the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Mar 2025
Impacts of wildfire smoke aerosols on near-surface ozone photochemistry
Abstract. Wildfires have been an increasing concern for the environment, yet the ozone (O3) production from wildfires remains poorly characterized. Here, we aim to elucidate the role of aerosols from wildfire smoke in near-surface O3 photochemistry by integrating insights from 0-D box model (F0AM) to 3-D chemical transport model (GEOS-Chem). While smoke aerosols typically inhibit O3 production through heterogeneous chemical and radiative pathways, we find that the positive effects of precursor emissions outweigh the negative effects of aerosols for most fires. The relative importance of the two aerosol effects varies, with the heterogeneous chemical effect generally overshadowing the radiative effect in the far field of fires. However, near the sources of extremely large fires, the radiative effect dominates, leading to an overall suppression of O3 production. By assessing the chain termination of hydrogen oxide radicals (HOx) and introducing the “light-limited” regime determination in GEOS-Chem, we find that a significant portion of O3 production occurred within light-limited and heterogeneous chemistry-inhibited regimes during the 2020 wildfire season in California. Building on the discovery that both aerosol and nitrogen oxide (NOx) concentrations modulate aerosol influence, we demonstrate that the surface PM2.5 to tropospheric NO2 column ratio—a metric retrievable from satellite—can serve as an indicator for identifying aerosol-dominated regimes through observations.
- Preprint
(2157 KB) - Metadata XML
-
Supplement
(1716 KB) - BibTeX
- EndNote
Status: open (until 15 Apr 2025)
-
RC1: 'Comment on egusphere-2025-706', Anonymous Referee #1, 26 Mar 2025
reply
General Comments
This study investigates the impact of aerosols from wildfires on near-surface ozone production and suppression. It uses a variety of tools – a global model, box model, surface data, and satellite data – to perform a comprehensive analysis of the aerosol-driven processes that impact ozone levels in fires of various sizes. The authors have considered many scientific, physical, and model-driven processes that could affect their results and have done a good job explaining their reasoning. I think this paper will be acceptable for publication after minor revisions.
Specific Comments
- The paper would benefit by adding a figure to the Introduction that shows the extent of the fires over California in 2020. The authors could consider figures that show ground heat, smoke extent, etc., averaged over Sep, Aug-Oct, periods of high fire activity, etc. This figure would be useful to cite when introducing the 2020 fires on line 75 and again in line 349 after “reflects the central areas of megafires”. In the paragraph beginning on line 75, you should also reference Text S1.
- Table S1 is introduced on line 121 but it feels out-of-place (or unnecessary) at this point in the manuscript. I understood its inclusion once I got to Section 3.4 and you performed a sensitivity test on . I recommend stating earlier in the paper that you will be testing the impact of so that Table S1 makes more sense. Alternatively, you could remove Table S1 since you list a range of values in lines 369-376. If you keep Table S1, you can just reference it in lines 369-376 instead of reiterating the values in the main text.
- Your method begins all HYSPLIT simulations at 18 UTC (line 147) and all figures are shown at 13:30 LT. Can you explain that further?
- Consider commenting on the impact of using a monodisperse size distribution for all aerosol types in F0AM (line 155) compared to 2 or 7 size bins in GEOS-Chem (line 122).
- The cutoff between near-field and far-field fire effects is 3 hours (Figure 1). How did you choose this cutoff value? Similarly, in lines 233-234 you state that the radiative effect diminishes by about half within 5 hours. Can you provide support (e.g., a figure in the SI) to support this?
- Line 231: Avoid using the word “significant” if you aren’t referring to a statistical analysis.
