the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Feb 2024
Intercomparison of GEOS-Chem and CAM-chem tropospheric oxidant chemistry within the Community Earth System Model version 2 (CESM2)
Abstract. Tropospheric ozone is a major air pollutant and greenhouse gas. It is also the primary precursor of OH, the main tropospheric oxidant. Global atmospheric chemistry models show large differences in their simulations of tropospheric ozone budgets. Here we implement the widely used GEOS-Chem atmospheric chemistry module as an alternative to CAM-chem within the Community Earth System Model version 2 (CESM2). We compare the resulting simulations of tropospheric ozone and related species to observations from ozonesondes, the ATom-1 aircraft campaign over the Pacific and Atlantic, and the KORUS-AQ aircraft campaign over the Seoul Metropolitan Area. We find that GEOS-Chem and CAM-chem within CESM2 have similar tropospheric ozone budgets and concentrations usually within 5 ppb but important differences in the underlying processes including (1) photolysis scheme (no aerosol effects in CAM-chem), (2) aerosol nitrate photolysis, (3) N2O5 cloud uptake, (4) tropospheric halogen chemistry, and (5) ozone deposition to the oceans. Global tropospheric OH concentrations are the same in both models but there are large regional differences reflecting the above processes. Carbon monoxide is lower in CAM-chem (and lower than observations) because of higher OH concentrations in the northern hemisphere and insufficient production from isoprene oxidation in the southern hemisphere. CESM2 does not scavenge water-soluble gases in convective updrafts leading to some upper tropospheric biases. Comparison to KORUS-AQ observations shows successful simulation of oxidants under polluted conditions in both models but suggests insufficient boundary layer mixing in CESM2. The implementation and evaluation of GEOS-Chem in CESM2 contributes to the MUSICA vision of modularizing tropospheric chemistry in Earth system models.
-
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
(2363 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2363 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-470', Anonymous Referee #1, 08 Apr 2024
This study compares tropospheric ozone simulations using GEOS-Chem and CAM-chem in CESM2. While both models show similar ozone budgets and concentrations within 5 ppb, they differ in key processes like photolysis schemes, aerosol effects, and halogen chemistry. Evaluation against observations suggests successful simulation of oxidants under polluted conditions but highlights potential biases in boundary layer mixing in CESM2. This integration supports the MUSICA vision of modularizing tropospheric chemistry in Earth system models.
Main comments
- There are numerous global-scale atmospheric chemistry transport models available today. Could you provide more context on why selected these two particular models for comparison? Giving a more in-depth discussion on the scientific significance behind this choice would be helpful.
- This study has extensively compared vertical profiles, but what about comparisons of surface observational elements? My suggestion would be to include data from ground monitoring stations to assess ozone and nitrogen oxides.
- In figure 4, I am wondering why surface ozone were rather high in western China (especially in regions like Tibet, almost the highest around the world) where anthropogenic emissions, i.e, NOx, were relatively low. Did you compare the surface simulations with surface observations?
- Have these two models taken into account the impact of halogen chemistry mechanisms on the formation of photochemical ozone? If they have, please provide some discussion.
Minor suggestions
- Line 155-160 “Fast-JX includes aerosol extinction but TUV does not, which explains the larger differences over polluted and open fire regions” Please give some examples to indicate these regions
- Line 200-205, Please extend more about the recent update in isoprene oxidation chemistry why isoprene does not titrate OH in GEOS-Chem
- Why GEOS-Chem simulated NOx in oceans were notably higher than CAM-chem?
- Figure 5 uses pressure while figure 6 and 7 use altitude (km) to show height, it is suggested to use the same unit of height, for instance, kilometers.
Citation: https://doi.org/10.5194/egusphere-2024-470-RC1 -
RC2: 'Comment on egusphere-2024-470', Anonymous Referee #2, 23 Apr 2024
Li et al. implemented the GEOS-Chem atmospheric chemistry module into the CAM-Chem and evaluated the difference in O3 and OH chemistry between model simulations. They also conducted the model sensitivity analysis and discussed the difference in O3, NOx, CO, and OH due to underlying processes including the photolysis scheme, aerosol nitrate photolysis, N2O5 cloud uptake, halogen chemistry, and ozone deposition to the oceans. I appreciate the tremendous technical work involved in implementing the CESM-GC capacity. I suggest minor revisions before this paper is accepted in ACP.
- The authors provided some high-level explanations of the underlying processes that lead to the difference across the model simulations. However, more detail is recommended. For instance, what is the process of scavenging water-soluble gases in convective updrafts? Are these gases NOx and less reactive VOCs? Do they mainly affect O3 formation in the upper troposphere?
