Review comments on “Multi-year high time resolution measurements of fine PM at 13 sites of the French Operational Network (CARA program): Data processing and chemical composition” by Chebaicheb et al. submitted to Earth System Science Data.

##################################################

General comments:

This study by Chebaicheb et al. present a unique dataset with multiannual (2016-2021) measurements of ACSM and AE33 collected at 13 urban sites in France including submicron (PM1) aerosol species, OA, NO3, NH4, SO4, Cl and eBC. Detailed description of the measurement instrument, data handling and data quality control are documented. This is of particular importance for data users. In addition, the authors conducted comprehensively analysis of the datasets, for example composition fractions, seasonal and diel cycles of each site are presented. Also, the dataset is used to evaluate the chemical transport model (CHIMERE) simulation. Overall, the topic is suitable for ESSD, and the manuscript is clear written. I would support for publication in Earth System Science Data after some corrections and clarifications. 

1. About measurement uncertainty, since the authors trying to deliver an important dataset to the community, it’s crucial to discuss about the uncertainty of each reported variable. Is there any limitation of the measurement that the data user should be aware?

1. Data availability: according to ESSD policy, I think there should be an individual section describing the dataset structure, unit, user guide, etc.

1. The manuscript gives me the feeling that it is describing data (composition fractions, seasonal and diel cycles). While it seems lack of scientific contributions except for the data itself. Maybe this is fine for ESSD. I would suggest some more in depth discussion (for example, next point).

1. General suggestion about Section 3.3: I would suggest extend more analysis in this section by utilizing the dataset to evaluate model simulation and provide more insights to the model development. For example, it would be interesting to provide some maps about spatial distribution, and focus on evaluation of CTM instantaneous simulation of hourly mass concentration?

##################################################

Specific comments:

Abstract L37: 43-60%, mass or volume? It’s good to clarify.

Abstract L46: please expand CHIMERE

L75: About BC effects on climate (Jacobson et al., 2001), I believe there are many recent research results here.

L89: please specify what’s CARA program.

Section 2.2.1: here you mentioned ACSM, Q-ACSM, ToF-ACSM, AMS, I’m not familiar with them. In my opinion, it would be helpful for readers to provide a table to summarize some key aspects (advantages/disadvantages, uncertainties, etc.).

L244: MAC_ACTRIS 7.5 m2/g, I think you refer to BC. In my opinion, this value is with high uncertainty that can vary from about 4 to >10, and it could have huge effects on your measured eBC values. I would suggest some more discussions about it.

L250: could you justify here why you decide the acceptable AAE 0.7 to 3.0? What’s your explanation of measured values outside this range?

L264: what do you mean NR-PM1? Non-Refractory?

Figure S1: sometimes PM1>PM2.5, could you please clarify and elaborate?

L290: Wang, (2023) -> Wang (2023)

L338-339: the measurements are conducted in daytime only?

L351: it’s not clear what are the two sides you are comparing (22-30 % vs 9-20 %)?

Section 3.3: -> Comparison between observations and CHIMERE Chemical Transport Model

L468: I guess the spatial resolution of emission inventories you input into CHIMERE is even coarser than 7 km.

L482: OA is underestimated by a factor of 2-3 at all sites, while eBC is more or less ok (a factor of < 1.5). Could you elaborate more about this? Or could you suggest something about OA/BC ratio?

General suggestion about Section 3.3: I would suggest extend more analysis in this section by utilizing the dataset to evaluate model simulation and provide more insights to the model development. For example, it would be interesting to provide some maps about spatial distribution, and focus on evaluation of CTM instantaneous simulation of hourly mass concentration?

Data availability: according to ESSD policy, I think there should be an individual section describing the dataset structure, unit, user guide, etc.

The current work describes a new dataset of high-resolution PM1 measurements from the CARA program. The methodology is clearly described and the manuscript is well written. Measuring the chemical composition of PM is important in air pollution science, being a vital step towards the source allocation of air pollutants. I think the article is suitable for publishing in ESSD, but after the below comments are addressed by the authors.

Section 2.5

Since the SOA mechanism of Wang et al. (2024) (using GENEO and SSH-aerosol) is described in Wang et al. (2024) as using the SOAP mechanism of Couvidat (2015), I suggest moving the text describing the SOAP mechanism to after Wang et al. (2024) is referenced. Please then also define the SOAP acronym, while also highlighting that this mechanism concerns the gas-particle partitioning of only the organics. In addition, please also include reference to the thermodynamics module that is used for the gas-particle partitioning of the inorganic species (which I assume is ISORROPIA).

As a general note, please do not forget to updated the Wang et al. (2024) reference to include a link to the now published DOI.

Please specify the CHIMERE model version used for this study.

I would suggest moving the sentence “Boundary conditions were taken from CAMS CIFS global model simulations (Flentje et al., 2021).” to after the IFS acronym is defined, and to then define CIFS to highlight that these reflect boundary conditions for the chemical species.

Section 3.3

I think this section, and the CHIMERE model comparison as a whole, could be restructured. While the introduction contains background information on why PM measurements are valuable assets in the validation of atmospheric chemistry-transport (CTM) models, the exact motivation of including a model comparison in the context of the CARA dataset could be more clearly defined. In its current form, it is unclear to me why the model configuration is described in section 2.5, as it does not seem to fit the narrative of the paper at this point.

