elsewhere, and it is not entirely clear what is new compared to earlier model versions (e.g., Jöckel et al., 2016). For the interactive gas–particle chemistry, more detail would be beneficial, as the current description is incomplete for interpreting SO₂ lifetime and deposition. I.e. the statement that “the simulation did not involve an interactive aerosol submodel” needs clarification. Does this mean that interactions with ammonia are excluded? If so, this should be explicitly stated, as ammonia strongly influences sulfur oxidation pathways, cloud pH, and the partitioning and deposition of sulfur.
Specific comments
For the emissions it is important to notice that earlier global inventory for China have underestimated the reductions after around 2010. I am not sure if CMIP6 emissions have taken this into account? In Line 35 it is written that the emissions in China remain high, which is true, but the authors should also mention the substantial reductions in recent years.

Why is the comparison made with EDGAR 5 instead of the more recent EDGAR 6.1 (2024)? Using EDGAR 6.1 would enable comparison over the full 2000–2019 period. It would strengthen the analysis to run short sensitivity tests (1–2 simulations) using alternative inventories (e.g., EDGAR 6.1) to quantify whether biases in CMIP6 emissions explain the overestimation. If such tests are beyond the study’s scope, this limitation should at least be discussed.

Why are you not comparing to the more recent EDGAR6.1 (from 2024) instead of EDGAR5? Then you would be able to compare it with the whole 2000-2019 period. It would have been useful to run the model with EDGAR and not only compare the emissions inventory to evaluate more directly if it biases in the CMIP6 emissions that cause the overestimation. Not necessarily a full rerun of the model but quantify the bias by re‑running 1–2 short sensitivity test with different emissions. Though I understand if this is beyond the capacity of this work.

Sulfur data from Africa (INDAAF; https://indaaf.obs-mip.fr) and from Canada (CAPMoN) could have been included to provide a more complete global picture. In the U.S., CASTNET is responsible for air and aerosol data, whereas wet deposition data originate from the National Atmospheric Deposition Program (NADP). It appears NADP data were used, but this is not stated explicitly. NADP also includes many more sites than those used here.

The model response to volcanic emissions is evaluated only against TROPOMI satellite data. You write that “it is difficult to ascertain whether the differences originate near the surface or higher up in the atmosphere.” It would strengthen the analysis to use in-situ data from 2019, which are available for that year and would allow a more direct comparison.

It is not written how trends were calculated, i.e. linear regression, Sen’s Slope . Mann Kendall. This should be added to the methods

For trend comparisons, it would be appropriate to refer to regional studies covering similar periods. Some suggestions:
In Europe recent work done by EMEP: https://doi.org/10.4209/aaqr.230237. Seems like you have somewhat larger trend for SO2 (0.05 ug/m3/y) compared to EMEP (0.034 ug/m3/y). The difference may stem from site selection, trend method, or emissions used.

Several studies in North America. Eg.: NADP data: https://doi.org/10.1016/j.atmosenv.2023.119783, CASTNet: https://doi.org/10.5194/acp-22-12749-2022 and from Canada by CAPMoN: https://doi.org/10.5194/acp-22-14631-2022

Japan: https://doi.org/10.1016/j.envpol.2021.117842

China: https://doi.org/10.3390/su17198815

Technical corrections/spelling errors
Line 70: “caclulated” to calculated

Line 80-81: “sun‑synchrinously” to sun‑

Line 89: “isopren to isoprene.

Line 146: “histrotical” to historical.

Line 149: “soley” to solely.

Line 230: “comparatative“ to comparative.

Line 260: “correpsonding” to corresponding.

Line 294: “drived“ to derived

Line 328: “depostion“ to deposition

Line 364: “heve” to have.

Line 543; “cylce“ to cycle

Line 604: “retrieveal” to retrieval.

Line 658: “resepctively“ to respectively.

Line 723: “anlyses” to analyses

Line 724: “caclualted” to calculated

Line 746 “undr” to under. Multiple instances of “acces” and “reslts” to be replaced with access and results.