- Could you provide some discussion on the halogen chemistry mechanism implemented in the model? How large is the uncertainty in this mechanism? Is the uncertainty introduced through the chemical mechanism smaller than the difference observed here with and without halogen chemistry?
- For the calculation of OH in Table 2, is it air mass-weighted column OH?
- In Table 2, both models generate the same OH. However, from Figure 3, the difference in OH is considerably large, with -3.2% at the surface and -10.1% at 500 hPa. Could you explain this discrepancy?
- The spatial resolutions of both models are coarse, how does the model’s spatial resolution affect the model comparison against observations, especially those from ATOM1 and KORUS-AQ observations?
- Both models consistently show a high bias in O3 compared to observations, except for GEOS-Chem without the PNO3 photolysis. Could you discuss more on the possible causes of the high O3 bias? For instance, how do stratospheric O3 and lightning NOx emissions contribute to O3 in the upper troposphere?
- The figures are vague, please update them to a higher resolution.
Citation: https://doi.org/10.5194/egusphere-2024-470-RC2 -
RC3: 'Comment on egusphere-2024-470', Anonymous Referee #3, 06 May 2024
My apologies for this late review. I congratulate the Editor for pushing this paper along.
The paper 2024-470 by Haipeng Lin et al is a very well written study that examines the tropospheric ozone budget and sensitivities for two very different chemistry modules (GEOS-Chem & CAM-chem) operating within a common framework (CESM2). The writing is incredible; I did not find any typos or very awkward sections! (Well except for BCC?) There are a few minor fixes and one more substantive issue that might be fixed before publication.
L27ff: this list of 5 specific items are not treated equally or evaluated equally well in the paper and hence do not belong so strongly as in the abstract. e.g., N2O5 cloud uptake and oceanic ozone deposition.
L30ff. Yes, CAM-chem is notorious for being very low in CO compared with many models and it seems to be something deeper than attributed here simply to isoprene or OH.
L39: “nonlinear” is used too easily in our community and often incorrectly. It is also meaningless since most of these processes are quasi-linear. how about: “by second-order processes that couple across hydrogen….”
L41: It is “chemistry-transport models (CTMs) NOT chemical-transport since these models do more than just transport the chemicals, they are chemistry models. You got it right with chemistry-climate models so fix this please.
L53 “chemistry is coupled to transport” What do you mean here? it is always, unless you are talking about models that are NOT operator split? i.e. simultaneous chemical rates and transport? If you mean that the resulting chemicals can change the dynamics through radiation then this is a CCM, OK. BUT you model here is forced with MERRA-2 fields and hence is NOT a CCM. Overall the distinction between offline and online is becoming blurred and your use of it here is not helping. If you run within CESM with forced met fields then it is not “online”, it is the same as a CTM that you call offline. I really recommend you drop off/online and come up with a simpler expression. Anyway all the runs here are the same so why bother. Are you running as a full CCM forced by GHGases and SSTs in all these cases?
L65ff: This large number -- 10-30% lower trop O3 from Br reactions -- seems like it is over exaggerated and should not be repeated since later you say that this was due to a major mistake with iodine, not bromine?
L115ff – again the use of offline v. online is only confusing here.
L227: how can a modern CCM like CESM2 have SSTs that are so far off from observed???
L230ff – Table 2 & Table 3: This gets very interesting. Looking hard at these tables, I am beginning to believe that CESM2 simply does not conserve O3. Is this so? Surely you have some diagnostics for this? I have heard talks saying that CESM loses 200 Tg/y of trop O3 but not seen it documented.
The idea of using a residual to get STE fluxes is very old and should not be propagated here. Since there is not evidence that you can calculate the STE flux, questionable evidence that you conserve O3, and obvious evidence that POx-LOx is only good to 5-10% (see Prather Elementa paper on the POX & LOX not being the same as dO3/dt, doi: 10.1525/elementa.2023.00112). You should not presume the accuracy of all you terms by declaring the residual to be STE. It does not matter that this is traditional practice, at least call it unknown residual, but give us an uncertainty in the terms here.
The Table 3 sensitivity results raise some interesting questions – the P-L does not change with 'J-nitrate', but it increases by 100 Tg/y for 'no-N2O5-uptake'. Why? Likewise, the Fast-JX in CAM-chem increases P-L from 587 to 764 Tg/y – where does that excess O3 go? to the surface? STE should not change, or does the stratosphere also change with Fast-JX? You should document that here.
