The manuscript presents a study comparing three plume rise parameterizations on their ability to capture observed injection heights, and how they differ in their impacts on air quality and photochemistry on the short and long range. This study represents good contributions to the field and it’s within the scope of ACP. I think the paper needs more work before it’s ready for publication based on the comments below.

My main comments are the following:

• The section evaluating injection heights is very short and could be greatly improved in many ways, including the addition of more cases that are more representative of the really extreme conditions that happened during this period. Also, the only injection height data used is that from MISR capturing fresh plumes. This does not capture the peak of fire activity that usually happens on the afternoon. For this I recommend including data from CALIPSO. While the chances for CALIPSO to capture fresh smoke are much lower, this event was so massive that most CALIPSO overpasses captured some section of the smoke emitted on a previous day. Thus an analysis could be done for the regional smoke heights rather than for smoke from individual fires, complementing that from MISR. The analysis should move a bit beyond a few study cases and try to capture the whole extend of the fire

• The article could be improved by being more thorough in it’s literature review and using reference to better backup some statements. See some examples in the comments by line below

• Some sections of the results were very qualitative on it’s description, where I think a better job on being quantitative and using statistical metrics could have been done. More details in the comments below.

Comments by line:

Intro. There have been multiple studies evaluating some of these plume injection height schemes beyond the Ye et al. (2021), so these previous findings need to be summarized. Some that come to mind can be found below, also look for work from Joe Wilkins. This literature review also can be used to motivate this study (i.e., what hasn’t been done). Some uniqueness I see from these work include the comparison of these 3 schemes and the type of event studied (record-breaking wildfire season)

-Mallia, D., Kochanski, A., Urbanski, S. & Lin, J. Optimizing Smoke and Plume Rise Modeling Approaches at Local Scales. Atmosphere 9, 166 (2018).

- Wilmot, T. Y., Mallia, D. V., Hallar, A. G., and Lin, J. C.: Wildfire plumes in the Western US are reaching greater heights and injecting more aerosols aloft as wildfire activity intensifies, Scientific Reports, 12, 12400, 10.1038/s41598-022-16607-3, 2022.

-Sessions, W. R., Fuelberg, H. E., Kahn, R. A. & Winker, D. M. An investigation of methods for injecting emissions from boreal wildfires using WRF-Chem during ARCTAS. Atmos. Chem. Phys. 11, 5719–5744 (2011).

-Roy, A. et al. Effects of Biomass Burning Emissions on Air Quality Over the Continental USA: A Three-Year Comprehensive Evaluation Accounting for Sensitivities Due to Boundary Conditions and Plume Rise Height. in Energy, Environment, and Sustainability 245–278 (Springer Singapore, 2017). doi:10.1007/978-981-10-7332-8_12

76. Add a reference to support these statements

95-104. This paragraph is lacking any referencing, please add. Also, the notion that primary organic aerosol is also quite dynamic needs to be included as well.

Section 2.2. A description of how model smoke injection height is derived is missing. Please be specific, e.g., what variable and what threshold is used, how is it mapped to the MISR pixels (location and time). Also include info on how AOD is derived and mapped to VIIRS.

Section 2.3. There are some details missing from the explanations of the injection schemes that could be useful to understand results. For instance, how is the plume distributed once the injection is computed? Is there a height of the bottom of the plume computed as well or how is it assumed? Is there a fraction of emission placed at the surface (so called “smoldering” emissions as stated in Freitas 2007)? If so, what % for each scheme? Due any of the parameterization consider different parameters for different fuels? Is the FRP used from GBBEPx a daily value? If so, is it applied as constant throughout the day or a given diurnal cycle is specified?

161-162. Please expand a bit more on why the factor of 10 is applied.

167-168. Please add references for this sentence

Section 2.4. VIIRS AOD data was not described. Please include any quality flags applied

180. MISR injection heights have the limitation that MISR overpass is in the morning while peak fire behavior (and thus deeper injections) tends to be in the afternoon. This has been highlighted in previous work (paper below for instance). Since this is the only dataset used for evaluation in this work, this needs to be mentioned and taken into consideration when discussing results and deriving conclusions.

Kahn, R. A. et al. Wildfire smoke injection heights: Two perspectives from space. Geophys. Res. Lett. 35, (2008).