Table 2 caption: “aersol” to aerosol.

Figure caption 12. EMEP should be replaced with EANET

Standardize Tg(S)/yr vs Tg(S)/a

There is a mix of units used for air concentration and fluxes (e.g., µg m⁻³ vs mmol m⁻² and kgS/ha). These should be standardized throughout the paper for clarity and comparability.
Citation: https://doi.org/10.5194/egusphere-2025-3915-RC1
- AC2: 'Reply on RC1', Patrick Jöckel, 26 Nov 2025
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-3915/egusphere-2025-3915-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2025-3915-AC2
RC2:
'Comment on egusphere-2025-3915', Anonymous Referee #2, 07 Oct 2025

This manuscript comprehensively evaluates sulfur dioxide in the EMAC chemistry-climate model. This is done by first verifying that the sulfur cycle is closed, meaning mass is neither lost nor gained. The SO2 depositions and vertical column densities are then evaluated with a set of ground-based and satellite observations. The structure of the manuscript is generally very well laid out and the methods are explained thoroughly and well executed. However, there are some areas where more concrete explanations are needed.

P4, L90: The authors mention that the aerosol is not interactive in the simulations used for this study and refer to the aerosol radiative effects and heterogeneous chemistry. Is sulfate aerosol represented in some form in these simulations? If so, please state for clarity. If not, what does this imply for the sulfur cycle?

P16, Figure 5: Looking at this figure, the SO2 burden appears to decline exponentially in the TROPOMI retrieval, whereas all model simulations depict a more gradual, almost linear decay. The authors elaborate that the differences might stem from repetitive injections of SO2 into the atmosphere. However, the different changes in the decline rates suggest a difference in removal processes as well.
Fig 7, 9, 11: The near surface SO2 concentrations in EMAC seem to be more spread out, is there an underestimation of initial removal close to the source or could it be related to how SO2 is emitted in the model or could the distribution be “flatter” due to e.g. numerical diffusion?
Minor comments:
L6: close → closed
L6-8: This sentence is a bit long and confusing
L60: Jöckel et al. in parentheses
L89: isopren → isoprene
Table 4: Biomasse → Biomass
L204: citet → cited
L230: comparatative → comparative
L365: heve → have

Citation: https://doi.org/10.5194/egusphere-2025-3915-RC2
- AC1: 'Reply on RC2', Patrick Jöckel, 26 Nov 2025
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-3915/egusphere-2025-3915-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2025-3915-AC1

Status: closed

RC1:
'Comment on egusphere-2025-3915', Anonymous Referee #1, 06 Oct 2025
General comments
This manuscript is a thorough evaluation of the EMAC v2.55 sulfur simulations and makes a useful contribution by (i) closing a model‑internal sulfur budget (ii) documenting how the model compares with satellite data in 2019 to evaluate how it responds to volcanic emissions and (iii) evaluating against long‑term measurements, 2010–2019
The paper is well organized and generally clear. However, it is unnecessarily long, and the presentation of results is at times too detailed, making it difficult to extract the main messages and scientific significance. The description of the model setup partly repeats work published elsewhere, and it is not entirely clear what is new compared to earlier model versions (e.g., Jöckel et al., 2016). For the interactive gas–particle chemistry, more detail would be beneficial, as the current description is incomplete for interpreting SO₂ lifetime and deposition. I.e. the statement that “the simulation did not involve an interactive aerosol submodel” needs clarification. Does this mean that interactions with ammonia are excluded? If so, this should be explicitly stated, as ammonia strongly influences sulfur oxidation pathways, cloud pH, and the partitioning and deposition of sulfur.
Specific comments
For the emissions it is important to notice that earlier global inventory for China have underestimated the reductions after around 2010. I am not sure if CMIP6 emissions have taken this into account? In Line 35 it is written that the emissions in China remain high, which is true, but the authors should also mention the substantial reductions in recent years.

