Implementation of the ORACLE (v1.0) organic aerosol composition and evolution module into the EC-Earth3-AerChem model

Kakavas, Stylianos; Myriokefalitakis, Stelios; Tsimpidi, Alexandra P.; Karydis, Vlassis A.; Pandis, Spyros N.

doi:10.5194/egusphere-2026-37

Preprints

https://doi.org/10.5194/egusphere-2026-37

Preprints

22 Jan 2026

| 22 Jan 2026

Status: this preprint is open for discussion and under review for Geoscientific Model Development (GMD).

Implementation of the ORACLE (v1.0) organic aerosol composition and evolution module into the EC-Earth3-AerChem model

Stylianos Kakavas, Stelios Myriokefalitakis, Alexandra P. Tsimpidi, Vlassis A. Karydis, and Spyros N. Pandis

Abstract. Simulating the composition and evolution of organic aerosol (OA) in Earth System Models (ESMs) presents significant challenges due to the high computational demands of detailed chemical mechanisms. The computationally efficient ORACLE module employs the volatility basis set framework and can simulate secondary organic aerosol (SOA) formation from a range of precursors, including volatile (VOCs), intermediate-volatility (IVOCs), semi-volatile (SVOCs), and low-volatility organic compounds (LVOCs). In this study, a lite configuration of the ORACLE v1.0 module (ORACLE-lite) is implemented into the TM5-MP global chemical transport model (CTM), which represents the chemistry-transport component of the EC-Earth3-AerChem ESM. SOA formation from anthropogenic VOCs is neglected to reduce the number of surrogate species and further improve computational efficiency. For the standalone TM5-MP simulation, the global annual mean surface total OA concentration using ORACLE-lite is approximately 1.1 μg m⁻³, representing a 25 % increase compared to the previous version of the model. The annual atmospheric OA burden also increases by 50 %, reaching 3.67 Tg. Corresponding predictions from EC-Earth3-AerChem are slightly higher, with a surface total OA concentration of 1.16  μg m⁻³and an atmospheric burden of 3.83 Tg, representing increases of 30 % and 60 %, respectively, compared to the previous version of the model. Comparison of monthly measured PM_2.5 OA concentrations from Europe and the US with the corresponding predictions shows that the models bias is reduced by approximately half in the standalone TM5-MP simulation and by a factor of three in EC-Earth3-AerChem when ORACLE-lite is implemented. These enhancements enable more accurate and computationally feasible assessments of the climate impacts of individual organic aerosol components in future ESM studies.

Received: 04 Jan 2026 – Discussion started: 22 Jan 2026

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.

Download & links

Preprint (PDF, 2182 KB)

Supplement (2330 KB)

Download & links

Stylianos Kakavas, Stelios Myriokefalitakis, Alexandra P. Tsimpidi, Vlassis A. Karydis, and Spyros N. Pandis

Status: open (until 19 Mar 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-37', Anonymous Referee #1, 20 Feb 2026 reply
The manuscript “Implementation of the ORACLE (v1.0) organic aerosol composition and evolution module into the EC-Earth3-AerChem model” by Kakavas et al. describes briefly the implementation ORACLE (v1.0) organic aerosol model in TM5-MP atmospheric chemistry model. It is analyzed and evaluated using it as a standalone model driven with reanalysis data as well as coupled to EC-Earth3-AerChem Earth system model.
The manuscript fits in the scope of the journal addressing the need for a computationally efficient and scientifically accurate description for organic aerosol. It presents a novel implementation of ORACLE-lite module into TM5-MP which shows improvements in simulated OA concentrations in both configurations it is evaluated. Methods are scientifically valid, although the manuscript relies on its description a lot on previous papers published on ORACLE-lite. However, for example Pandis et al. (1993) paper is behind a pay wall. The results mainly support the interpretations and conclusions. The description could be more detailed so that the reader shouldn’t have to go through so much of previous publications on ORACLE. The paper is well structured, the language is fluent and precise.
I can recommend publishing the paper after the following issues have been addressed:
Line 88: “ORACLE reduces the computational cost by utilizing a small number of surrogate OA species by employing a novel lumping method.”. When compared to what?