- The authors have shown that the PM5 : NO2 column ratio could be an indicator of aerosol-dominated regimes (Section 3.5). The authors have clearly explained their methods and results regarding this topic, but I suggest reconsidering defining a specific threshold value (20 (ug m-3)/(1015 molecules cm-2)). Previous studies seem to show that the HCHO : NO2 ratio is better used as a qualitative indicator of NOx and VOC influences on a region, rather than a quantitative indicator, and I think that this PM2.5 : NO2 ratio may be similar. You studied this indicator over a specific region, time, and fire(s) which is useful, but may not be applicable in other regions and times. You mention these considerations in the Conclusion, but I recommend changing the discussion at the end of Section 3.5 to focus on why this ratio is generally useful rather than stating why a specific value of the ratio is useful. Additionally, please comment on why a ratio between a surface quantity (PM2.5) and a column quantity (NO2) is or isn’t useful.
Technical Corrections
- Line 16: The terms “positive” and “negative” are somewhat confusing here, since positive could mean positive ozone formation potential, or it could mean positive human impact by reducing ozone. I recommend modifying these terms to “ozone-producing” and “ozone-mitigating”, or something similar.
- Line 61: Remove comma between “…precursor emissions” and “have been used…”
- Line 85: Rephrase “delve into the reasons underlying the processes” to “delve into the underlying processes”.
- Line 101: add “the”: “anthropogenic emissions in the US are represented…”
- Line 102: You need references for the NEI data and when saying that the NEI data was “scaled based on the national interannual variation in emissions”.
- The paragraph beginning on line 124 would be better if the information was put into a table.
- Line 136: Remove “SI” before “Figure S1”. There are some other locations in the text where this should also be done.
- Lines 136-138: This information should be somewhere in the Results section, not the Methods section. Also, the AQS data should be described/cited in Section 2.4.
- Line 152: Spell out the acronym for F0AM the first time you use it.
- Line 153: Add “the” before “Master Chemical Mechanism”.
- Line 161: Cite Table S2 after saying “…and some VOCs”.
- Table S2 is somewhat confusing; I think that what it’s telling the reader is what GEOS-Chem output variables are used as inputs to various F0AM runs. I recommend renaming the column labels: “F0AM inputs” “GEOS-Chem output variables used as inputs to F0AM”; “F0AM-base” “GEOS-Chem simulation corresponding to F0AM base case”; etc.
- Line 163: define TUV.
- Lines 183-185: Consider adding more detail about the PM5 surface data from Wei et al. (2023).
- Line 188: Consider adding “(see Section 2.5)” at the end of this sentence.
- Line 198: Consider changing “” to “” since it represents a difference in reaction rates rather than a difference in concentrations.
- Line 215: Should “NOx titration, sequestration of NOx” be “NOx titration or sequestration of NOx”?
- Lines 216-217: Is this sentence contradictory? If ozone suppression extends to distant areas of extreme fires, why would that lead to an ozone increase?
- Line 260: At the end of the sentence, reference Figure 1: “…consistent with findings from GEOS-Chem (Figure 1)”.
- Lines 264-265: Should “highest” and “lowest” be swapped?
- Line 267: Rephrase: “NOx in particular” “particularly NOx”.
- Line 282: Add “that”: “Figure 1 indicates that for PM”.
- Line 290: Why is a steeper decline expected for surface NOx compared to column NOx? Do you have data or a reference to support this?
- Why are Figure 4 and Figure S6 split into 2 separate figures? If you keep them separate, modify the Figure 4 caption to list the correct months.
- Lines 369-376: Listing these studies and data values is repetitive with Table S1.
- Lines 380-382: Reference Figure S1.
- Lines 387-389: This sentence (“To assess the potential…model-observation comparisons”) requires more description, either in the main text or in the SI.
- Line 399: How is the “probability of each regime at various PM5 levels” calculated? This should be described in the Methods section.
- Line 434: Change “calcification” to “classification”.
- Line 443: Add “and”: “differ substantially in emissions and aerosol composition”.
