This paper provides an improved framework for modeling isoprene emissions in the Finnish boreal forest by incorporating high-resolution, species-specific tree and emission data. The authors found that the new framework can better reproduce the measured patterns and further claimed an overall improvement in secondary organic aerosol estimates. Based on Figure 11, the performance on OA shows only a marginal improvement, so please revise this statement to be more accurate.

My major comments are:

(1) The authors compared the “simple” and “advanced” species distribution, noting 20% differences, but did not run the WRF-CHIMERE model using this simple specification. As this 20% differences might result in negligible differences in the CTM, I see the need to run a control simulation with this “simple speciation” to logically claim that highly detailed species distributions are essential for atmospheric chemistry modelling;

(2) Why only focus on isoprene changes in this study, as these species are not strong isoprene emitter at all, so the changes/improvement we see here might be overturned/non-valid if the authors also account for main-emitted compounds and their atmospheric interactions;

(3) “EFs calculated by MEGANv3.2 represent the leaf-level EF with units of nanomoles m−2 s−1 LAI−1”. I don’t think this is correct. If it is leaf-area EF, it should be in units of per leaf area. In Section 2.2, the manuscript states that a "careful conversion is required" to translate leaf-level EFs from MEGANv3.2 to canopy-scale EFs for MEGANv2.1. Equation 2 is presented for this conversion, with both LAI_std and LAI_max defined as 5 m² m⁻². Consequently, the scaling factor (LAI_std / LAI_max) equals exactly 1, meaning the numerical values of the EFs remain mathematically unchanged. For me, this is not correct, and equation 1 cannot convert the unit from per leaf area to per ground area in my understanding.

Detailed comments:

L110: All the simulations are performed "online". What does this mean? Do meteorological variables affect emissions at each time step?

L145-147: In MEGAN 2.1, EF is at the ecosystem level, but not at the gridcell level. The EF is still at the PFT level. It is important not to mix this. When used in CLM, for instance, the EFs are only used at the PFT level!

L149-150: Not sure I understand here. First of all, why only focus on isoprene emissions, as isoprene emissions are generally low or absent in these listed trees. Then, about this EF processor, what does it do with your measured EF? Before introducing the emission factor to be used in the model. It is important to describe how these measured EFs were derived/measured.

Equation 6: Why integrate SRR over a fixed vertical depth of 500 meters? I assume this fixed depth will systematically bias the AME metric by underestimating daytime but overestimating nighttime. How does this fixed depth affect Fig. 12?

L296-297: Where does the isoprene come from in the night?        

Fig. 11 the almost unchanged patterns when compared with the observed OA. Back to my question again: why improve only isoprene emissions in this manuscript?

M. Bettineschi and co-authors present a well-thought and valuable study addressing the model estimation of isoprene emissions and concentrations over Finland. Several previous studies have hinted at a significant overestimation of isoprene fluxes calculated using the MEGAN algorithm. Here, high-resolution distributions of the major tree species in Finland are used in MEGANv3.2 to provide improved emission estimates. These emissions are used in a regional model to calculate the concentrations of BVOCs and organic aerosols (OA) over Finland and neighboring areas. The isoprene fluxes and BVOC and OA concentrations are evaluated against in situ and flux data over Finland. Improved agreement for the isoprene fluxes and concentrations is obtained when the updated distributions are used, in conjunction with best estimates of the emission factors, as well as with a canopy correction to the wind speed and diffusion coefficients within the canopy. The study demonstrates quite well the importance of using local information on tree distributions in biogenic emissions models. The canopy correction also proves to be very valuable. The results for OA abundances are comparatively disappointing. The analysis suggests an underestimation of secondary OA formation from the oxidation of monoterpenes. An improved model representation of SOA formation is one of the main recommendations for future studies.



Although the study is interesting and very well-made, I have a few reservations, as explained below. They mostly concern the model description and the discussion of results. I'll recommend the paper for publication, if the authors address the following points.



Major comments



1) The MEGANv3.2 algorithm is not properly described. A reference (Guenther et al. 2020) is provided, but it is a book chapter (conference paper) and does not fully describe the MEGANv3.2 algorithm. There is some confusion regarding the units and definition of the emission factors - providing equations with properly defined variables would make it easier for the reader to understand the algorithmic differences between the versions 2.1 and 3.2 of MEGAN. Also, a conversion is required  for the reported emission factors of the main tree species, since the literature generally gives EF per gram of dry matter, instead of per square meter. I'm also not sure whether the standard conditions of MEGAN (either 2.1 or 3.2) are identical to the standard conditions of the reported emission factors. These aspects should beclarified.