Line 116: There is a description paper by van Noije et al., (2023) on this EC-Earth3-Aerchem which might be a more appropriate reference for this model. In addition, Döscher citation is for year 2021, but the reference is 2022.

Lines 180-186: “Note however, that the SOA formation from biogenic VOC emissions (isoprene and monoterpenes) is already represented in the models, as described by Bergman et al. (2022), while SOA formation from anthropogenic VOC emissions is neglected. As a result, the number of surrogate species used to represent OA and its volatility in ORACLE-lite was reduced from 18 to 9. An overview of the characteristics of the lite configuration of the ORACLE module used in this study is shown in Table 1.” This is unclear what is meant by this. Is SOA formation from biogenic VOC emissions treated separately using the approach by Bergman et al. (2022)?

It is also unclear how OA is partitioned between different modes. Pandis et al., (1993) is referred to, but I don’t have the access to the paper.

Are SOA species affecting aerosol radiative properties? If so, what assumptions have been used for optical properties?

Do SOA species affect cloud droplet activation in EC-Earth?

Line 231: “data were available from only 3 stations for the simulated period.” Could you have used data for another year or ran a year that would have more station data?

Page 9: Emissions between TM5-MP and EC-Earth look remarkably similar. Is there a good reason to show both of them? It would also be good to use some other colormap and scale, and put them in the same figure to see if there are any differences. Jet colormap is also not recommended in modern data visualization. The same comment applies to Figure 3.

Lines 456-458: “The incorporation of ORACLE-lite significantly improved the representation of OA formation and atmospheric behavior both in the standalone TM5-MP and the EC-Earth.“ This is an inaccurate sentence since the default TM5-MP does not simulate OA formation, rather they emit OA as primary particles. In addition, overall significance of improvement depends on which evaluation metrics we are looking at in Table 3. Bias shows clear decrease, but different error metrics not necessarily.

Lines 477-479: “The seasonal and spatial variability of SOA was also better captured, with higher concentrations predicted in regions with intense biomass burning and anthropogenic activity, such as India, China, and sub-Saharan Africa.” How do you determine better seasonal and spatial variability? To me, these are not evident from the manuscript.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-37-RC1
RC2: 'Comment on egusphere-2026-37', Anonymous Referee #2, 26 Feb 2026 reply

