the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
CAMx-UNIPAR Simulation of SOA Mass Formed from Multiphase Reactions of Hydrocarbons under the Central Valley Urban Atmospheres of California
Yujin Jo
Sanghee Han
Azad Madhu
Bonyoung Koo
Yiqin Jia
Zechen Yu
Soontae Kim
Jinsoo Park
Abstract. The UNIfied Partitioning-Aerosol phase Reaction (UNIPAR) model was established on the Comprehensive Air quality Model with extensions (CAMx) to process Secondary Organic Aerosol (SOA) formation by capturing multiphase reactions of hydrocarbons (HCs) in regional scales. SOA growth was simulated using a wide range of anthropogenic HCs including ten aromatics and linear alkanes with different carbon-lengths. The atmospheric processes of biogenic HCs (isoprene, terpenes, and sesquiterpene) were simulated for the major oxidation paths (ozone, OH radicals, and nitrate radicals) to predict day and night SOA formation. The UNIPAR model streamlined the multiphase partitioning of the lumping species originating from semi-explicitly predicted gas products and their heterogeneous chemistry to form non-volatile oligomeric species in both organic aerosol and inorganic aqueous phase. The CAMx-UNIPAR model predicted SOA formation at four ground urban sites (San Jose, Sacramento, Fresno, and Bakersfield) in California, United States during wintertime 2018. Overall, the simulated mass concentrations of the total organic matter, consisting of primary OA (POA) and SOA, showed a good agreement with the observations. The simulated SOA mass in the urban areas of California was predominated by alkane and terpene. During the daytime, low-volatile products originating from the autoxidation of long-chain alkanes considerably contributed to the SOA mass. In contrast, a significant amount of nighttime SOA was produced by the reaction of terpene with ozone or nitrate radicals. The spatial distributions of anthropogenic SOA associated with aromatic and alkane HCs were noticeably affected by the southward wind direction owing to the relatively long lifetime of their atmospheric oxidation, whereas those of biogenic SOA were nearly insensitive to wind direction. During wintertime 2018, the impact of inorganic aerosol hygroscopicity on the total SOA budget was not evident because of the small contribution of aromatic and isoprene products that are hydrophilic and reactive in the inorganic aqueous phase. However, an increased isoprene SOA mass was predicted during the wet periods, although its contribution to the total SOA was little.
- Preprint
(2135 KB) - Metadata XML
-
Supplement
(8299 KB) - BibTeX
- EndNote
Yujin Jo et al.
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2023-93', Anonymous Referee #1, 10 May 2023
The paper “ CAMx-UNIPAR Simulation of SOA Mass Formed from Multiphase Reactions of Hydrocarbons under the Central Valley Urban Atmospheres of California” by Yujin Jo et al. presents a one month simulation of organic aerosol build-up with the CAMX/UNIPAR system including not only gas-phase, but also in particle organic and water phase SOA build-up in the Central valley over California.
Although this topic is rather new, in the sense that most CTM models only include the gas phase formation of semi and low-volatile organics, I think that several points should be better addressed, before the paper can be considered for publication in ACP.
- The authors should better state how innovative their work is. Except the Yu et al., 2022 paper for the Korus campaign in South Korea, are their other examples where a detailed multiphase aerosol model like UNIPAR was integrated into a 3D model system? This should be pointed out.
- While the Central valley might be an interesting place with significant organic aerosol pollution, the available data from four sites of the standard air quality network are in my sense rather poor for detailed model evaluation. They only provide daily averages each three days, so we have 12 OC data points for each of the four sites. I would propose two options: either apply the model to an intensive field campaign giving information on precursor gases, oxidants, and organic aerosol composition, ideally including tracers for different formation pathways and precursors. This would be in my mind highly interesting, and I recommend thinking in this way.
As an alternative, authors could conduct a more operational evaluation. To my knowledge, the CSN and IMPOROVE networks contain more the 100 sites with OM data. Performing a simulation over a wider area and period would allow a truly statistical analysis of model ability to simulate SOA. From a computer point of view, is a longer situation possible? - I think that figures 4 and 5 are not coherent with figure 6. From what I see, figure 6 shows a bit less than half of aromatic SOA with respect to alkane SOA averaged over the period (roughly both contributions have similar spatial contributions) while in figures 4 and 5 the aromatic contribution is very low. The same incoherence seems to appear for sesquiterpene SOA, in figure 6 it is with terpene the major SOA contributor in San José, but in figures 4 and 5 it is nearly absent. In section 2.5 it is said, that sesquiterpene emissions are negligible. Please check these figures.
- Beyond this incoherence, it is astonishing that the aromatic contribution is so low, and the alkane one so high. In the conclusion it is said that these are partly long alkane chains, probably in the IVOC range. This should be made clearer already before the conclusion, when talking about emissions in section 2.5
- The detailed model results about species contribution cannot be evaluated from the masurements available from this paper. They should be discussed in depth based on intensive campaigns elsewhere, beyond the discussion with the KORUS campaign. In the same order of ideas, is the 10 -35% simulated contribution of in-particle formation expected from literature, has such a contribution already been simulated, observed in the field or in a chamber?