Citation: https://doi.org/10.5194/egusphere-2025-706-RC1 -
RC2: 'Comment on egusphere-2025-706', Anonymous Referee #2, 27 Mar 2025
reply
Summary:
This study examines how wildfire smoke aerosols influence near-surface ozone production during 2020 California fire season, using a combination of 3D chemical transport model (GEOS-Chem), 0D box model (F0AM), as well as surface and satellite observations. The authors quantify aerosol impacts through two pathways: (a) heterogeneous uptake of HOₓ radicals, which suppresses ozone formation, and (b) radiative effects that reduce photolysis rates. They further assess how the relative importance of these pathways varies by location (near-field vs. far-field), chemical environment (low-NOx vs. high-NOx), and biomass fuel type. Finally, the study introduces the conceptual definitions of “heterogeneous chemistry-inhibited” and “light-limited” regimes and proposes observational thresholds to diagnose these regimes.
Major Comments:
The manuscript addresses a timely and important scientific question, and it is clear that the authors have invested considerable effort and thought into this work. While many aspects are well-executed and meets publication quality, I have substantive concerns for certain parts that need to be addressed before publication, and thus recommend major revisions.
First, the authors employ the box model (F0AM) in two different configurations. In the first setup, concentrations are initialized by GEOS-Chem, and key inputs (meteorology, boundary conditions, photolysis rates, and dilution rate) are all constrained by GEOS-Chem at each time step. As expected, the F0AM results largely mirror GEOS-Chem outputs, and I don’t feel it added any new insight or validation. If the authors have their reasoning to retain this set of results, I’d suggest assessing the uncertainties around dilution rate (which is not explicitly stated now) more thoroughly. If the dilution rate is set large, F0AM might be basically reproducing GEOS-Chem-derived boundary condition values; a sensitivity analysis to dilution rate will be necessary for any F0AM-derived conclusions.
In the F0AM second setup, it is driven by GFED emissions without concentration constraints. This raises greater concern, as box models driven solely by emissions, particularly in the absence of observation constraints, can produce unrealistically high or low concentrations due to large inherent uncertainties. The manuscript does not provide any form of performance evaluation for this setup, yet several important conclusions (e.g., the effects of biomass fuel type or NOₓ levels on ozone chemistry) are derived from it. A performance evaluation against observations is needed to establish the credibility of these results.
Second, the uptake coefficient for HOₓ radicals, which is central to the heterogeneous chemistry pathway, is assumed to be 0.2, near the upper end of reported values (10⁻³ to 10⁻¹ magnitudes). Although a sensitivity test is presented at 0.1, this remains within the same order of magnitude and does not capture the full span of possible values. Since many conclusions rely on this parameter, a broader sensitivity analysis including lower uptake coefficients is recommended, along with clarification on which conclusions remain robust under more conservative assumptions.
Minor Comments
Line 102: NEI is annual - how was the temporal allocation done? What data sources in particular were used for interannual scaling?
Line 201: Please clarify the choice of 1:30 PM local time.
Line 205: Consider adding a brief summary of fire distribution (spatial & temporal) in the main text for context.
Line 208: Suggest changing “positively influence O₃” to “increase O₃” for clarity.
Line 211: How many fires fall into the “extreme” category? Is there a threshold where aerosol effects outweigh emissions effects?
Figure 1: You mentioned GEOS-Chem is run for the whole 2020 - why does figure 1 only include September 2020 results?
Line 220: I found this term “plume age” somehow confusing; based on your lines 149-150 definition, my understanding is that it's grid-specific and reflects the transport time (and thus distance) from fire source, not the post-fire evolution time, right?
Line 222: is PM enhanced defined based on daily averages?
Line 300: I thought it's the emission effects dominate here, right? Maybe you mean “near the source, heterogeneous chemical may or may not outweigh radiative effects depending on NOx levels”?
Line 345: Are values averaged over full month or fire days only?
Line 395: Please confirm whether results are from GEOS-Chem.
Line 401: You note heterogeneous update effects dominate at PM2.5 >30 µg/m³, but Fig 1 shows net O₃ increases up to 200 µg/m³. Please reconcile and add more explanation here.
Line 404: Replace “significant” unless referring to statistical significance.
Line 410: Are results shown for fire days or all of 2020?
Figure 6: Consider repositioning the color legend in panel (a).
