the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Nov 2024
Significant spatial and temporal variation of the concentrations and chemical composition of ultrafine particulate matter over Europe
Abstract. Ultrafine particles have attracted interest as perhaps the most dangerous fraction of atmospheric PM. This study focuses on the characterization of ultrafine particulate matter (PM0.1) mass concentrations and their chemical composition during a summer and winter period in Europe.
Predicted levels of PM0.1 varied substantially, both in space and in time. The average predicted PM0.1 mass concentration was 0.6 μg m-3 in the summer, higher than the 0.3 μg m-3 predicted in the winter period. PM0.1 chemical composition exhibited significant seasonality. In summer, PM0.1 was mostly comprised of secondary inorganic matter (38 % sulfate and 13 % ammonium) and organics (9 % primary and 32 % secondary). During the winter, the fraction of secondary inorganic matter increased, with sulfate contributing 47 % and ammonium 19 %, on average. Primary organic matter contribution also increased from 9 % in summer to 23 % in winter, while secondary organic matter decreased significantly to 6 % on average during winter.
During summertime, the model performance at 12 sites for daily average ultrafine particle volume (PV0.1) concentrations was considered good, with normalized mean error (NME) equal to 46 % and normalized mean bias (NMB) equal to 15 %. For the winter period, the corresponding values for daily average levels were -27 % for NMB and 64 % for NME, indicating an average model performance.
Correlations between PM0.1 and the currently regulated PM2.5 were generally low. Better correlations were observed in cases where the primary component of PM0.1 was significant. This suggests that there are significant differences between the dominant sources and processes of PM0.1 and PM2.5.
- Preprint
(1672 KB) - Metadata XML
-
Supplement
(2041 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3357', Anonymous Referee #1, 21 Feb 2025
General comment
Mataras et al. present in their manuscript a study on the characterization of ultrafine particulate matter (PM0.1) mass concentrations and their chemical composition during a summer and winter period in Europe. Ultrafine particles have attracted interest because they may be the most dangerous fraction of atmospheric particulate matter. In general, very few modelling attempts aimed at ultrafine particle mass have been conducted, especially over Europe. The Authors used PMCAMx-UF model, an Eulerian regional three-dimensional chemical transport model that is an extension of the PMCAMx model. They predicted/modelled the levels of PM0.1 and its chemical composition over Europe and specifically at a series of 12 urban and rural sites. They, further, evaluated the performances of the PMCAMx-UF model for daily average ultrafine particle volume (PV0.1) concentrations. I think the paper is well written and it is neatly exposed. The literature cited is adequate and so are the graphics. It presents valuable results and conclusions and contributes to the growing knowledge on ultrafine particles pollution. Therefore, I consider this paper suitable for publication on Atmospheric Chemistry and Physics. Given the paucity of published model works about ultrafine mass concentration, and especially in Europe region, this paper can represent a useful addition to the field if these issues are addressed.
I suggest a minor revision of the manuscript according to my specific comments before to consider it for publication on the Journal. There are a few issues that may warrant some additional thought and perhaps some further analysis.
Specific comments:
Title: The title of the manuscript does not seem fully reflect the core content of the study. It lacks key terms that would help identify the specific focus and contributions of the work. I suggest revising the title to better align with the main findings and scope of the research.
In Line 101-102 the Authors stated that an extremely low volatility secondary organic aerosol (ELSOA) component was simulated and then a related results was reported in Table S3. However, nothing was reported and discussed about it in the manuscript. Why? It should be important to discuss about this component.
From Figures S2 and S4, it is evident that several time series (i.e. Hohenpeissenberg, Aspvreten, Dresden, Finokalia) have a temporal gap and they are incomplete for the period under consideration. This could certainly affect the comparison between the measurements and the model outputs. It might be worth addressing this issue in your work or at least explicitly mentioning it in the article.
Throughout the manuscript, several acronyms are not explicitly defined, which may affect readability. I recommend defining each acronym upon first use to improve clarity for the reader.
Abstract: Line 12-14. This sentence appears to mislead the reader regarding the true focus of the study. The work is not centred on measuring the mass concentration of ultrafine particles but rather on modelling these concentrations and evaluating the model’s performance through a comparison with measurements provided in literature. I recommend revising this sentence to better reflect the core objectives and contributions of the study.
Line 65-66. Authors reported as a reference Bernardoni et al, 2017, but their measurements were in north Italy and not in California.
Line 73-75: This sentence may be misleading for the reader. Studies on ultrafine particles outside the United States are numerous and cover a wide range of topics beyond just number concentration or size distribution. Perhaps the authors intended to refer specifically to the mass concentration of ultrafine particles.
Line 151-156. The authors use the volume concentration (PV₀.₁) directly to avoid estimating the average density of ultrafine particles when converting to mass concentration. However, they later take the mass concentration from the model output, estimate the density based on chemical composition, and convert it back to volume concentration. This approach raises questions about its advantage, as the added steps seem to undermine the initial reasoning for using PV₀.₁ directly. It seems that this approach is taken because there are no measured data on chemical composition to derive the particle density, while the predicted chemical composition from the model is available. However, this reasoning should be explicitly stated, as the phrase 'to avoid complications' is too vague and does not clearly justify the methodological choice.
Line 204, 214. Here (line 204), it is stated that elemental carbon in Paris is the dominant component, accounting for 30%, and the same behaviour (line 214) is observed in winter. Is this just a coincidence, or is there an underlying explanation for this pattern?
Line 235-237. This sentence is ambiguous and may confuse the Reader. It is unclear what it refers to and seems to contradict the statement in line 232. I suggest clarifying it to ensure consistency and avoid misunderstanding.
Line 253-254. In this sentence it should be convenient to explicit that the UFPs emission were “overestimated” and not “overpredicted” to avoid confusion.