- The paper’s quality in English wording is not sufficient. This can easily be improved. I put some examples below.
Specific comments:
Line 55: "To compensate the underestimation of SOA, the SOA model employed high partitioning-base model parameters, emerging gas mechanisms to form low-volatile products via autoxidation (Mayorga et al., 2022; Jokinen et al., 2015; Pye et al., 2019), or nighttime oxidations of HC with nitrate radical and ozone (Zaveri et al., 2020; Gao et al., 2019)."
This suggests that the mentioned processes do not occur in the atmosphere, or at least are overestimated. Both statements are not proven.
Line 82: “In this study, the CAMx–UNIPAR model was extended to alkane SOA and nighttime chemistry of biogenic HCs.” and Line 87: “In addition, the UNIPAR model has recently been expanded to simulate biogenic SOA based on three major paths (i.e., OH radicals, ozone and nitrate radicals) being capable of nighttime SOA formation (Han and Jang, 2023) that was dominated by oxidation with ozone and nitrate radicals.”
From this it is not clear where these changes are first presented, in the present work or already in Han and Jang (2023). Or the first comparison to field data is done here? (it becomes clear from table 1 and looking into the papers but should be made clear in the text).
Around line 190 : deriving POA from a fixed POA/EC ratio is uncertain, even more since the used ratio has been derived some 20 years ago. No control is possible using daily measurements. This is in line with my above remark that the used observational data-set is not suitable for in depth model evaluation.
Line 270: it is not specifically figure “Fig. 4(e–h)” here.
Line 274 : « For example, the fraction of alkane SOA was higher than that of terpene SOA in all sites except Sacramento.” Is this SOA from the ALK5 lumped species? Does this include long chain alkanes, of intermediate volatility. Please state this before the conclusion section.
Lines 410 – 427: there is a long discussion on the potential NOx and SO2 reduction impact on BSOA formation. I encourage the authors to add a test simulation where they test such scenarios over their model domain.
In table 1, I guess that SOA data means chamber data for validation of parts of the UNIPAR model. Please correct.
I do not understand why 50 *4 stoechiometric coefficients are needed.
For regional simulations, I think Yu et al. 2022 should appear more often except for alkanes.
Wording and typos :
Line 45: “The SOA model has been simulated in regional and global scales ……”
This is a strange formulation. You could say : “the ….model has been to simulate …. at regional and global scales ….”
Line 69: “established on UNIPAR » in, within?
Line 99 : “as follow » with an « s ».
Line 101 : “For the SOA formation in multiphase …” -> “For the multiphase SOA formation …”
Line 107 : “ volatility-reactivity base » -> « volatility-reactivity based”
Line 115 : “are used to manage their multiphase partitioning” -> may be : “are used to determine their multiphase
partitioning”
Line 422: “where aromatic SOA is MORE dominant than that in California”
Line 436: “can yield lower SOA yields” please reformulate
Citation: https://doi.org/10.5194/egusphere-2023-93-RC1 - AC1: 'Reply on RC1', Myoseon Jang, 15 Sep 2023
-
RC2: 'Comment on egusphere-2023-93', Anonymous Referee #2, 10 Aug 2023
Review of Jo et al (2023)
Jo et al present an evaluation study of the CAMx–UNIPAR against primary and secondary organic aerosol data from 4 sites in and near the Central Valley of California. The CAMx–UNIPAR has been updated to include alkane SOA and nighttime biogenic VOC chemistry. Analysis of the composition of the modelled organic aerosol in terms of its precursors (terpenes, isoprene, sesquiterpene, alkanes and aromatics) is discussed as well as the pathways by which precursor mass can be transferred to the aerosol phase and the role of hygroscopicity.
The CAMx–UNIPAR model is an important development in simulation of SOA and this is useful evaluaton of the model. However, there are several areas which need to be addressed before the manuscript is ready for publication.
Section 1
In the introduction reference is made to “the SOA model” and I’m an unclear if this referring to a particular model or the more general idea of modelling SOA. This might be a case of just refining the language.
“Volatility-reactivity based lumping species originating from explicit gas mechanisms allow to estimate their physicochemical parameters that process multiphase partitioning and in-particle reactions (i.e., oligomerization and acid catalyzed reactions).” I’m not sure I fully understand this sentence. Do you mean that by adopting an approach where species from a gas phase chemical mechanism are lumped based on their volatility and reactivity allows their physicochemical parameters, which are highly influential for the partitioning to multiphase aerosol and the subsequent in-particle reactions to be estimated?
I don’t understand the use of the word “emerging” in line 56 – please could you clarify?
