the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jul 2026
Species-explicit inventory of forest biogenic volatile organic emissions across China with MEGAN v3.2
Abstract. Biogenic volatile organic compounds (BVOCs) are important precursors of secondary organic aerosols and tropospheric ozone, yet their emissions are commonly modelled at the plant functional type (PFT) level, which obscures substantial interspecies variability. This limitation is particularly acute for China, where high forest diversity and ambitious afforestation initiatives are expected to reshape national BVOC budgets. Here, we present a species-explicit BVOC emissions inventory for China covering 234 tree species (https://doi.org/10.5281/zenodo.20396128, Liu et al., 2026). The inventory was developed within the Model of Emissions of Gases and Aerosols from Nature (MEGAN) v3.2 modeling framework, with refined inputs comprising species-specific emission factors and high-resolution forest composition data for China. Total forest BVOC emissions were estimated at 10.26 Tg in 2019, with summer emissions accounting for about 50 % of the annual total and a clear decreasing gradient from southern to northern China. Emissions were highly concentrated among a limited number of species: the five largest contributors, Pinus massoniana, Quercus liaotungensis, Phyllostachys edulis, Cunninghamia lanceolata, and Quercus variabilis accounted for 41.7 % of total emissions while occupying only 25.4 % of forest area. At the compound-class level, Quercus liaotungensis dominated isoprene emissions, whereas Pinus massoniana was the leading contributor to monoterpene emissions. Applying this species-explicit framework to two future afforestation scenarios with identical planted areas but contrasting species composition, we found that BVOC emissions may increase by 4.65 Tg yr⁻¹ with the biomass-maximization tree species and by 5.10 Tg yr⁻¹ under the most environmental-suitability species. The dominant compound class and species contributors differed markedly between scenarios, indicating that afforestation-driven BVOC responses depend strongly on species selection. These results demonstrate the importance of incorporating species-specific emission traits BVOC models and suggest that future afforestation strategies could substantially reshape both the magnitude and chemical composition of biogenic emissions, with implications for atmospheric chemistry and air quality.
- Preprint
(2364 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 04 Sep 2026)
- CEC1: 'Comment on egusphere-2026-3708', Astrid Kerkweg, 17 Jul 2026 reply
-
RC1: 'Comment on egusphere-2026-3708', Anonymous Referee #1, 27 Jul 2026
reply
This study presents an impressive dataset applicable for BVOC emissions modeling in China, providing estimates of tree species distribution maps, emission factor database along with improvements in LAI and land cover classification over China. The Appendix provides the EF’s for isoprene and monoterpenes, while the associated Zenodo archive provides estimates of total BVOC emissions at grid box level. Some evaluation is provided by intercomparing emission isoprene emission fluxes at available measurement sites, as well as SOA concentrations. The isoprene emission totals are put into perspective by comparing them with literature values.
While this manuscript is very valuable, and provides important input to the discussion of impact of different vegetation on the BVOC emissions in different scenarios, and to the science community by providing a compilation of EF’s, still some aspects can be improved.
The evaluation section should be strengthened by giving more emphasis to the isoprene flux evaluation, which is the primary means of evaluation of the improvement obtained here. Presentation of the evaluation results in the Table should be promoted to the primary text rather than the Appendix.
Contrary, the evaluation against SOA observations should either be strengthened as well, either by presenting, analyzing and intercomparing the results from different emission scenarios, and better understanding and interpretation of the observations and modeling errors, or alternatively giving this less weight by moving the results to the appendix.
Also, while the authors present EF’s for isoprene and monoterpenes, they ignore the presentation of EF’s for other types (sesquiterpenes, and other BVOC’s), while they still report on their overall magnitude. Also sesquiterpenes are considered important for SOA formation. Therefore it is not traceable how they obtain the EF’s for these values; please provide more details on this aspect here.
More generally, it would help in the judgement if the authors provide traceability as to how all the numbers (EF’s) are derived, either by providing a reference (citation) of one of the 184 peer-reviewed studies, or by an alternative datasource.
