the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Aug 2026
Comparison of the GECKO-A and SAPRC MechGen Atmospheric Chemical Mechanism Generators
Abstract. Atmospheric chemical mechanisms describe the photolysis and oxidation reactions of volatile organic compounds (VOCs), including bimolecular and unimolecular reactions of product radicals. Fully explicit representation of a single VOC can require thousands to millions of chemical reactions and species. Consequently, explicit chemical mechanism generation has proceeded through the use of automated tools that rely on observational data and structure activity relationships (SARs) to estimate rate coefficients and branching ratios. Here we compare, for the first time, two automated mechanism generators, GECKO-A (Generator of Explicit Chemical Kinetics of Organics in the Atmosphere and SAPRC MechGen (SAPRC Mechanism Generator). We specifically compare mechanisms for ten representative VOCs and the resultant properties of generated product mixtures as relevant for atmospheric chemistry. The observed differences in product mixtures can be explained by systematic differences between the mechanism generators after the initial reaction steps. GECKO-A has a more detailed representation of photolysis, including Norrish type-2 photolysis, which results in the formation of reactive alkenes from aldehydes. MechGen includes more alkoxy radical isomerization processes and autoxidation via H-shift reactions, the latter of which are not currently included in GECKO-A. Also MechGen assumes fast cyclization of alkyl radicals to form lactones or carbonates, which is not assumed by GECKO-A, and for which available theoretical predictions are inconsistent. These differences highlight needs for additional experimental data, particularly for uncertain products and yields, as well as for near continuous updating of these mechanism generation tools as sufficient data and new SARs become available.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(3753 KB) - Metadata XML
-
Supplement
(1297 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-4908', Anonymous Referee #1, 23 Sep 2026
-
RC2: 'Comment on egusphere-2026-4908', Anonymous Referee #2, 19 Oct 2026
This paper describes in detail a box model comparison study of chemical mechanisms for a structurally diverse range of representative organic species of atmospheric interest, derived using two of the most commonly used automatic chemical mechanism generators; GECKO-A and SAPRC MechGen. Observed differences are explained through identified differences in the structure activity relationships applied, thereby highlighting areas for further research focus.
The paper is overall written well and relatively easy to follow. It provides, for the first time, an in-depth observational description into how these mechanisms are derived and highlight differences in the generation methodologies of these two well known tools. This work will certainly be highly valuable to the atmospheric chemistry community and a useful companion to the respective model description publications. It should be published in a suitable Copernicus journal.
However, the work described in this paper is of a detailed, technical nature and therefore should be aimed to be published in a suitable more technically oriented journal, such as Geoscientific Model Development (https://www.geoscientific-model-development.net/). Both the GECKO-A and SAPRC MechGen description publications are published in GMD. If the work included comparisons to observations, such as detailed chamber experiments or constrained chemical process box models from field campaigns, then this work could also be suitable for publication in ACP.
I only have a few specific comments to address, listed below:
- One simulation scenario was chosen for the comparison (“rural” continental conditions several hours downwind of a major urban area). Given that this is a detail chemistry comparison of the two different generation methods, it would have been (potentially more) useful to compare under more extreme “remote” and “urban” conditions (see Aumont et al., 2026; https://doi.org/10.5194/egusphere-2026-4908). It would be interesting to see the potential differences under “low-NO” and “high-NO” conditions.
- I know this is a technical paper comparing two different mechanism generation systems/methods, but it does not give a sense as to which gives to more correct, under the comparison conditions. This would be difficult without direct comparisons to observational data (see above). Are there any plans for some systematic evaluations against chamber data? Have any been carried out previously? If so, it would be useful to mention here any previous evaluation work, and also a strategy for any future evaluation work (against the extensively available chamber databases already available or suitably designed bespoke chamber experiments).
- I agree with the set of increasingly chemically complex, structurally representative set of VOC species investigated in this work. However, what is an obvious omission is a representative aromatic and/or phenolic species? Why were these species emitted from this study?
