Review for Pan et al.: The Sensitivity of Smoke Aerosol Dispersion to Smoke Injection Height and Source-Strength in Multiple AeroCom Models

General Remarks: This study presents a case study of a Siberian wildfire event in April 2008 using four models and three different experiments using those models (with an additional no biomass burning simulation). The methods section could be rearranged some to improve flow, and I have a few questions regarding the methodology. The text of the results section is well written; however, the discussion could be expanded. Lastly, I believe there is room for improvement with some of the figures. Although many of the results in this paper have been documented previously (importance of biomass burning injection height, uncertainty in wet removal, and variability in biomass burning emission strength between inventories), I find there still to be novel aspects of this paper. The primary novel aspect of this paper is unlike most biomass burning studies, it evaluates model-observation agreement across multiple models. I believe that this paper could be an appropriate fit for ACP after major revisions to address the following concerns.

Specific Comments:

Introduction: I find the literature review of this study to be too brief. Given the number of studies that have reported on the impact of biomass burning plume injection height; the effects of BBIH on model-observation agreement/air quality found in prior studies should be discussed more. I think that at a minimum the impacts of BBIH on air quality and the ways different models simulate or assume BBIH could be separate more in-depth paragraphs.

Section 2.1: Could the authors explain the reasoning behind not including an experiment that uses the FEERv1.0-G1.2 and MISR plume injection height (a BBIH+BBEM simulation)? If this simulation also does not reproduce observations, I think it would strengthen the conclusion that changing the emissions and injection height are not enough to accurately simulate biomass burning plumes pointing towards biases in transport and/or deposition.

Page 6, Line 34: “It is assumed that, within each land cover region, the sampled plume profiles are representative of the entire region. This assumption is supported in part by statistical consistency across multiple cases within most land cover types.” I think at least some discussion on the limitations of using a monthly data set is warranted, given that fire strength impacts plume height and varies with time and space.

Page 7, Line 7: The final two paragraphs of this section feel as though they come in the wrong place. I think this section could be arranged in this general order: introduce MISR, introduce MINX, introduce how it is applied to the models used in this study, then discuss Figure 2.

Figure 2a: The boxes here are difficult to read since they are similar colors to the map below? Could the boxes just be black (or another color with large contrast to the map underneath)?

General Methods question: How is missing data from the observations handled when regionally and taking the April average of the model simulations? Are days with missing CALIOP data excluded from the model for that comparison? Are the latitudes North of the MODIS and MISR boundary excluded in the regional average? Is model data averaged at the time of the satellite overpass? How are the differences in satellite data availability incorporated into the Table 3 presentation where the median of all satellites is presented? These discussion points could be a part of Section 2.3.

Figure 3a & 3b have separate captions and should be separate figure numbers. Same comment for Figure 5a & 5b.

General Figures comment: Switch to a different sequential color map that is not jet based for Figures 2a and 3a, the colormap used for Figure 8 is ok. Switch to a diverging colormap for Figures 5 and 6 (e.g. goes from blue to red with white in the middle without the green/yellow/orange colors).

Page 9, Table 3: Include BBEM in this table with a reminder in the caption that BBEM doesn’t include GFDL. Either remove the “BASE/Satellites” and “BBIH/Satellites” rows or update the table caption to include this information.

Section 3.4: The summary vertical profile metrics of Z_a and F_2km are useful. However, I think a non-normalized metric would also be helpful since in terms of Z_a and F_2km the nobb simulations are often the closest to CALIOP.

General model-observation comparison: From the introduction and the discussion in Section 3.4 it seems that part of the decision to focus on April 2008 was the ARCTAS and ARCPAC field campaigns showing that smoke originating from the Boreal Asia fires impacted Alaska. I think the study could be improved by including model-observation comparisons to these field campaigns.

Page 14, Table 4 & 5: Both BASE and BBIH use the GFED4.1 emission inventory, for a given model (ex. GEOS), shouldn’t the total emission be the same between BASE and BBIH? Additionally, correct the column headers of emibboa and emibbbc to use words and acronyms that are defined.