Likewise, one of the main outcomes of this study is a gridded dataset of forest tree species distribution. Yet this dataset appears not available through Zenodo. Also the actual MEGAN code, used to ingest this dataset, and convert this to emissions, is not shared in the Zenodo archive. I understand this is a requirement of GMD, following also the Executive Editor comment.
Detailed comments:
Line 388: Table C3 is briefly introduced and discussed. But this table is put together rather superficially, while it contains key information regarding the fidelity of the model and its improvements. Therefore I recommend to put more emphasis on this evaluation aspects. For instance, the table caption appears incomplete, not properly describing what is in the Table. (Missing units). It would additionally help to get information on the absolute average emissions from the observation, to judge if the mean bias / RMSE numbers are significant.
On linke 386 the description of the performance at LA is incomplete and should be checked more carefully. The improvement in performance is only seen for monoterpenes and not for isoprene. Then this table may be promoted to the main text rather than given in the appendix.
Likewise, Sec. 3.2 provides an assessment of SOA, which is a derivative of the alternating BVOC trace gas emissions, associated to many uncertainties. Currently it only reports on the quality of simulation Exist-SPE, but it would be logical to also discuss the other experiments, as done for the BVOC emissions. A closer assessment of these evaluations would be welcome, e.g. selecting only observations and time periods that are known to be affected by SOA from BVOC would be advised. Otherwise, this evaluation is not really informative, and could/should be left to the appendix instead.
On line 427 the authors report on sesquiterpenes emissions, as well as other VOC species. It is actually unclear to me how these numbers are derived, given that no species-specific information for other species than isoprene and monoterpenes is provided - do the authors use reference EFs from MEGANv3.2 for these?
Finally, in all figures showing maps of China, please exclude the inset with the wider area over the seas south of China mainland: This does not contain any information that is relevant to this work on forest emissions - at least it is not visible due to the small size, and only distracts the reader.
technical comments
Line 36: “1000 Tg of reactive carbon” …, change to “1000 Tg of trace gases containing reactive carbon” … ?
Line 79: please spell out (explain) the units once. Especially I’m unfamiliar with the ‘gdw’ unit.
Line 330: “simulate SOA”: change to: “simulate tropospheric chemistry and aerosol, including BVOCs and SOA” ?
Line 472: ‘Exist-PEC’ => ‘Exist-SPE’
Citation: https://doi.org/10.5194/egusphere-2026-3708-RC1 -
RC2: 'Comment on egusphere-2026-3708', Anonymous Referee #2, 28 Jul 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-3708/egusphere-2026-3708-RC2-supplement.pdf
-
RC3: 'Comment on egusphere-2026-3708', Anonymous Referee #3, 29 Jul 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-3708/egusphere-2026-3708-RC3-supplement.pdf
-
RC4: 'Comment on egusphere-2026-3708', Anonymous Referee #4, 03 Aug 2026
reply
Species-explicit inventory of forest biogenic volatile organic emissions across China with MEGAN v3.2
By Liu et al.
Biogenic volatile organic compounds (BVOCs) exert substantial impacts on climate change and air quality, as they act as critical precursors to tropospheric ozone and secondary organic aerosols (SOAs). To provide more comprehensive estimates of BVOC emissions, this study develops a species-resolved emission inventory built upon the MEGAN v3.2 modeling framework, combining with species-specific emission factors and high-resolution forest distribution datasets for China. Based on these methodological improvements, the manuscript further addresses China’s 2060 carbon neutrality goal and associated large-scale afforestation initiatives by comparing two prevailing afforestation pathways: biomass maximization (BIO) and ecological suitability (SUIT). This manuscript presents the first systematic evaluation of how tree species selection modulates BVOC emission magnitudes and chemical composition, giving the work both fundamental atmospheric science value and direct policy relevance for forestry planning. The scientific contribution lies in establishing the complete research chain from "afforestation species planning → BVOC emissions → ozone/SOA" and breaking the single-dimension evaluation of carbon-centric afforestation by introducing an integrated greening assessment framework that balances carbon sequestration potential against air quality implications. Overall, the study’s design and logical narrative are largely complete, yet several deficiencies remain: cursory mechanistic interpretations of scenario results, insufficient uncertainty analysis, non-standard figure annotations, and disorganized reference formatting. I recommend the manuscript be accepted with minor revisions.