Figure 3. Having a visible image (MODIS Terra) including hotspots would help to visualize each scene.

Figure 3. It would also help to have the model boundary layer height as reference to assess boundary layer injections versus into the free-troposphere. Given the range shown by MISR, I would assume it’s estimates contains a mixture of boundary layer smoke and injections. But the model doesn’t show this variability, so it would be nice to understand why

Figure 3. Are these heights capturing mostly freshly emitted smoke or is here any recirculated smoke from the same or other fires?

Figure 3. I find that evaluating only for 2 snapshots capturing 2 fires during August is a bit insufficient. There are likely many more opportunities during this 2 month period, especially during September where fires in the whole western US were exploding.

Figure 3. Also think in better ways of presenting the data, maybe aggregate MISR to the model resolution to avoid having so many repeated values for the model?

Figure 3. The model resolves the increase of height with distance, so you can do analysis to assess why is this happening. This is a bit counterintuitive as one would expect to earlier plume (i.e., further away) be emitted a lower altitudes. Or are the conditions such that the plume is just rising with time?

191-202. Please be more quantitative. Higher by how much? Show some statistics

206-207 Maybe do statistical testing on the mean o backup this statement?

208-213. Need to better backup these statements. For instance, what were the stability conditions the model used for these fires. Also note Briggs not always shows higher injections.

Figure 4. It would be nice to explore if these trends are also found on observations such at those from IMPROVE sites

Figure 5. It would be good to have a panel showing the average profile for one of the schemes to use it as a reference.

Section 3.3. While VIIRS AOD is included, it doesn’t seem to be used in the analysis. Satellite AOD tends to saturate around 5 so retrievals for the very fresh plumes are likely missing. If the model was not screened by these missing values then this likely explain why the model is overpredicting AOD on the locations of the fires. Ones you move away a bit from the fires the bias flip, with models tending to underpredict AOD. This discussion needs to happen before analysis is done comparing model runs.

295. Reference Figure 6 to make it clear you are comparing to that figure

310. Changes of 70% are described, but the color-scale of Fig 8 saturates at 30%. Exploring a scale that’s not linear (like Fig 5) might work better.

Figure 8. I t would be nice to have a surface PM2.5 map for one of the simulations as a reference to better interpret the differences.

Figure 8. How much of these differences are due to differences in injection height versus assumptions of fraction of emissions placed at surface levels vs injected?

319-331. While 35 um/m3 is the standard, this divides the "Moderate" from "Unhealthy for sensitive groups" categories. However, it would be nice to see how the models in predicting the more extreme categories ("Unhealthy", "Very unhealthy", "Hazardous".) where more authorities might take more stringent measures

319-331. This analysis is based on one day. A way to generalize this analysis could be to show a map or difference maps of the number of days the models predict exceedances.

Minor Edits

191: ACP won’t allow links, convert it to a reference

321-325. A lot of this text is already in Fig 9 caption so no need to repeat.

Summary:

This paper used an offline CMAQ model with three different plume rise schemes to discuss the impact of injection plume height on local and downwind aerosols’ chemical composition as well as the photochemical processes. The paper clearly explains the basic settings of the CMAQ model and the different input parameters in three plume models.

This work has compared the predicted AOD among three plume models to illustrate their different prediction performance in the source and downwind area, while a detailed comparison between VIIRS observed AOD and modeled AOD is expected to further validate the model prediction accuracy. This work has made the comparison of PM exceedance regions between the model prediction and the AirNow observation on Aug 20, 2020. The result shows good consistency.

The analysis of the plume models’ impacts on photochemistry mainly focuses on the photolysis of NO₂. We expect more observation evidence on NO₂ concentration to validate the model prediction. The relationship between NO₂ concentration or photolysis rate and other species concentration which have adverse effects on human health (e.g., ozone) needs to be further established.

General comments:

Introduction:

Authors explained the different parameters used in three schemes for plume height estimates. We expect the authors to explain how the later-published schemes of plume height simulation improve the modeling accuracy, in general. Also, what is the limitation of each model.

Method:

The CMAQ model domain has a spatial resolution of 12 km × 12 km. The scale of an wildfires in the western US is normally smaller than the spatial resolution of the defined CMAQ domain (Biomass burning emission is a 0.1 degree product). Please provide the dimension information (or related inforation) of the studied fire to explain the choice of domain resolution.