Citation: https://doi.org/10.5194/egusphere-2025-706-RC2 -
RC3: 'Comment on egusphere-2025-706', Anonymous Referee #3, 27 Mar 2025
reply
General comments:
Shen et al. simulate O3 production chemistry during the 2020 California wildfire season using a 3-D chemical transport model (GEOS-Chem) with fire emissions from the Global Fire Emissions Database. The authors find that wildfire smoke aerosols reduce modeled near-surface O3 production via both heterogeneous uptake of HO2 and suppression of photolysis rates, and they include these aerosol effects in O3 photochemical regime assignment based on modeled HOx radical termination. Supporting the GEOS-Chem insights with 0-D chemical box modeling of individual plumes, the authors determine that the relative contributions of smoke aerosol effects on O3 production depend on fire size, plume age, and NOx emissions. The authors propose the surface PM2.5 to tropospheric NO2 column ratio as an observation-based metric for determining whether O3 production occurs in an aerosol-dominated regime.
Overall, the manuscript is well written and the topic is within the scope of ACP. The synthesis of four different O3 photochemical regimes (NOx-limited, NOx-saturated, aerosol heterogeneous chemistry-inhibited, and aerosol light-limited) into a single framework is a novel approach for understanding wildfire O3 production chemistry. The analytical methods seem appropriate, although some clarification is required. We recommend the manuscript for publication pending the consideration of the following points as well as those listed in the Specific Comments section:
- It is not clear what measures were taken to isolate the fire plumes analyzed in GEOS-Chem and F0AM from urban influences on O3 and PM. As the authors state, the interaction between wildfire O3 and PM and urban air quality is not the topic of this paper, and therefore the steps taken to isolate wildfire plume chemistry from urban influences on O3 production should be described explicitly in the methods section.
- The manuscript does not address how vertical distributions of smoke and O3 are handled in the models and in the interpretation of the results. Although the analysis focuses only on near-surface O3, the altitude-dependent photochemistry throughout the vertical layers of fire plumes is still relevant. The manuscript would benefit from discussion of this topic; for example, how vertical mixing of lofted O3 contributes to surface enhancements, or whether the fire size categorization considers PM2.5 enhancements above the surface (or above the boundary layer) that contribute to aerosol shading effects.
Specific Comments:
Line 16: Suggest replacing “positive effects” with “O3 enhancement” and “negative effects” with “O3 suppression” here and throughout the text. The positive/negative terminology is not explicitly defined and thus could be misconstrued as good/bad impacts on air quality.
Line 40: Aldehyde photolysis and alkene photolysis are also important pyrogenic HOx sources (Robinson et al. ES&T 2021)
Line 69–70: Reference(s) are needed for the statement that “dense smoke can create a dark environment that makes O3 production limited by light” (possibly from the list of references on Line 51).
Line 73: The statement “introducing two new regimes” should be reworded, as it suggests that both the aerosol chemistry-inhibited and light-limited regimes are original to this work. The aerosol-inhibited regime has previously been described in the literature (e.g., Ivatt et al. 2022, as referenced on Lines 67–69).
Line 99–101. While the authors state specifically that ethene and ethyne chemistry are included in the GEOS-Chem mechanism, they do not mention whether the GEOS-Chem version they used includes furanoid chemistry (Carter et al. 2022). Furanoid compounds are a major product of biomass burning that contribute to O3 and PM formation (Romanias et al. 2024).
Line 105: The allocation of 65% of fire emissions to the boundary layer is not representative of all fires; the authors should clearly state that their model is specific to fires that mainly impact the boundary layer.
Line 121–123: Clarification is need on what “iterated over” means in this context.
Line 135–138: Further details are necessary in the text to clarify the time period (full year, 1 pm local?) and location (only pixels in which there is an EPA monitor?) over which the reported O3 values are averaged. The reported O3 standard deviations should only have one significant figure. It would be useful to include a scatter plot showing the correlation between model and in-situ O3 in Figure S1, especially since the R2 for this correlation is reported in the text.