1. The introduction only broadly describes China’s rich tree species diversity but fails to highlight the core statistic that "234 dominant species cover 90% of the national forest area" upfront. Since this statistic is the foundational justification for constructing a large-scale species-resolved inventory, placing it early in the introduction would immediately establish the study's necessity and prevent readers from questioning why exactly 234 species were selected for simulation.
2. The research objectives outlined in Lines 97–103 are overly general and fail to clearly articulate the three core innovations of this work. The authors should explicitly enumerate these advances to clearly distinguish their research from prior plant functional type (PFT)-based BVOC modeling studies:
(1) a localized species-resolved emission factor database covering 234 native species;
(2) a disaggregation algorithm for allocating mixed-stand areas to individual species;
(3) a compound-specific BVOC assessment under two carbon-neutrality-oriented afforestation scenarios.
3. The key compositional contrast, isoprene dominance in BIO versus monoterpene dominance in SUIT, is described in scattered paragraphs without a standalone summary statement; given that this is the most critical finding for policy audiences, the authors should extract and highlight this conclusion in a prominent standalone sentence to ensure that non-modeling practitioners (e.g., forestry and environmental agency staff) can grasp the core result at a glance.
4. Under the BIO scenario, the emission increment hotspots for Quercus aquifolioides are concentrated across Southwest China. The current explanation only references the species’ distribution without integrating emission factor properties and regional climatic controls. To fully interpret this spatial hotspot pattern, the authors should combine the high isoprene emission factor of this species (listed in Table C4) with the region’s characteristic warm summers and strong solar radiation to deliver a complete mechanistic explanation.
5. Under the SUIT scenario, Pinus yunnanensis occupies the largest planted area yet yields lower annual emissions than Pinus massoniana. The manuscript only briefly attributes this discrepancy to low temperatures at high altitudes. This is a typical case of species-environment coupling and deserves a more rigorous explanation. The authors should compare the two pine species’ monoterpene emission factors from Table C4, and explicitly quantify how low temperatures at high altitudes suppress metabolic activity, thereby co-driving the emission gap through both intrinsic emission capacity and environmental constraints.
6. Both scenarios adopt 2019 meteorology as a static baseline, with no discussion of biases introduced by excluding future climate warming effects. Rising temperatures will substantially enhance BVOC emissions across all tree species, meaning the current projected emission increments constitute conservative lower-bound estimates. The authors must explicitly acknowledge this limitation in the discussion section and note that ignoring warming-induced emission amplification imposes a negative bias on their emission projections.
7. The policy implication section lacks sufficient depth and only provides a generic recommendation that BVOC emissions should be considered during afforestation planning. Given the study’s strong applied value, the authors should provide more actionable, region-specific management advice. For example: moderately reducing the proportion of high-monoterpene species like Pinus massonianaand Hevea brasiliensis in southern afforestation projects, while concurrently evaluating ozone formation risks when deploying high-isoprene oak species for carbon sequestration in southwestern China.
Citation: https://doi.org/10.5194/egusphere-2026-3708-RC4 -
RC5: 'Comment on egusphere-2026-3708', Anonymous Referee #5, 18 Aug 2026
reply
The manuscript describes estimation of biogenic VOC emissions in China using the MEGANv3.2 model. The important achievement of presented work is a compilation of species-specific emission factors with high-resolution forest composition data for China. In combination with updated MODIS LAI and refined MODIS land cover data the authors prepared a detailed regional input data that were used to calculate Chinese BVOC emission inventory for the period of 2018-2022. The authors performed a very thorough analysis of the newly obtained emission dataset by comparison with emissions calculated with less precise land cover inputs (PFT level), evaluating also contribution from individual tree species, to show the added value of the detailed tree land cover description and species-specific emission factors. The species-resolved as well as the PFT based BVOC emissions were compared to measured emission rates at four forest sites. Additionally, the species-resolved BVOCs were used in the WRF-chem model and the resulting SOA concentrations were evaluated with measurements. The manuscript also discussed potential future changes in BVOC emissions if the afforestation and reforestation (A&R) scenarios are applied for China. The two presented scenarios vary in tree species to be planted either to maximize the biomass or to satisfy the highest environmental suitability. Indeed, it is interesting to add yet another parameter, i.e. BVOC emission change (and eventually its impact on the atmospheric chemistry), on top of other environmental implications when creating A&R strategies. I acknowledge the effort to simulate future LAI with series of models trained on current time data and dependent on identified parameters that will carry the future land cover change.