Why is the comparison made with EDGAR 5 instead of the more recent EDGAR 6.1 (2024)? Using EDGAR 6.1 would enable comparison over the full 2000–2019 period. It would strengthen the analysis to run short sensitivity tests (1–2 simulations) using alternative inventories (e.g., EDGAR 6.1) to quantify whether biases in CMIP6 emissions explain the overestimation. If such tests are beyond the study’s scope, this limitation should at least be discussed.

Why are you not comparing to the more recent EDGAR6.1 (from 2024) instead of EDGAR5? Then you would be able to compare it with the whole 2000-2019 period. It would have been useful to run the model with EDGAR and not only compare the emissions inventory to evaluate more directly if it biases in the CMIP6 emissions that cause the overestimation. Not necessarily a full rerun of the model but quantify the bias by re‑running 1–2 short sensitivity test with different emissions. Though I understand if this is beyond the capacity of this work.

Sulfur data from Africa (INDAAF; https://indaaf.obs-mip.fr) and from Canada (CAPMoN) could have been included to provide a more complete global picture. In the U.S., CASTNET is responsible for air and aerosol data, whereas wet deposition data originate from the National Atmospheric Deposition Program (NADP). It appears NADP data were used, but this is not stated explicitly. NADP also includes many more sites than those used here.

The model response to volcanic emissions is evaluated only against TROPOMI satellite data. You write that “it is difficult to ascertain whether the differences originate near the surface or higher up in the atmosphere.” It would strengthen the analysis to use in-situ data from 2019, which are available for that year and would allow a more direct comparison.

It is not written how trends were calculated, i.e. linear regression, Sen’s Slope . Mann Kendall. This should be added to the methods

For trend comparisons, it would be appropriate to refer to regional studies covering similar periods. Some suggestions:
In Europe recent work done by EMEP: https://doi.org/10.4209/aaqr.230237. Seems like you have somewhat larger trend for SO2 (0.05 ug/m3/y) compared to EMEP (0.034 ug/m3/y). The difference may stem from site selection, trend method, or emissions used.

Several studies in North America. Eg.: NADP data: https://doi.org/10.1016/j.atmosenv.2023.119783, CASTNet: https://doi.org/10.5194/acp-22-12749-2022 and from Canada by CAPMoN: https://doi.org/10.5194/acp-22-14631-2022

Japan: https://doi.org/10.1016/j.envpol.2021.117842

China: https://doi.org/10.3390/su17198815

Technical corrections/spelling errors
Line 70: “caclulated” to calculated

Line 80-81: “sun‑synchrinously” to sun‑

Line 89: “isopren to isoprene.

Line 146: “histrotical” to historical.

Line 149: “soley” to solely.

Line 230: “comparatative“ to comparative.

Line 260: “correpsonding” to corresponding.

Line 294: “drived“ to derived

Line 328: “depostion“ to deposition

Line 364: “heve” to have.

Line 543; “cylce“ to cycle

Line 604: “retrieveal” to retrieval.

Line 658: “resepctively“ to respectively.

Line 723: “anlyses” to analyses

Line 724: “caclualted” to calculated

Line 746 “undr” to under. Multiple instances of “acces” and “reslts” to be replaced with access and results.

Table 2 caption: “aersol” to aerosol.

Figure caption 12. EMEP should be replaced with EANET

Standardize Tg(S)/yr vs Tg(S)/a

There is a mix of units used for air concentration and fluxes (e.g., µg m⁻³ vs mmol m⁻² and kgS/ha). These should be standardized throughout the paper for clarity and comparability.
Citation: https://doi.org/10.5194/egusphere-2025-3915-RC1
- AC2: 'Reply on RC1', Patrick Jöckel, 26 Nov 2025
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-3915/egusphere-2025-3915-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2025-3915-AC2
RC2:
'Comment on egusphere-2025-3915', Anonymous Referee #2, 07 Oct 2025

This manuscript comprehensively evaluates sulfur dioxide in the EMAC chemistry-climate model. This is done by first verifying that the sulfur cycle is closed, meaning mass is neither lost nor gained. The SO2 depositions and vertical column densities are then evaluated with a set of ground-based and satellite observations. The structure of the manuscript is generally very well laid out and the methods are explained thoroughly and well executed. However, there are some areas where more concrete explanations are needed.