Line 143–148: The starting altitudes (injection heights) for the HYSPLIT forward trajectories are not stated. The authors should mention if the trajectory heights (i.e., whether the plume remained in the boundary layer) were considered in the F0AM – GEOS-Chem comparison.
Line 152–156: It is unclear whether all 470 fire plumes (described in Section 2.2) were modeled individually in F0AM, or just a subset of these plumes. F0AM version 4.3 includes a mechanism based on MCM v3.3.1 with additional biomass burning chemistry (“MCMv331_AllRxns_NOAABB”). The text should specify whether this additional chemistry has been included; if it has not, the authors should consider the sensitivity of the F0AM model results to the addition of this chemistry. Further details are needed about the aerosol sizes used in F0AM: Does the “monodisperse size distribution” refer to aerosol size within the different aerosol type/size categories described in Section 2.1? How was the aerosol radius determined for each aerosol type, and what are these radii? Is the F0AM model sensitive to the choice of aerosol radii?
Line 163–164: The authors may consider adding justification for scaling J-values in F0AM using the GEOS-Chem J-values for HONO and HCHO but not NO2, since the NO2 photolysis rate is also critical for O3 production.
Line 171–181: In the F0AM fresh plume analysis using GFED emission factor inputs, it is not clear how the aerosol effects were calculated. The authors should also indicate whether the scaling of pollutants/inputs “to achieve aerosol concentrations ranging from 1 to 300 µg m-3” is based aerosol concentrations at the time of emission or at a certain plume age.
Line 207: The PM enhancement size bins have not been defined. Section 2.2 (Fire Plume Evolution Analysis) may be a good location to define the PM enhancement size bins, along with clarification as to whether the fire size assignment is based on only the PM enhancement in the pixel nearest to the source for a given fire, or for all individual downwind pixels.
Line 219–227 (Figure 1): It has not been specified whether GEOS-Chem pixels containing smoke plumes and strong urban influence were identified and excluded before calculating the averages shown in Figure 1. In other words, if the NOx in the model plumes is not exclusively from fire emissions, then the authors should address how NOx from background or urban sources influences the results shown in Figure 1. Additionally, it would be useful to discuss whether there are any trends in O3 observed in plumes of the same fire size category (i.e., plumes that have the same absolute ∆PM2.5) that have different normalized excess mixing ratios of PM2.5 (∆PM2.5/∆CO) as a way to probe the impacts of relative rather than absolute aerosol loading.
Line 240-241: The statement “O3 suppression in surrounding areas is transported into these regions” needs clarification and justification.
Line 244–257: See comments below for Figure S4, which is confusing. The text should clarify whether the comparisons between F0AM and GEOS-Chem are based on averages of modeled plumes or on individual plumes, and if the latter, how plumes were selected for comparison.
Line 258–265: Due to the scaling of precursor emissions over two orders of magnitude, the OH exposure (chemical age) over 1 hour of physical time is likely quite different between the 1 µg m-3 ∆PM plume model and the 300 µg m-3 ∆PM plume model, for example. Running each model to a specific OH exposure time, rather than a specific physical time, would allow a more robust assessment of the dependence of ∆O3 on ∆PM in the plumes.
Line 269–274: The authors attribute the variation in aerosol effects on O3, for a given PM magnitude, to NOx concentrations. It is unclear, however, whether the PM to HOx-precursor (e.g., HCHO) emission ratio the same for all fuel types.
Line 276–281 (Figure 2): The caption should remind the reader that the model results are from F0AM and are near-field/1-hour run time. Including the PM/NOx emission ratio on each panel would facilitate the comparison described in the text (Lines 267–270).
Line 281: The statement that “the dependence of aerosol effects on NOx is also seen in GEOS-Chem” has not been supported by evidence at this point in the text.
Line 287–288: Suggest replacing “has a positive impact on” with “enhances” because the positive/negative terminology is especially vague in reference to Figure 2, where absolute values of O3 enhancement are used.