The authors made a great and important effort to compile data of detailed tree species distribution in China together with their assignment with species-specific emission factors. Despite some inevitable assumptions and simplifications, this represents a valuable input dataset for BVOC emission modelling in Chinese region. The newly input data were successfully used in the MEGANv3.2 model and the resulting emissions were thoroughly evaluated and discussed. The manuscript is written comprehensively, with clear structure and very good English. It falls well within the scope of GMD. I recommend the manuscript for publication after addressing the following questions.
Questions to be clarified:
- Can the authors please make clear what is the unit of the emissions? Is it Tg of BVOC species or Tg(C)? This is especially important when presenting the BVOC total (e.g. abstract, Table 2, Fig. 4, Conclusion), or the percentage fraction of individual BVOC to total BVOC (e.g. in Sect. 3.3, 4.1, 4.2), given the different molecular weights of different BVOCs. It would be clearer and without doubt or confusion, to state the emissions in Tg(C) and then calculate the percentage contribution of individual BVOC species to the emission total.
- The soil moisture parameterization of Pegoraro et al. (2004) implemented in MEGANv2.1, and I believe in MEGANv3.2 as well, was shown not to represent the drought stress on plants leading to isoprene reduction accurately (e.g. Opacka et al. (2022) and references therein). Can the authors please confirm that the soil moisture parameterization they used in MEGANv3.2 (gamma_SM) is based on Pegoraro et al. (i.e. depending on soil moisture and wilting point value) and if so, can they estimate how much the soil moisture parameterization impacts their final isoprene emissions?
- The performance of the WRF-Chem model with Exist-SPE to simulate BSOA is discussed (Sect. 3.2). Did the authors look at how would the model performance compare with measurements when using the Exist-PFT emissions with regard to BSOA?
Note:
It would be interesting to see the impact of the BVOC emission increase under the future A&R scenarios (BIO and SUIT) on atmospheric chemistry. I understand this was out of the scope of the current paper. Please see this as a suggestion that could be taken up in the follow-up study.
References
Opacka, B.; Müller, J.-F.; Stavrakou, T.; Miralles, D.G.; Koppa, A.; Pagán, B.R.; Potosnak, M.J.; Seco, R.; De Smedt, I.; Guenther, A.B. Impact of Drought on Isoprene Fluxes Assessed Using Field Data, Satellite-Based GLEAM Soil Moisture and HCHO Observations from OMI. Remote Sens. 2022, 14, 2021. https://doi.org/10.3390/rs14092021
Pegoraro, E.; Rey, A.; Greenberg, J.; Harley, P.; Grace, J.; Malhi, Y.; Guenther, A. Effect of Drought on Isoprene Emission Rates from Leaves of Quercus Virginiana Mill. Atmos. Environ. 2004, 38, 6149–6156.
Citation: https://doi.org/10.5194/egusphere-2026-3708-RC5
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 217 | 88 | 18 | 323 | 13 | 15 |
- HTML: 217
- PDF: 88
- XML: 18
- Total: 323
- BibTeX: 13
- EndNote: 15
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear authors,
although you provide the reference to a zenodo archive for the MEGAN code, I do not find a code availability section. Please add this before or after the data availability section in you article.
Best regards,
Astrid Kerkweg (GMD Executive Editor)