L258 & Fig 6: for comparing with profiles of OH and CO made by aircraft, you must simulate the correct time of day. How did you take these values form the model? Did you actually pull off the nearest hour to when the obs were made? Is the mean profile correctly weighted? Very tough to do this with a CCM, I am impressed. Just note that your flight-track sampling included the correct local solar time ! (L290).
By the way, producing a mean profile when all the points that went into it were sampled most likely in a systematic, weirdly biased time-of-day makes this a pretty much useless diagnostic for anyone else.
L312: typo? what is a BCC ESM, spell out.
Overall, a very nice piece of hard work. It could be cleaned up.
Citation: https://doi.org/10.5194/egusphere-2024-470-RC3 - AC1: 'Comment on egusphere-2024-470', Haipeng Lin, 21 Jun 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-470', Anonymous Referee #1, 08 Apr 2024
This study compares tropospheric ozone simulations using GEOS-Chem and CAM-chem in CESM2. While both models show similar ozone budgets and concentrations within 5 ppb, they differ in key processes like photolysis schemes, aerosol effects, and halogen chemistry. Evaluation against observations suggests successful simulation of oxidants under polluted conditions but highlights potential biases in boundary layer mixing in CESM2. This integration supports the MUSICA vision of modularizing tropospheric chemistry in Earth system models.
Main comments
- There are numerous global-scale atmospheric chemistry transport models available today. Could you provide more context on why selected these two particular models for comparison? Giving a more in-depth discussion on the scientific significance behind this choice would be helpful.
- This study has extensively compared vertical profiles, but what about comparisons of surface observational elements? My suggestion would be to include data from ground monitoring stations to assess ozone and nitrogen oxides.
- In figure 4, I am wondering why surface ozone were rather high in western China (especially in regions like Tibet, almost the highest around the world) where anthropogenic emissions, i.e, NOx, were relatively low. Did you compare the surface simulations with surface observations?
- Have these two models taken into account the impact of halogen chemistry mechanisms on the formation of photochemical ozone? If they have, please provide some discussion.
Minor suggestions
- Line 155-160 “Fast-JX includes aerosol extinction but TUV does not, which explains the larger differences over polluted and open fire regions” Please give some examples to indicate these regions
- Line 200-205, Please extend more about the recent update in isoprene oxidation chemistry why isoprene does not titrate OH in GEOS-Chem
- Why GEOS-Chem simulated NOx in oceans were notably higher than CAM-chem?
- Figure 5 uses pressure while figure 6 and 7 use altitude (km) to show height, it is suggested to use the same unit of height, for instance, kilometers.
Citation: https://doi.org/10.5194/egusphere-2024-470-RC1 -
RC2: 'Comment on egusphere-2024-470', Anonymous Referee #2, 23 Apr 2024
Li et al. implemented the GEOS-Chem atmospheric chemistry module into the CAM-Chem and evaluated the difference in O3 and OH chemistry between model simulations. They also conducted the model sensitivity analysis and discussed the difference in O3, NOx, CO, and OH due to underlying processes including the photolysis scheme, aerosol nitrate photolysis, N2O5 cloud uptake, halogen chemistry, and ozone deposition to the oceans. I appreciate the tremendous technical work involved in implementing the CESM-GC capacity. I suggest minor revisions before this paper is accepted in ACP.
- The authors provided some high-level explanations of the underlying processes that lead to the difference across the model simulations. However, more detail is recommended. For instance, what is the process of scavenging water-soluble gases in convective updrafts? Are these gases NOx and less reactive VOCs? Do they mainly affect O3 formation in the upper troposphere?
- Could you provide some discussion on the halogen chemistry mechanism implemented in the model? How large is the uncertainty in this mechanism? Is the uncertainty introduced through the chemical mechanism smaller than the difference observed here with and without halogen chemistry?
- For the calculation of OH in Table 2, is it air mass-weighted column OH?
- In Table 2, both models generate the same OH. However, from Figure 3, the difference in OH is considerably large, with -3.2% at the surface and -10.1% at 500 hPa. Could you explain this discrepancy?
- The spatial resolutions of both models are coarse, how does the model’s spatial resolution affect the model comparison against observations, especially those from ATOM1 and KORUS-AQ observations?
- Both models consistently show a high bias in O3 compared to observations, except for GEOS-Chem without the PNO3 photolysis. Could you discuss more on the possible causes of the high O3 bias? For instance, how do stratospheric O3 and lightning NOx emissions contribute to O3 in the upper troposphere?
- The figures are vague, please update them to a higher resolution.
Citation: https://doi.org/10.5194/egusphere-2024-470-RC2 -
RC3: 'Comment on egusphere-2024-470', Anonymous Referee #3, 06 May 2024
My apologies for this late review. I congratulate the Editor for pushing this paper along.