- As a fair comparison of the organic chemistry generated by these two tools, the inorganic chemistry scheme should be the same. Why was the same inorganic chemistry not used in the model comparisons? How different are these schemes? It would be useful to give these in the supplementary (or at least a link to where they are available).
- One of the main outcomes of this work is the evaluation of the impact of differences in SARs on mechanism development. It would be useful to discuss this in more detail, to list in general where the main differences come from (in a table?), pointing towards where fundamental kinetic and experimental work should focus to aide in the developments and updates of the SARs.
- When first introducing the 10 VOCs that this work focuses on, it would be useful to the reader to give a table with structures of these compounds (add to table S2?).
- Line 125. I think it is unhelpful to the reader to describe the MCM v3.3.1 (https://mcm.york.ac.uk) as the “Leeds Master Chemical Mechanism”. Remove “Leeds” from this sentence.
Citation: https://doi.org/10.5194/egusphere-2026-4908-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 222 | 110 | 80 | 412 | 61 | 82 | 88 |
- HTML: 222
- PDF: 110
- XML: 80
- Total: 412
- Supplement: 61
- BibTeX: 82
- EndNote: 88
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors present a modelling study comparing the chemical mechanisms describing the atmospheric oxidation of a series of compounds representative of typical species in the atmosphere generated from two automated mechanism generators, GECKO-A and SAPRC MechGen. The manuscript is well written and provides useful detail regarding the differences between results of simulations. The manuscript will be of interest to the atmospheric science community and I have only minor comments that should be addressed prior to publication.
Although the manuscript is rather long, it would benefit from the inclusion of a brief discussion outlining previous work on mechanism validation, particularly involving comparison to experiment and/or theory. Examples of previous work in the literature using either of the mechanism generators and any previous mechanism intercomparison studies would also be beneficial.
In addition, although the authors have identified key assumptions in the mechanism generators that lead to differences in simulations for some systems, it would be helpful, if possible, to identify which mechanism might be more reliable in each instance.
Recommendations are made for additional experiments to improve mechanism development and to help resolve such differences, it would be helpful to provide some specific detail of the type of experiments (temperature-dependent kinetics, product studies, or site-specific kinetics for example) or systems of interest (classes of compound for example) that would be most beneficial for the development of automated mechanism generators.
The summary is a little focussed towards future developments in GECKO-A, can the authors comment on any needs or plans for development for MechGen?
Other minor comments:
Line 101 onwards: Are the cut-offs in MechGen fixed or user supplied?
Line 138: How does aggregation of properties in post-processing as opposed to consideration of heterogeneous processes at run-time potentially impact results? Are any physical losses included in the simulations?
Line 160: What are the differences between the inorganic chemistry in the two mechanisms?
Line 160-165: Please clarify how the constraint on RO2 and RCO3 potentially impacts results. What are the eight RO2 types?
Line 171: Please quantify ‘approximately the same’.
Line 181: Is OH reactivity defined here as the rate of OH consumption (-d[OH]/dt) or the rate coefficient describing the rate?
Line 213: Section numbers need checking.
Line 234: How much difference does assessment of mPOHR compared to summing the reactivity from species in the mechanism make? How significant are the differences in the amount of “lost carbon” between the two mechanisms?
Line 242 onwards (and elsewhere): Are there any chamber measurements that can be compared to the simulations, particularly for simulations where the two mechanisms differ?
Line 362: Are the units correct for 4-6 × 10-12?
Line 427: Is there any value in comparisons between mechanisms containing more than one parent VOC? Is this an area of potential interest for future work?
Line 450: Are values taken for specific time intervals in the simulations?
Line 464: What are the values based on in each mechanism? Are there experimental values available for comparison?
Line 473 (and elsewhere): The image quality for some of the figures should be improved before final publication.
Line 559: Are experimental data available for comparison or validation of either mechanism?
Line 573-574: Over what time period are the yields time-integrated?