The current work provides a description of the implementation of the ORACLE-lite secondary organic aerosol (SOA) formation mechanism in the TM5-MP and EC-Earth modeling systems. Accurate modeling of SOA concentrations is important for both air quality and climate purposes, even though SOA modeling is often associated with considerable uncertainty and computational cost. The implementation and evaluation described in the current work serves as a useful source of documentation, both for the currently described models and for other models that may wish to implement the ORACLE-lite module in the future. The comparison against surface observations, while limited to North America and Western Europe, is an important element of this work. I can recommend this article for publication after the below comments have been considered.
Main comment
1. The authors note that only 3 stations from the EBAS database are available for the year 2005. However, following the link provided to the EBAS database finds that PM2.5 OC measurements are available from 4 stations (one in Spain, Italy, Germany, and Norway). The Italian, German, and Norwegian sites all have nearly continuous measurements over the course of the year. The Spanish site only has two daily mean measurements per month. Given that it is typical to calculate monthly means only when around 60-70% data availability is reached, I think the measurements from the Spanish site should be excluded from the analysis. At least so long as the analysis focuses on monthly mean values.
The Italian site (Ispra) is clearly influenced by anthropogenic wood burning emissions, reaching a monthly mean OC concentration of 22 ug/m3 in January (or 40 ug/m3 OM using the employed OA/OC ratio of 1.8). This value would skew the average monthly mean across all sites (whether the Spanish site is excluded or not) beyond the average of 2.2 ug/m3 currently shown for January in Figure 8. It therefore seems like the Italian site was excluded from the analysis. However, since this is a rural site (as are the German and Norwegian sites), and since the high wintertime concentrations are likely to have large contributions from the anthropogenic biomass burning emissions (and the resulting oxygenated OA formation) that is central to the current work, it is not clear to me why this station should be excluded.
The availability of OC measurements in Europe increases greatly in later years. For example, for 2010 there are 8 stations with good data availability, whereas for 2015 there are 20. I think the simulations should at least be compared to EBAS/EMEP observations for the year 2010 (implying that a new simulation with TM5-MP would have to be performed). Ideally, however, more recent years for both the TM5-MP and EC-Earth simulations would have to be considered, preferably for 2015 or later.
If the authors decide to continue using the TM5-MP simulation for 2005, I think the stations in Germany (Melpitz), Italy (Ispra), and Norway (Birkenes) should be evaluated in detail on an individual basis rather than as the average across all three sites.
Minor comments
1. It would be helpful if the ORACLE-lite description would be expanded to make the article more self-contained, even though more detailed descriptions can be found in the cited works. For example, it would be helpful if the K_OH rates that modify the volatility of the emitted organic compounds are included in Table 1.
2. Could the motivation behind neglecting SOA formation from anthropogenic VOC emissions be expanded? These are said to contribute only 15% to total global average surface OA concentrations, but this seems like a considerable amount (similar also to the calculated annual SOA-sv mass, roughly based on Table S1). The measurements are also underestimated by 8-13% with the current ORACLE-lite setup, suggesting that the additional SOA from VOCs would improve/reduce the model bias. The current runtime increase of 8% seems modest. How much slower would the model become with the addition of SOA from anthropogenic VOCs?
3. When the IMPROVE network is introduced, it would be helpful if it is mentioned that the main purpose of the network is to measure aerosols in remote areas of the United States, with the measurements therefore being representative of rural conditions.
4. The authors mention that measurements of OC are converted to OA using an OA/OC ratio of 1.8, while the traditional model setup is to assume that OA from all emission sources (treated as POA) has a ratio of 1.6. For reference, it would be helpful to include a description of the OA/OC ratios calculated/assumed using ORACLE-lite, and the resulting calculated ratios of the total OA.
5. In Fig. S13, could the y-axis be changed to show hPa rather than model level? Currently it is difficult to make out the vertical scales.
Editorial
ECMWF is defined on both lines 120 and 131.
SOA-sv is said to correspond to a C* value of 10^-2 ug/m3 on line 332. However, Table 1 shows a value of 10^1 ug/m3.

Reply

Citation: https://doi.org/10.5194/egusphere-2026-37-RC2

Stylianos Kakavas, Stelios Myriokefalitakis, Alexandra P. Tsimpidi, Vlassis A. Karydis, and Spyros N. Pandis

Supplement

https://doi.org/10.5194/egusphere-2026-37-supplement

Stylianos Kakavas, Stelios Myriokefalitakis, Alexandra P. Tsimpidi, Vlassis A. Karydis, and Spyros N. Pandis

Viewed

Total article views: 206 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
121	67	18	206	33	18	20

HTML: 121
PDF: 67
XML: 18
Total: 206
Supplement: 33
BibTeX: 18
EndNote: 20

Views and downloads (calculated since 22 Jan 2026)

Month	HTML	PDF	XML	Total
Jan 2026	74	38	9	121
Feb 2026	47	29	9	85

Cumulative views and downloads (calculated since 22 Jan 2026)

Month	HTML	PDF	XML	Total
Jan 2026	74	38	9	121
Feb 2026	47	29	9	85

Viewed (geographical distribution)

Total article views: 181 (including HTML, PDF, and XML) Thereof 181 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 28 Feb 2026

Short summary

The computationally efficient configuration of the ORACLE v1.0 module (ORACLE-lite) is implemented into the TM5-MP global chemical transport model, which represents the chemistry-transport component of the EC-Earth3-AerChem ESM. The models bias is reduced by approximately half in the standalone TM5-MP simulation and by a factor of three in EC-Earth3-AerChem when ORACLE-lite is implemented.


Total:	0
HTML:	0
PDF:	0
XML:	0