General Discussion: There is little discussion on how this study fits into prior studies of biomass burning emission and biomass burning injection height. I think additional discussion on this front would be beneficial to understanding how this study expands the knowledge presented by prior studies. For example, Zhu et al. (2018) used GEOS-Chem and the MISR-based injection heights.

Technical Corrections:

Page 1, Line 17: “Each model performed four simulations: (1) BASE, using a common emission inventory with default injection height; (2) BBIH, with vertical distribution adjusted using MISR plume heights; (3) BBEM, with an alternative emission inventory; and (4) NOBB, excluding biomass burning emissions.” For (2) and (3) I would switch the word with to using.

Page 1, Line 29: “These findings suggest that increasing emission strength alone is insufficient; improving vertical injection near-source to loft more smoke above 3 km in Siberia and reducing excessive aerosol wet removal during transport are critical.” Clarify that “insufficient” and “critical” are referring to improving model-observation agreement of the biomass burning event. Also, I would argue that all three of these things need to be improved to simulate the Siberian wildfire event, the wording of this sentence makes the emission strength seem unimportant.

Page 3, Table 1: Since the column header is “Default emission altitude in boreal Eurasia” and the text explains that CAM5 and GFDL assume different emission altitudes outside this region, I think the table could be made easier to read by removing the “30° - 60°N:” and “Global:” parts of the default emission altitude in boreal Eurasia column.

Page 5, Line 10: “(used in the BASE run)” GFED4.1s is also used in the BBIH simulations.

Page 6, Line 7: “Pan et al (2020)” should read Pan et al. (2020).

The study investigates the sensitivity of biomass burning aerosol dispersion to injection height and source strength at four models participating in the AeroCom Phase III intercomparison. Particular focus is on an intense event of Siberian wildfires in April 2008. Simulations employing the default representation of plume injection height are compared with those using MISR satellite-derived plume heights, and results based on two different emission inventories are analyzed. The model outputs are evaluated against multiple active and passive satellite remote sensing datasets.

The manuscript explores an important and timely topic: the representation of biomass burning aerosol, in particular wildfire smoke, in climate and Earth system models. In light of the extreme fire events observed worldwide in recent years, and the expected increase in their frequency and intensity under climate warming, the study is highly relevant.

The manuscript is well structured and clearly written. The language is at a good level; however, most of the figures require revision. The multi-panel maps and vertical profiles are too small and hardly legible. Some overlaid boxes and certain legends are also difficult to read.

There are also some concerns about the content. While the approach with multiple sensitivity simulations is appropriate and has been well implemented on a seasonal basis, the focus on one month of a single past event seems too narrow. The available climate models and simulation results would allow a more comprehensive and statistically robust analysis. For example, the sensitivity of smoke to injection height and source strength in the models could be analysed for average and extreme events over a multi-annual period and for different vegetation types and climate regions. In their current form, the results do not differ significantly from what would be expected and is already known from previous studies. What is certainly new here is that several models were tested. By describing the model uncertainties in more detail and with more specificity, the paper could be improved. Beyond these more general considerations, several specific aspects of the manuscript need further clarification and elaboration:

(1) In the introduction, a more detailed discussion of the options for parameterizing smoke injection heights in models would be useful, as well as a clearer explanation of the range of the different emission inventories (including their rationale and uncertainties), since this is ultimately one of the main motivations of the study.

(2) Neither in the model description nor in the description of the emission datasets is there any reference to the specific biomass-burning aerosol species that are modeled. Are there differences in the emission composition across the inventories, and if so, how might these affect the results?

(3) The section on dry and wet deposition is interesting and relevant, but too brief. That wet deposition constitutes the dominant removal pathway for smoke aerosol is not surprising, given the typical particle size of smoke aerosol compared to, for instance, desert dust or volcanic ash.