In section 2.1 Experiment Design, authors have mentioned the reaction pathway from VOCs to SOA. However, in the result analysis part, section 3.2, the contribution of SOA in the total OM has not been mentioned. The potion of OM in the total PM2.5 seems to be completely regarded as the primary emission. Combining primary OA and SOA together may introduce errors in the further discussion on particle/gas transport issue.

Results:

Figure 5 plots the specific chemical component of PM2.5 against the distance (km) from the source point. Is this distance along certain smoke transport pathway? If so, which specific pathway you chose to sample the modeled concentration of different species.

Specific comments:

Line 48: PM2.5 definition: Particle’s aerodynamic diameter is less than 2.5 μm

Line 48: “47%” in mass or other types of measure?

Line 62: “Irregular large point sources”. What does this terminology mean? The boundary of the source is irregular? Then why is a point source?

Line 78: the unit of “3720”. Daily, hourly cases?

Line 156: What’s the result of this reason?

Line 209: Define “ABL” before using it

Line 225: Is “OM” here the same as organic carbon you defined in section 2.1, which only refers to primary organic carbon?

Line 226: Clarify that the composition of PM2.5 in this section is surface PM2.5, or PM2.5 under PBL, or column PM2.5. Line 232 mentioned “surface PM2.5”, and the conclusion of this section is “integrated over all vertical layer”.

Figure 4: The negative sign before the longitude is unnecessary

Line 249: Unify the representation of longitude: either 115° W or -115° throughout the paper

Figure 6: The comparisons between VIIRS AOD and modeled results may be needed to demonstrate the prediction accuracy of different plume models.

Line 286: Thicker smoke in this study doesn’t necessarily mean higher AOD. Thicker smoke somehow may be attributed to a diluted plume because of the different plume height modeled by different schemes. A basic assumption in this study is the primary biomass burning emissions among three models are similar (or identical).

Typos or other improvement suggestions:

Line 25: lower case “western”

Line 46: two “annual”

Line 63: Start a new sentence to declare the second limitation of Briggs scheme.

Line 69: Please provide the reference of Siberia study

Line 69: “height”

Line 72: Suggest start a new sentence here.

Line 162: “from… from…” redundancy

Line 185: lower case

Line 263: lower case

Figure 6: Increase the tick font size of the colorbar

Line 296: The difference ratios in NO2 is higher than the ones of AOD can only prove the concentration of NO2 is not linearly proportional to AOD.

Line 297: The reaction rate of NO2 (for this NO2 -> NO + O reaction) is the product of <NO2_IUPAC10> and the concentration of NO2. The comparison of reaction rates between different plume schemes is needed to support your conclusion.

Figure 7: Increase the tick font size of the colorbar

Other comments

Line 67: This sentence means Sofiev scheme used the MISR observed plume height to determine the modeled plume height. I am a bit confused about it (Line 67 - 69).

Line 76: Please provide the reference of the burned area. (I remembered the burned area in the western US in 2020 is high but below 10 million acres. The entire US is larger than 10 million acres)

Review comment on “Impacts of estimated plume rise on PM2.5 exceedance prediction during extreme wildfire events: A comparison of three schemes (Briggs, Freitas, and Sofiev)

The authors compared three popular plume rise schemes, namely Briggs 1969, Freitas 2007 and Sofiev 2012, and their impacts on the simulated plume heights, AOD, PM2.5 and NO2 photochemistry using the CMAQ model driven by WRF meteorology data for the 2020 western U.S. wildfire season. With global warming, the increasing trend in western U.S. fire activities, and the need to predict hazardous air quality associated with wildfires, the study would make a timely and significant contribution to wildfire and air quality modeling science. So publication is recommended. However, I believe, the presentation can be significantly improved to increase the scientific impact of this study.

Major comments:  

The descriptions of each of the plume rise schemes are short. More details of the schemes could be provided to help readers know better of the differences of the schemes (length of description can be doubled or tripled). Also, the authors focus mainly on plume top height, however plume extension (top and bottom of a plume) in the vertical at emission is as important as plume top. Information about plume vertical extension at emission and how emission mass is distributed in the vertical (e.g. evenly or weighted) from the schemes should be provided.