2) One important aspect of the study is overlooked: the chemical mechanism of isoprene oxidation in WRF-CHIMERE and in the model studies (Ciarelli et al., 2024; Cholakian et al., 2022) cited in the introduction to back the claim of isoprene overestimation against in situ data. The oxidation mechanisms in these studies (including this one) are obsolete, and lack OH-recycling mechanisms (epoxide formation, peroxy radical isomerisation, etc.) that are now widely accepted to occur (see e.g. Wennberg et al., https://doi.org/10.1021/acs.chemrev.7b00439; Novelli et al., https://doi.org/10.5194/acp-20-3333-2020). The absence of such reactions might lead to isoprene overestimation in source regions, especially at remote locations such as the boreal forest. The manuscript should be amended to acknowledge this limitation, and if possible, to provide a crude estimate of the possible impact. More caution is advised when discussing the causes of model overestimation of isoprene levels. Among the recommendations of the paper, the implementation of a more appropriate chemical mechanism is advisable.



3) Another important aspect is the large variability in the reported isoprene emission factors for Norway spruce (see Table 2 in Hakola et al., 2023). This variability implies an uncertainty that might exceed the differences found here between different emission estimation methods. I think that this point requires a further examination and discussion of the available data.



Minor comments



l. 31 Isoprene reacts also with O3. The relative contribution of OH, O3 and NO3 to the total isoprene sink has been estimated using models (https://doi.org/10.5194/acp-19-9613-2019, https://doi.org/10.5194/gmd-12-2307-2019).



l. 45 I would add that these differences highlight the complexity and uncertainties of SOA formation, as well as the importance of accurate estimation.



l. 51 Harrison et al 2013 is not the best reference for this aspect.



l. 65-71 I could not find any mention of isoprene overestimation over boreal forests in Zhao et al. (2024); in fact, their model was found in good agreement with measurements along the Trans-Siberian.



l. 116-117 What are these reactions rates? Please clarify.



Figure 1 legend "the shading represents the percentage of the ceell covered by forest": I cannot see this on the plot.



l. 141 In the Supplement, it would be useful to see a plot of typical vertical profiles of the correction factors to the diffusion coefficient and windspeed.



Figure 2. Please add color bars to each subplot. It would be interesting to compare the subplot d) with the EF from the unmodified MEGANv3.2 model.



Eq. (1) Define LAIstd and LAImax.



l. 297. The temperature underestimation being of the order of 1 degree at night, it probably cannot  explain the large flux underestimation.



l. 325-330 The calculated effect of isoprene on OH might be unrealistic.



Fig. 10. I am not convinced that the normalized concentration plot makes much sense. The subplots a and c are sufficient for the discussion.

 

l. 393-397 If typical pyrogenic compounds were measured at the site (e.g. CO or CH3CN), filtering of the BVOC dataset based on this filter could help removing the effect of fires.



l. 460 This OH reduction is dependent on the chemical mechanism, which is not a strong feature of the model used here.



Sections 4 and 5 of the paper are somewhat long, with some repetition. I suggest revising, and possibly merging, these two sections.





Technical/language corrections



l. 6 "reduction in bias": not clear. With respect to what?

l. 8 insert "the"  before "vertical", and delete the word "concentrations"

l. 9 replace "additionally" by "further"

l. 10 delete the commas

l. 11 "transportation" --> "transport"

l. 11 and 14: avoid repetition

l. 16 and 17: delete "the" before "atmospheric"

l. 19 insert a comma after "globally"

l. 22 delete "the" before "biogenic"

l. 24 "among which"

l. 25 mention the formula of monoterpenes (C10H16)

l. 35 "a different range of ..." : many words, little meaning. Maybe replace by "lower-volatility compounds, which might partition to the aerosol phase..."

l. 35 Replace the comma by a new sentence.

l. 41 replace "bigger" by "higher"

l. 42 "the oxidant"

l. 52 delete "it"

l. 55 delete the second "emissions"

l. 56 "... factors accounting for..."

l. 58 The unique EF value is not just challenging, it is impossible. I think the idea is clear from the first part of the sentence.

l. 59 "... in the same PFT coexist..." weird, rephrase. Maybe "the same PFT includes..."

l. 90 "BVOCs driven aerosol dynamics" weird wording, please rephrase.

l. 109 "the parent domain has". Same mistake on l. 110.

l. 158 "bigger" -> "larger"

l. 193 "can be chosen"

l. 258-259 The correction is not described in the methods section but in the Supplement.