Line 376–378: The sensitivity test with an HO2 uptake coefficient of 0.1 is a valuable addition to the analysis. However, this uptake coefficient value is still higher than many of the values in Table S1 as well as other studies in the literature (e.g., Tan et al. 2020). It would be useful include a sensitivity test with an uptake coefficient lower than 0.1, as well as the extension of the uptake coefficient sensitivity tests to other components of the analysis (e.g., Section 3.1).
Line 380–391: This section addresses some of the uncertainty in the regime calculations by applying a correction to GEOS-Chem PM so that it matches ground site PM observations. It may be worthwhile to test the sensitivity of the average fire effects on O3 (Section 3.1) to this same GEOS-Chem PM correction.
Line 413–415: A comment on the variation, or lack thereof, in HOx or HOx precursors across the different NOx bins would add useful context to Figure 6a.
Line 431–445: Use of the same map projection and/or grids for Figures 4 and 7 would greatly aid the visual comparison described in this section. The combination of a surface measurement (PM) with a tropospheric column measurement (NO2) in the PM2.5/NO2 ratio used to identify regimes in Figure 7 may contribute to some of the discrepancy in regime assignment between Figures 4 and 7. The authors should provide justification for using this combined surface-PM2.5/column-NO2 ratio rather than a purely surface-based (ground site PM2.5/NOx) or satellite-based (tropospheric column AOD/NO2) metric.
Supplement
Figure S4: For the 3 panels for a given fire size, there is no description of the difference between the panels (e.g., for the small fire panels, what is the difference between a, b, and c?). Calling the no-fire model results “base” is confusing because in Section 2.1 and Table S2 “base” refers to the with-fire model. Arrange the panels in 4 rows (each corresponding to fire size) by 3 columns, and increase the size of each panel.
Technical Corrections:
Line 59: Unnecessary comma after “VOCs”
Line 61: Unnecessary comma after “emissions”
Line 66: Need another dash between “areas” and “often” if using a dash between “loadings” and “typical”
Line 68: Missing article between “to” and “strong”
Line 115–117: Variable A is not defined. Italicize variable a. Avoid starting sentences with variables.
Line 192: Replace “=” with “-->” in the HNO3 reaction.
Line 289: Rephrase “concentrations of NO2 column decay” as “NO2 columns decay” for consistency with the units in Figure 3.
Line 325: Replace dash with comma for consistency with comma on Line 326 (after “light-limited regime”)
Line 434: “Calcification” seems to be a misspelling of “calculation”
References
Carter, T. S., et al. (2022). "An improved representation of fire non-methane organic gases (NMOGs) in models: emissions to reactivity." Atmospheric Chemistry and Physics 22(18): 12093-12111.
Ivatt, P. D., et al. (2022). "Suppression of surface ozone by an aerosol-inhibited photochemical ozone regime." Nature Geoscience 15(7): 536-540.
Robinson, M. A., et al. (2021). "Variability and Time of Day Dependence of Ozone Photochemistry in Western Wildfire Plumes." Environ Sci Technol 55(15): 10280-10290.
Romanias, M. N., et al. (2024). "Emissions and Atmospheric Chemistry of Furanoids from Biomass Burning: Insights from Laboratory to Atmospheric Observations." ACS Earth and Space Chemistry 8(5): 857-899.
Tan, Z., et al. (2020). "No Evidence for a Significant Impact of Heterogeneous Chemistry on Radical Concentrations in the North China Plain in Summer 2014." Environ Sci Technol 54(10): 5973-5979.
Citation: https://doi.org/10.5194/egusphere-2025-706-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 169 | 48 | 9 | 226 | 17 | 3 | 3 |
- HTML: 169
- PDF: 48
- XML: 9
- Total: 226
- Supplement: 17
- BibTeX: 3
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 141 | 62 |
| China | 2 | 18 | 7 |
| undefined | 3 | 10 | 4 |
| South Korea | 4 | 8 | 3 |
| Australia | 5 | 6 | 2 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 141