It’s not clear how model and MISR plume heights were compared. The model and MISR observations don’t have the same spatial and temporal resolutions, and MISR observations are not continuous in time and space. So some spatiotemporal interpolation is expected. The treatment of the model and MISR data for comparison should be clearly stated.

Section 3.3 and conclusion: Why do F07 and S12, which tend to have lower plume height than B69 near source region, have higher AOD than B69 near source region? What is the column total PM2.5 differences in the source region for the three schemes? Is the difference small? (You could consider providing total/average PM2.5 or even better dry mass in Figure 4 for the different regions and schemes) If so, what causes the 20-30% AOD difference in source region?  Is it purely because of different vertical distributions of same mass? For example, there could be more aerosols in the lower altitude in F07 and S12 (and RH tends to be higher than higher altitude), so that hygroscopic growth of smoke in the lower layer leads to the higher AOD? Or is it due to different SOA production rate? The authors should be able to provide some discussions through analysis.

Figure 9: This is a case study of PM2.5 exceedance. Using a color wheel with overlapping colors to represent simulated PM2.5 exceedance regions from the three schemes is brilliant. I do have a few questions though: Why August 20^th is chosen as the case? The authors should provide a reason. Since this is a case study, some background of the wild fires and PM transport should be provided. Did you compare the plume heights with MISR (This case was not included in the earlier section or Figure 3)? Why F07-B69 difference in daily surface PM2.5 is provided, but not S12-B69? The authors should provide the reasoning of leaving this comparison out or making this comparison. 

Minor comments:

Line 69:  “heigh” should be “height”.

Line 77-78: Please define “PM2.5 exceedance”. Is it based on daily-mean or hourly PM data? Also for the 3720 observations, how many sites are the observations based on?

Line 79-80: There is no direct visual link between “hazy” and AOD shown in Figure 1. I would suggest adding a matching VIIRS true color image and/or define “hazy” in terms of AOD and PM values.

Figure 2: What is the data source of this figure and pie chart in the figure. The authors should cite some papers on fire emission chemistry (currently there are none) in the introductory paragraph (line 95-105) of experiment design. Below are a few examples:

Koppmann, R., von Czapiewski, K., and Reid, J. S.: A review of biomass burning emissions, part I: gaseous emissions of carbon monoxide, methane, volatile organic compounds, and nitrogen containing compounds, Atmos. Chem. Phys. Discuss., 5, 10455–10516, https://doi.org/10.5194/acpd-5-10455-2005, 2005.

Schlosser, J. S., Braun, R. A., Bradley, T., Dadashazar, H., MacDonald, A. B., Aldhaif, A. A., Aghdam, M. A., Mardi, A. H., Xian, P., and Sorooshian, A.: Analysis of aerosol composition data for western United States wildfires between 2005 and 2015: Dust emissions, chloride depletion, and most enhanced aerosol constituents, J. Geophys. Res.-Atmos, 122, 8951–8966, https://doi.org/10.1002/2017JD026547, 2017. 

Line 145-146: To be more accurate, the vertical profiles of PM2.5 would match the “vertical profiles of backscatter” from CALIPSO.

Line 156-157: Not a complete sentence.

Line 183: “70°” instead of “70”.

Line 232: “….which include nitrate formation from both wildfires and anthropogenic emission”. This is confusing. I would expect no anthropogenic influence , as “the impact of other PM2.5 sources was removed by subtracting the results of NoFire run” from line 224.

It may be worth labeling the longitudes on the upper x axis on the geographical plots (e.g. at least Figure 1), so that readers would know the projection of the maps and where the division lines lie between the regions. This information is currently not straight forward. An alternative is to plot the division lines on the maps.

Line 298-299: “The consumption of NO2 is slowed down so that the NO2 concentration is higher in the high AOD area.” I think you meant “so” instead of “so that”.

Line 271: You could consider updating the subsection title to include the impact of plume rise on photochemistry besides AOD.

Line 321-325: The description of color scales is already included in the figure caption, which is the right place. It is redundant here in the text. 

Figure 8: It would be helpful if a plot of the average surface PM2.5 from B69 overlaid with AirNow measurement is provided, as the difference plots here are based on B69 surface PM2.5. Also the addition of AirNow would provide some kind of evaluation for the model.