I think it would be worth mentioning that uncertainty in emissions of primary OM and secondary OM precursors may also contribute to model biases.
In the final paragraph you mention the expansion of the CAMx–UNIPAR model to include alkane SOA and nighttime chemistry of biogenic HCs in this work. There are then further descriptions of other changes made to CAMx–UNIPAR in other studies. It is not clear if these changes are also included in the version of CAMx–UNIPAR used in this study or not. This needs to be clarified. It would also be helpful if the additions to specific to this study were described in more detail and listed in the SI.
Section 2
“These physicochemical parameters are universalized for five major precursor groups in UNIPAR (Table 1).”
What does this mean?
Where are SOA precursor emissions coming from? You mention they are “SAPRC07-based” but I don’t know what this means.
Have the emissions been validated? A bias in these emissions could mean the SOA model is getting the right/wrong answers for the wrong reasons in some cases. Fig 3 suggests a low bias in SOA which might come in part from an emissions bias.
“The lumping species in the lowest volatility is treated as non-volatile OM in this study.” I don’t understand this sentence – do you mean that the lumped species with the lowest volatility is automatically treated as non-volatile OM and so irreversibly partitions into the aerosol phase?
In terms of OMP and OMH, I am unclear which relates to volatility-driven partitioning, which relates to reactive uptake and which relates to the dissolution of gases in aqueous phase aerosol. Differentiating between these is a key part of this model and so more detail is needed here.
Section 3
Line 245 – suggest you replace “degradation” with “decrease”
Throughout this paragraph I would use “bias” in place of “deviation”. For example, “The low bias of the predicted SOA is generally greater than the high bias of the POA, which drives the low bias of the total OM from the observations.”
“The underestimation of SOA mass can be attributed to missing precursor HCs and unidentified chemistry in the gas and aerosol phases. For example, the SOA model is currently missing phenols, branched and cyclic alkanes, and polyaromatic hydrocarbons (i.e., naphthalene).” I’m not clear this is true without an evaluation of the emissions of the precursor species.
“The simulated SOA/POA ratios were relatively lower than those in the observed ratios, as discussed for the different deviations of the predicted POA and SOA from the observations.” I’m not sure I understand the second clause here. Do you mean that the lower SOA/POA ratio from the model is in line with the general model high bias of POA and low bias of SOA?
“A strong wind appeared in the northern area, decreasing the residence time of pollutants, which reduced secondary products of pollutants in this region.”
What time period are you discussing here? From Fig S3, I can see wind speeds at San Jose are persistently higher than for the other locations.
I understand that you cannot do anything about the 3 day averaging of the observational data but I do think the resulting lower number of observation data points means extending these simulations for at least another month or two would be warranted.
It would also be helpful to have a timeseries plot of emissions at each site in a similar format to the line plots of Fig 3.
In Figure 2(b-h), you give units of moles or g per second. I think the units of emission should be moles or g per second per unit area (I admit the final column in (b) could stay as moles/s). While I understand that the magnitude of the emissions of the different species vary considerably such that it would not be sensible to have a single common colorbar range for c-h, could cleanly separated ticks (e.g. 0, 2, 4, 6, 8, 10) be used for each to make comparison easier?
Similarly for Fig 6, I would strongly encourage the authors to consider either a log scale for the colorbar or make it much clearer that the concentrations span 2 orders of magnitude.
“Additionally, the model includes the low volatility products originating from autoxidation of α-pinene ozonolysis products (Roldin et al., 2019; Crounse et al., 2013; Bianchi et al., 2019). The importance of autoxidation mechanisms on terpene SOA formation was in a recent study by Yu et al. (2021c) for the daytime chemistry (Yu et al., 2021c).” While it is very good that you are considering highly oxidised species from α-pinene, could you provide any information about the yields you are using or whether you are using the full Roldin scheme which is substantial. Furthermore, the reference to the paper by Yu et al (2021c) is vague – what did this paper show?
Unless I have misunderstood the difference between OMH and OMP, I would have thought that the alkane autoxidation products would be in the OMP category given their highly oxidised structure and low volatility – please could you clarify?
A better color scale is needed for O3 in Fig S10 since most of the region is off the top end of the scale.
I am surprised by the low yield of SOA from aromatics. Can you provide any more detail about why this is quite so low?
Data Accessibility
In the interests of community modelling and FAIR principles, I would like to see the code and model data uploaded to a freely accessible repository such as Github or Zenodo.
Citation: https://doi.org/10.5194/egusphere-2023-93-RC2 - AC2: 'Reply on RC2', Myoseon Jang, 15 Sep 2023
Yujin Jo et al.
Yujin Jo et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 489 | 119 | 22 | 630 | 59 | 8 | 12 |
- HTML: 489
- PDF: 119
- XML: 22
- Total: 630
- Supplement: 59
- BibTeX: 8
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1