I would recommend adding a Section 4 divided into subsections that 1) motivate why a model-to-measurement comparison adds valuable information/results within the context of the CARA dataset, 2) model description moved from section 2.5 to here, 3) discussion of the model-to-measurement results, targeting clear implications or recommendations for the model configuration/implementation (also following comments of reviewer 1), to support the conclusion regarding the usefulness of the CARA dataset. A clear motivation for this model comparison should then also be included in the introduction of the paper (the CHIMERE model is routinely validated against (organic and inorganic) PM observations, also in France within the context of the referenced RI-Urbans project, so what makes the lessons that can be learned from the CARA comparison especially novel?). In the current manuscript, this motivation is mostly limited to the final paragraph of the conclusion.

In the context of the current work, I don’t see the added value of including the comparison to the four filter samples (outside of the CARA dataset) shown in Figure S11, and would suggest leaving this out altogether. If the authors nevertheless decide to include the filter sample measurements, I would suggest including a comparison and discussion of why and how these measurements are different than those made during the CARA program elsewhere in the text, and not in the model-to-measurement comparison section. Going by Fig. S11, the difference in these co-located measurements seem very large to me (as is also apparent from the reported difference in correlation statistics with respect to modeled concentrations), in my opinion warranting a detailed discussion.

Figure 7. Why is the model-to-measurement comparison limited to summer and winter? Currently it reads as if this is a limitation of the model, but if a simulation for the year 2018 was performed, surely this should not be a problem? The spring and autumn seasons could at least be included in the supplementary material for consistency with the seasons shown in Figure 5.

Data (availability)

A quick scan of the data reveals that there are many data entries for Cl and NH4 having values of 0.0055 and 0.142 ug/m3, respectively. Since this corresponds to exactly half their detection limits (defined as 0.011 and 0.284 uh/m3 in the readme file, respectively), I am not entirely sure what the origin of these numbers could be. Are they result of reaching the detection limit together with some kind of upper limit on the CE? For the CE the lower limit is specified at 0.5, but with the recurring entries scaling as 1/2 (=1/CE) times the detection limit, it would seem like there is also an upper limit of CE = 2 in place? But surely the CE can't be greater than 1?

As a general comment, I would recommend explicitly specifying the units (I assume ug/m3) in both the measurement dataset as well as the readme file.

If the ATOLL (Lille), Marseille, and Paris data described in the manuscript are derived from the same ACSM measurements as those described in Chen et al. (2022) (their Fig. 2), I think this should be mentioned somewhere in the manuscript text.

Since the CHIMERE model calculations are discussed in the paper, I think the relevant model outputs should also be included in the zenodo upload, even though they are not part of the CARA dataset. (at least following the submission guidelines for Geophysical Model Development – I can’t seem to find whether it’s different for ESSD).

The current link to the zenodo page containing the dataset points towards Version V1 which has restricted access.

Other comments

I suggest removing “Such chemically-speciated multi-year datasets have significant value for the scientific community, offering opportunities for future research, including source apportionment studies, trend analyses, and epidemiological investigations. They are also vital for evaluating and validating regional air quality models. In this regard,” from the abstract and instead leaving it for the introduction, where I think this type of information is more fitting.

Since the CARA measurements are presented as PM1, I would suggest also stating this in the title (rephrasing from “fine PM”).

A brief explanation of how the ACSM measurements of PM1 relate to total PM25 might be good in the context of air pollution (typically measured in PM25) early on in the text, for example by referring to the discussion on lines 270-280.

Line 368. I think the difference in eBC between Chen et al. (2022) and the results of the current work is an important result. While some of the reasons behind this are discussed in section 2.3.2, I think it would be good to expand upon this (e.g., can it be verified which method is more correct? What is the relative importance of the MAC vs. harmonization factor?). Not in the least because Chen et al. (2022) is also a recent dataset, with many of its measurements and instrumentation participating in the same RI-Urbans project referred to in section 2.3.2, but also because the reported (methodological) 41% reduction in eBC levels strike me as very large.

L456. The increase in OA/eBC ratios during the day in summer seem to be the result of decreasing eBC concentrations rather than increased summertime biogenic (S)OA concentrations, as argued for in the text. For example, in Figure 5 bottom-left, the summertime OA diurnal profile remains almost perfectly flat, while that of eBC shows a noon-time minimum.

Line 123. Please define the acronym ‘NR’ (in NR-PM1) here.

Line 149. I would suggest adding “(as discussed below)” behind “Middlebrook et al. (2011)” to highlight that the Middlebrook algorithm is discussed in more detail further on.

Line 252. I would suggest leaving out the words “carefully” and “thoroughly” to make the wording of this sentence more scientific.

L533-541. In the second paragraph of the conclusion, it is noted that OA is the predominant compound for the highest concentration levels in summertime at all sites, probably due to photochemical production. It is not clear to me where exactly this conclusion can be drawn from. In Figure 4, OA shows a pronounced summertime minimum, consistent with the daily mean (and nearly flat diurnal cycle) of OA being the lowest in spring and summer as shown in Figure 5. From Figure 4 and 5, it seems instead like the other species are low during summer, rather than OA being high.

References

Chen, G., Canonaco, F., Tobler, A., Aas, W., Alastuey, A., Allan, J., ... & Prévôt, A. S. (2022). European aerosol phenomenology− 8: Harmonised source apportionment of organic aerosol using 22 Year-long ACSM/AMS datasets. Environment international, 166, 107325. DOI: https://doi.org/10.1016/j.envint.2022.107325

Wang, Z., Couvidat, F., & Sartelet, K. (2024). Response of biogenic secondary organic aerosol formation to anthropogenic NOx emission mitigation. Science of the Total Environment, 927, 172142. https://doi.org/10.1016/j.scitotenv.2024.172142