P4, L90: The authors mention that the aerosol is not interactive in the simulations used for this study and refer to the aerosol radiative effects and heterogeneous chemistry. Is sulfate aerosol represented in some form in these simulations? If so, please state for clarity. If not, what does this imply for the sulfur cycle?

P16, Figure 5: Looking at this figure, the SO2 burden appears to decline exponentially in the TROPOMI retrieval, whereas all model simulations depict a more gradual, almost linear decay. The authors elaborate that the differences might stem from repetitive injections of SO2 into the atmosphere. However, the different changes in the decline rates suggest a difference in removal processes as well.
Fig 7, 9, 11: The near surface SO2 concentrations in EMAC seem to be more spread out, is there an underestimation of initial removal close to the source or could it be related to how SO2 is emitted in the model or could the distribution be “flatter” due to e.g. numerical diffusion?
Minor comments:
L6: close → closed
L6-8: This sentence is a bit long and confusing
L60: Jöckel et al. in parentheses
L89: isopren → isoprene
Table 4: Biomasse → Biomass
L204: citet → cited
L230: comparatative → comparative
L365: heve → have

Citation: https://doi.org/10.5194/egusphere-2025-3915-RC2
- AC1: 'Reply on RC2', Patrick Jöckel, 26 Nov 2025
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-3915/egusphere-2025-3915-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2025-3915-AC1

Ismail Makroum, Patrick Jöckel, Martin Dameris, Nicolas Theys, and Johannes De Leeuw

Data sets

RD1SD: EMAC CCMI-2022 hindcast simulations with specified dynamics, ERA-5, 1979-2019 P. Jöckel et al. https://doi.org/10.26050/WDCC/ESCiMo2_RD1SD

RD1SD: EMAC CCMI-2022 hindcast simulations with specified dynamics, ERA-5, 1979-2019 (additional data) P. Jöckel et al. https://www.wdc-climate.de/ui/entry?acronym=DKRZ_LTA_853_dsg0002

SO2 data of EMAC sensitivity simulations (eruption of Mt. Raikoke, 2019) I. Makroum et al. https://zenodo.org/records/15655676

Ismail Makroum, Patrick Jöckel, Martin Dameris, Nicolas Theys, and Johannes De Leeuw

Viewed

Total article views: 1,541 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
1,420	99	22	1,541	19	22

HTML: 1,420
PDF: 99
XML: 22
Total: 1,541
BibTeX: 19
EndNote: 22

Views and downloads (calculated since 05 Sep 2025)

Month	HTML	PDF	XML	Total
Sep 2025	1,157	22	3	1,182
Oct 2025	142	29	10	181
Nov 2025	81	30	5	116
Dec 2025	40	18	4	62

Cumulative views and downloads (calculated since 05 Sep 2025)

Month	HTML	PDF	XML	Total
Sep 2025	1,157	22	3	1,182
Oct 2025	142	29	10	181
Nov 2025	81	30	5	116
Dec 2025	40	18	4	62

Viewed (geographical distribution)

Total article views: 1,583 (including HTML, PDF, and XML) Thereof 1,583 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 19 Dec 2025

Short summary

We use a state-of-the-art numerical chemistry-climate model to study the atmospheric sulfur dioxide budget. We simulate the atmospheric concentration of sulfur dioxide (SO₂) and corresponding sulfur deposition fluxes and compare the results with observational data from a satellite instrument and with ground-based in-situ measurements. For the evaluation of the simulated atmospheric lifetime of SO₂, we also simulate the fate of SO₂ emitted by two volcanic eruptions that happened in 2019.


Total:	0
HTML:	0
PDF:	0
XML:	0