The paper 2024-470 by Haipeng Lin et al is a very well written study that examines the tropospheric ozone budget and sensitivities for two very different chemistry modules (GEOS-Chem & CAM-chem) operating within a common framework (CESM2). The writing is incredible; I did not find any typos or very awkward sections! (Well except for BCC?) There are a few minor fixes and one more substantive issue that might be fixed before publication.
L27ff: this list of 5 specific items are not treated equally or evaluated equally well in the paper and hence do not belong so strongly as in the abstract. e.g., N2O5 cloud uptake and oceanic ozone deposition.
L30ff. Yes, CAM-chem is notorious for being very low in CO compared with many models and it seems to be something deeper than attributed here simply to isoprene or OH.
L39: “nonlinear” is used too easily in our community and often incorrectly. It is also meaningless since most of these processes are quasi-linear. how about: “by second-order processes that couple across hydrogen….”
L41: It is “chemistry-transport models (CTMs) NOT chemical-transport since these models do more than just transport the chemicals, they are chemistry models. You got it right with chemistry-climate models so fix this please.
L53 “chemistry is coupled to transport” What do you mean here? it is always, unless you are talking about models that are NOT operator split? i.e. simultaneous chemical rates and transport? If you mean that the resulting chemicals can change the dynamics through radiation then this is a CCM, OK. BUT you model here is forced with MERRA-2 fields and hence is NOT a CCM. Overall the distinction between offline and online is becoming blurred and your use of it here is not helping. If you run within CESM with forced met fields then it is not “online”, it is the same as a CTM that you call offline. I really recommend you drop off/online and come up with a simpler expression. Anyway all the runs here are the same so why bother. Are you running as a full CCM forced by GHGases and SSTs in all these cases?
L65ff: This large number -- 10-30% lower trop O3 from Br reactions -- seems like it is over exaggerated and should not be repeated since later you say that this was due to a major mistake with iodine, not bromine?
L115ff – again the use of offline v. online is only confusing here.
L227: how can a modern CCM like CESM2 have SSTs that are so far off from observed???
L230ff – Table 2 & Table 3: This gets very interesting. Looking hard at these tables, I am beginning to believe that CESM2 simply does not conserve O3. Is this so? Surely you have some diagnostics for this? I have heard talks saying that CESM loses 200 Tg/y of trop O3 but not seen it documented.
The idea of using a residual to get STE fluxes is very old and should not be propagated here. Since there is not evidence that you can calculate the STE flux, questionable evidence that you conserve O3, and obvious evidence that POx-LOx is only good to 5-10% (see Prather Elementa paper on the POX & LOX not being the same as dO3/dt, doi: 10.1525/elementa.2023.00112). You should not presume the accuracy of all you terms by declaring the residual to be STE. It does not matter that this is traditional practice, at least call it unknown residual, but give us an uncertainty in the terms here.
The Table 3 sensitivity results raise some interesting questions – the P-L does not change with 'J-nitrate', but it increases by 100 Tg/y for 'no-N2O5-uptake'. Why? Likewise, the Fast-JX in CAM-chem increases P-L from 587 to 764 Tg/y – where does that excess O3 go? to the surface? STE should not change, or does the stratosphere also change with Fast-JX? You should document that here.
L258 & Fig 6: for comparing with profiles of OH and CO made by aircraft, you must simulate the correct time of day. How did you take these values form the model? Did you actually pull off the nearest hour to when the obs were made? Is the mean profile correctly weighted? Very tough to do this with a CCM, I am impressed. Just note that your flight-track sampling included the correct local solar time ! (L290).
By the way, producing a mean profile when all the points that went into it were sampled most likely in a systematic, weirdly biased time-of-day makes this a pretty much useless diagnostic for anyone else.
L312: typo? what is a BCC ESM, spell out.
Overall, a very nice piece of hard work. It could be cleaned up.
Citation: https://doi.org/10.5194/egusphere-2024-470-RC3 - AC1: 'Comment on egusphere-2024-470', Haipeng Lin, 21 Jun 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 566 | 382 | 44 | 992 | 13 | 16 |
- HTML: 566
- PDF: 382
- XML: 44
- Total: 992
- BibTeX: 13
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Louisa K. Emmons
Elizabeth W. Lundgren
Laura Hyesung Yang
Ruijun Dang
Shixian Zhai
Yunxiao Tang
Makoto M. Kelp
Nadia K. Colombi
Sebastian D. Eastham
Thibaud M. Fritz
Daniel J. Jacob
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2363 KB) - Metadata XML