l. 309 high-emission area

l. 323 "when compared to..."

l. 429 "Further improvements for future work" weird, please rephrase.

l. 456 "The absence, or too simplistic representation, ..."

l. 457 "in underestimated BSOA production"

l. 479 "shows a significant improvement"

The manuscript by Bettineschi et al. tests the effects of using tree species-specific emission factors in the calculation of isoprene emissions by the bottom-up model MEGAN and the resulting simulated concentrations of BVOCs and OA at different various heights above the ground. It also examines the possible causes of model underestimates in non-isoprene BVOC (especially monoterpenes) and OA concentrations such as lateral boundary conditions, missing emissions from wetlands and fires, and the limited representation of SOA formation chemistry. To achieve these, they apply multiple models including MEGAN v2.1 and v3.2, WRF-CHIMERE, and FLEXPART. Based on their results, the authors show the importance of accurate EFs for bottom-up BVOC emission estimate (like in MEGAN), canopy correction, other emission sources and chemistry updates. Overall, the manuscript presents an interesting study of the effect of isoprene emission updates, and is well written and relatively easy to read (despite with some difficulty in methodological descriptions). It fits the scope and requirement of the journal. Below are some suggestions to clarify the methodology and strengthen the findings of the manuscript.

1. The authors have done a thorough investigation of the effect of EFs, with a caveat that the effects of activity factors are absent. I understand the focus of this study is EF, but since the activity factors could also substantially affect the resulting emissions, it would be valuable to examine the (potential imperfectness in) activity factors and discuss their potential effects on emissions. The simulated temperature and wind speed is examined already at one site, and it would be useful to look at other factors (such as solar radiation) and at other locations. I understand the ground-based measurements may not be available, but some comparison with existing assimilated meteorological datasets might still be useful to get a rough estimate the potential uncertainty in these activity factors and their impacts on calculated emissions.
2. Given the limited number of ground measurement sites, it would be useful to include some evaluation of modeled VOC concentrations based on satellite measurements of VOCs such as HCHO, CHOCHO, and isoprene. Those data are often freely available. This evaluation would provide a more comprehensive evaluation of the uncertainty and your improvements in BVOC emissions, to complement your current results from ground sites.
3. Although the manuscript is overall clearly written, the methodological descriptions related to the use of SRR and SRC could be further clarified. What is the exact meaning and formula for SRR? What is the meaning of “air mass arrival time” in Eq. 3? Does SRC consider the effect of the characteristic lifetime of OA (spent from the emissions of its precursors to the deposition of OA)? If so, how? What is the meaning of omega (I guess it means some sort of spatiotemporal domain, but it should be clearly defined)?
4. In addition, the difference between leaf-level EF and canopy-level EF could be better described. What is the physical rationale of the conversion in Eq. 1? Eq. 1 suggests the converted EF to have exactly the same unit as the EF before the conversion, which appears to be contradictory to the difference between leaf-level EF (per LAI basis) and canopy-level EF (not per LAI basis) shown in Line 182-183. What are the meanings of LAI_max and LAI_std, and where are their specific values from?
5. Furthermore, different sets of experiments are shown for isoprene emissions and WRF-CHIMERE simulations, and it is hard to link the emission-focused experiments to WRF-focused experiments. To help understand these experiments, it would be good to add a table (or expanding Table 1) to describe both emission-focused experiments and WRF-focused experiments side by side.

Some minor suggestions:

The boundary layer mixing scheme, along with convection scheme and the fact that the canopy process is not well represented, could be added in Method (i.e., at the beginning of Sect. 2.1), to improve the understanding of results, particularly on the vertical profiles.

Line 174-175: “This simplified version was tested because detailed information on tree species distribution is often unavailable in many regions.” – it would be good to show the extent of such limitation (for example, in the SI).

Line 179-181: “Specifically, for Norway spruce we used the average value of all the EFs available, while for Scots pine and Silver birch we assigned the minimum value present in MEGANv3.2, as for both species extremely small EFs are reported in Finland.” – add references.

Figure 4: Change “+ Trees emission factor” to “+ Tree emission factor updates”

Figure 10: To improve readability, it would be better to reduce the number of colors, for example, use red color dashed line for CC Day.

Line 423-425: “Because the standard MEGANv3.2 configuration includes species information only for the contiguous United States, this limitation introduces considerable uncertainty in global applications.” -- This information could be already mentioned in the introduction or method section.

Line 456-457: “The absence, or more detailed representation, of LVOCs, ELVOCs, and ULVOCs in the VBS scheme used in this study can therefore lead to two main issues” – Grammar error.