Summary

The authors perform a series of sensitivity tests with the GEOS-Chem chemical transport model to explore how various model processes impact model performance over Southeast Asia during two biomass burning periods in 2014. The authors primarily conclude that the addition of SVOCs to default biomass burning emission inputs notably improves model performance, especially during the March time frame. The authors also note the general dearth of chemical measurements available across Southeast Asia and the need for additional observational constraints to inform model-based assessments of public health impacts from biomass burning in the region. 

This is a useful analysis that could inform future work on mitigating public health impacts from regional and global biomass burning. However in my view there are two lines of inquiry and discussion currently missing from the analysis that should be included to justify the conclusions drawn by the authors, especially given the relevance for future public health applications.

General Comments

1.    While the authors clearly spent substantial time and resources testing the sensitivity of modeled biomass burning PM2.5 and AOD to the emission inventory used as well as several other emissions-related assumptions, there is very little discussion in the paper about the impacts of assumptions made within the AOD calculation itself. For example, Hammer et al 2016 found that increasing the assumed absorption from biomass burning aerosol in GEOS-Chem to better match satellite observations meaningfully impacted global OH. In my view it is insufficient to draw conclusions about improved model performance based on model-satellite AOD comparisons without some exploration of the AOD assumptions themselves. If an additional sensitivity simulation or two to further explore key assumptions in the AOD calculation is not feasible, perhaps at a minimum a robust discussion of these factors/potential uncertainties could be added to the text to more fully contextualize the results presented. Assumed aerosol composition (smoke vs. urban vs. biogenic vs dust), optical properties by PM2.5 component (OC, EC, sulfate, nitrate, dust), and distinguishing between cloud and smoke are a few of the key areas that would seem relevant to me when it comes to comparing modeled AOD to satellite-retrieved values. 

Hammer, M. S., Martin, R. V., van Donkelaar, A., Buchard, V., Torres, O., Ridley, D. A., and Spurr, R. J. D.: Interpreting the ultraviolet aerosol index observed with the OMI satellite instrument to understand absorption by organic aerosols: implications for atmospheric oxidation and direct radiative effects, Atmos. Chem. Phys., 16, 2507–2523, https://doi.org/10.5194/acp-16-2507-2016, 2016.

2.    Perhaps this was explained and I missed it - it seems contrary to me to first emphasize the lack of available observational constraints in Southeast Asia, but then proceed to advocate for a more chemically complex modeling approach to address smoke-related applications in the region. Wouldn’t the lack of measurements regionally warrant at least an exploration of simpler methods to capture biomass burning PM2.5 compared to the more sophisticated representation of organic species suggested by the authors? Maybe the answer is simply needing to dial up emissions of primary OC to begin with to address fires/burned area missed by the satellite products? Given the nature of the conclusions drawn by the authors this strikes me as another line of inquiry that should be addressed somewhere, ideally through a separate sensitivity simulation to compare a simplified scaling of primary OC with both the satellite AOD and the surface PM2.5.

 

Line-by-line

Line 84-85. Since you mention testing other biomass burning inventories you tested, perhaps list them quickly here for clarity/easy reference?

Line 162. Just a suggestion given the sparsity of the AERONET data points apparent in Figure 3 - did you also look at the AERONET 1.5 data? My understanding is that especially in this region a lot of data get filtered out between 1.5 and 2 during severe smoke episodes, but the filtered smoke episodes are still evident sometimes in the separation between the v1.5 fine vs. coarse product.  

Marvin et al. investigated the sources of uncertainty for biomass burning aerosols in Southeast Asia with GEOS-Chem and tried to reduce the uncertainties by adjusting several parameters in the model. The uncertainties were further related to public health impacts in the region. The manuscript is well-written and well-motivated. However, I have some concerns about the results that I think should be addressed in the revised version.

Major comments:

1. The evaluation of model aerosols seems to be limited by the availability of PM observations in the SEA. I wonder if some additional insights could be obtained from evaluation using observations of other biomass burning-related species, e.g., carbon monoxide? Since the simulations were full-chemistry in the troposphere, there should be a handful of choices.

2. As mentioned in the manuscript, the GEOS-Chem nested grid simulations rely on boundary conditions from a global spin-up run. Maybe I missed this point in the paper, but were the modifications listed in Table 1 applied to the global spin-up runs or only to the nested grid runs? Namely, I wonder if the authors could clarify this or discuss the potential impact of uncertainties in biomass burning from other regions close to the boundaries via long-range transport, e.g. , the southern part of China and North Australia. As a follow-up comment, how well could the insights from this study be generalized to other regions of the world?

3. The impact of uncertainties in vertical transport was briefly mentioned in the paper. A recent study (Wizenberg et al., 2023) showed that the underestimation of model fire emissions relative to surface observations could be largely attributed to the injection height scheme in GEOS-Chem, which suggests transport could be a very important source of uncertainty. If another sensitivity run with changed injection height is not feasible, could the authors slightly expand the discussion on this in the paper?

Wizenberg, T., Strong, K., Jones, D. B. A., Lutsch, E., Mahieu, E., Franco, B., & Clarisse, L. (2023). Exceptional wildfire enhancements of PAN, C2H4, CH3OH, and HCOOH over the Canadian high Arctic during August 2017. Journal of Geophysical Research: Atmospheres, 128, e2022JD038052. https://doi.org/10.1029/2022JD038052

Other points:

Figure 1(b): It would be better to use scientific notation in the colorbar.

Figure 3: Is there a reason why only SS1 is shown for March and only SS2 is shown for September?

L86: It would be great if the authors could clarify how the daily and 3-hourly scaling factors were applied.