Paragraph 4.4. The objective of the study presented in this paragraph is not entirely clear. The authors provide a detailed discussion of the comparison results between PM₀.₁ and PM₂.₅ (ultrafine and fine particles), but the ultimate purpose of this comparison remains rather unclear. It might be helpful to better explain the aim of this analysis in the text. While the purpose of this comparison is somewhat explained in the Conclusions (lines 334-335), it would be beneficial to elaborate on it more clearly in this paragraph to provide better context and understanding for the Reader.
Figure 4. The labels (a) and (b) are missing between the two panels.
Figure 6. In this figure it should be convenient for the Reader to insert the standard error (as error bars) in the measured data and an error bar for the modelled data.
Figure 8. As in Figure 6.
Figure 9. In the caption it should be convenient to explicit that R2 is the Pearson coefficient of the scatter plots reported in Figure S5.
Figure S1. As in Figure 6.
Figure S3. As in Figure 6.
Citation: https://doi.org/10.5194/egusphere-2024-3357-RC1 -
RC2: 'Comment on egusphere-2024-3357', Anonymous Referee #2, 08 Apr 2025
Mataras et al. have performed simulations with a chemical transport model (PMCAMx-UF) to compare mass/volume concentrations and composition for particles smaller than 0.1 µm at several cities across Europe over 2 distinct seasons. They find the model performance is modest in some cities with specific biases in other cities over winter. The authors provide strong motivation for this work, including a robust set of methods to answer the questions posed. A minor objection to the study is the use of dated episodes and model versions, both of which could have been updated to present findings for the modern atmosphere. My primary concern, however, is that the paper runs a single configuration of the model and compares predictions to evaluations on mostly a time-averaged basis (over the episode) across a dozen sites. The text reads more like a technical report with little depth in analysis and interpretation. In my view, this is a missed opportunity and a lot more insight could have been extracted from these comparisons. With some effort, it should highlight what the model gets right and where it fails, with implications for how emissions, chemistry, and deposition are represented in chemical transport models, especially for particles smaller than 0.1 µm. At this point, I do not recommend publication of this work in ACP and I would like the authors to either expand the analysis and interpretation significantly before resubmission or consider submission as a technical report.
Major comments:
- Lines 119-121: More needs to be said about why episodes from 13 and 16 years ago were simulated and used for evaluation. How would the findings from this work change for the present (i.e., 2025) given significant changes in the sources and processes linked to UFP over the past decade?
- Lines 122: This seems like a much-dated version of WRF, which has now moved to 4.6 and beyond. At a minimum, the key WRF updates between the version used here and the latest version need to be discussed.
- Section 2 needs to have a separate paragraph describing and discussing the various processes that are likely to strongly influence PM0.1 concentrations, size distribution, and composition. It should be made clear where the model is reasonably accurate and where some of the outstanding uncertainties lie.
- Model-measurement comparison for UPF mass: There are uncertainties built into both the modeled (e.g., numerical diffusion) and measured (e.g., shape factor, UPF density) concentrations of UPF mass/volume. These need to be considered during the evaluation. I understand why 100 nm is considered but I feel that this threshold is arbitrary. Ideally, like PM2.5, a cutoff size based on the low point in the ‘trough’ between the Aitken and accumulation models would be better suited.
- Sections 4.1 and 4.2: The results from this study need to be placed in context of other PM fractions (e.g., PM1 and PM2.5 concentrations and composition) (some of which has been done in Figures S5 and S6) and compared with past work, with the literature review for studies from CA and US as a starting point. The concentration correlations and compositional differences (or similarities) with other PM fractions will provide answers on why and if smaller particles might be more important for health endpoints.
- PM1 and PV0.1: The model predictions for these and comparisons against measurements are perhaps the most important output of this paper. While Section 4.3 offers a qualitative description of the comparisons, I found it lacked depth in interpretation and insight. Many of my questions remained unanswered: (i) why are the timeseries data so variable and how does it compare with that for, say, PM2.5?. (ii) what can we learn about the emissions, transport, and physical/chemical pathways for PM0.1 and PV0.1?, (iii) in what way are these source terms different between <0.1 and <2.5 µm, (iv) what causes the spatial and seasonal variability in model performance?, (v) are there ways in which the model predictions can be presented that allow for more information to be extracted (e.g., if <0.1 µm particles are tied to anthropogenic combustion sources, would presenting the data as a ratio to EC or CO make more sense)?, (vi) how do the raw size distributions compare between predictions and observations (missed opportunity for a more thorough comparison) and what do they say about aerosol microphysical processes (e.g., nucleation, growth)?
Minor comments:
- Line 36: how is specific surface area defined and what are the units?
- Line 46: Always expand the first use of abbreviations (i.e., UCD-P here). Please review this for other abbreviations throughout the manuscript.
- Section 2: How many size bins were used to represent the aerosol size distribution?
- Line 135: A better approach, documented in Murphy et al. (ACP, 2023) and references therein, is to tie the IVOC emission rate to the total VOC/NMOG instead of POA.
- Line 169-170: A useful way to average would be to only consider continental regions.
- Line 199: SOA was only 2-10%. This isn’t true during the summer and the text does not make it clear if you are talking about summer or winter results. This comment applies elsewhere in the manscript as well.
- Scatter plots corresponding to Tables 1 and 2 could be useful to assess the overall performance across cities as well as the model skill in reproducing PV0.1.
Citation: https://doi.org/10.5194/egusphere-2024-3357-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 300 | 59 | 50 | 409 | 49 | 14 | 16 |
- HTML: 300
- PDF: 59
- XML: 50
- Total: 409
- Supplement: 49
- BibTeX: 14
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1