A mosaic of aerosol sources in the central Mediterranean: insights from dry deposition and single-particle analyses

Morán, Marcos; Schneiders, Kilian; Eknayan, Melanie; De Tomasi, Ferdinando; Ielpo, Pierina; Scerri, Mark; Nolle, Michael; Kandler, Konrad

doi:10.5194/egusphere-2026-5232

Preprints

https://doi.org/10.5194/egusphere-2026-5232

Preprints

11 Sep 2026

| 11 Sep 2026

A mosaic of aerosol sources in the central Mediterranean: insights from dry deposition and single-particle analyses

Marcos Morán, Kilian Schneiders, Melanie Eknayan, Ferdinando De Tomasi, Pierina Ielpo, Mark Scerri, Michael Nolle, and Konrad Kandler

Abstract. Flux-based dry deposition observations are scarce in the central Mediterranean, leaving unresolved how transport shapes deposited composition. Dry deposition was sampled concurrently at Lecce, southern Italy, and San Lawrenz, Malta, from spring 2017 to winter 2018 within the XMed-Dry network, using flat-plate collectors with 48–72 h exposures. Up to 5000 particles per sample were classified by scanning electron microscopy with X-ray fluorescence, and source regions were derived from FLEXPART backward simulations weighted first by hourly PM₁₀ and then by class-resolved deposition rates. Mineral dust dominated the deposited mass in 95 % of samples at inland Lecce but 69 % at coastal San Lawrenz, where the sea-salt flux was an order of magnitude higher. Dust sources peaked over northern Algeria and Tunisia at Lecce but over Libya at San Lawrenz, and sea salt at Lecce reached the northeastern Atlantic, increasingly with size, while at San Lawrenz it remained within the western Mediterranean. Intensity instead separates the classes. The five strongest events carried about 20 % of the dust mass at both sites but 40 % of the sulfate mass. Within several intrusions, silicate-sulfate and silicate-sea-salt mixtures peaked after the dust maximum, indicating progressive processing as the event decayed, and these episodes carried higher iron proportions across all size fractions. Such processing reportedly raises iron solubility, increasing bioavailable iron delivered to surface waters. Mineral classes intercorrelate strongly except Ca-rich particles, whose mass at San Lawrenz held through boreal winter, when Saharan input reached its seasonal minimum, indicating a local carbonate source.

Received: 30 Aug 2026 – Discussion started: 11 Sep 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, 11212 KB)

Supplement (11409 KB)

Download & links

Preprint (11212 KB)
Metadata XML
Supplement (11409 KB)
BibTeX
EndNote

Marcos Morán, Kilian Schneiders, Melanie Eknayan, Ferdinando De Tomasi, Pierina Ielpo, Mark Scerri, Michael Nolle, and Konrad Kandler

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-5232', Anonymous Referee #1, 11 Oct 2026
In this manuscript, the authors present a dry deposition dataset of about 1.5 years, which was collected at two sites in the central Mediterranean at the same time. The single-particle analysis is combined with backward trajectory modelling to identify the source regions of different particle classes. In my opinion, this dataset is valuable and rare, because flux-based and class-resolved dry deposition observations are very limited in this region. The SEM methodology is based on the established works of the group. Several parts of the data processing are carefully done, such as the blank correction, the exclusion of artifacts, the window correction, the plausibility check of Pb and the Voronoi-based bootstrap for uncertainty estimation. The topic fits the scope of ACP well. However, I think some central conclusions of the manuscript are not well supported by the presented analyses. Some of them are even in conflict with the authors’ own numbers or statements. The main concerns are listed as follows.

Major comments
The FLEXPART simulations used an inert tracer without gravitational settling as mentioned at line 230. But the deposited mass is dominated by particles larger than 10 µm. The supercoarse fractions of dust and sea salt are important in the size range of larger than 10 µm, which are important contents in this paper. The atmospheric lifetime of such large particles is usually only several hours. So it is hard to accept that the supercoarse sea salt at Lecce comes from the northeastern Atlantic, which is thousands of kilometers away. A simpler explanation should be considered: strong westerly or northwesterly winds produce long Atlantic footprints and, at the same time, produce more local or regional sea spray. The CWT method cannot separate the source location from the transport pathway.

In section 4.2, the aging signal and the implication for iron bioavailability are overstated. The phase-lag argument (Fig. 7) is based on only 5 windows at Lecce and 3 windows at SLS. Each window has 5 samples with coarse time resolution, and each class is normalized to its own maximum. In several windows, the described sequence such as “precede an upward trend in both the sulfate mix and sea-salt mix curves toward their own relative maxima” from Line 366 to 369 is not observed. No statistical test was applied, but the Discussion mentions a “statistically detectable aging signal” (Line 616).

In method section: some basic information is missing: (1) The number of samples at each site, the total number of analyzed particles. (2) The lower and upper particle size limits of the analysis. (3) How the 48-72 hours data were aggregated for the MA-CWT, correlation analysis and pollution roses. (4) The distance between the sampling head and ground level. (5) How hourly PM₁₀ in situ was measured.

Section 4.3.3. The argument for a local Ca-rich source is reasonable, and it may be one of the most interesting results of this work. But it can be much more convincing if the authors use the data they already have: (1) Construct pollution roses for Ca-rich particles only, because the quarries are located to the SW, S and SE of the sampler. If the quarries are the main source, a hot spot is expected in these directions, especially at low wind speeds. However, the Saharan dust also mainly comes from the south, so a hot spot in the southern sector alone cannot distinguish these two sources. Therefore, it is better to compare the Ca-rich pollution roses with those of the silicate classes, or to use only the winter samples, when Saharan transport is weaker. Besides, since each sample integrates 48-72 h, the directional signal may be smoothed, and this limitation should be considered when interpreting the results. (2) Morphological screening to distinguish coccolithophore from angular quarry fragments. (3) The Ca-rich flux at SLS is the same in summer and winter (3.1 × 10³ mg m^-2, Lines 532–534). This should be double-checked.

Minor comments
Line 4 and Section 2.2.1: The technique used here is confusing. If it is XRF, the minimum particle size can be detected should be around 10 microns, but in this paper, there are results smaller than 1 micron. I assume the technique is SEM with energy-dispersive X-ray spectroscopy (EDX/EDS), which detects the X-ray emission excited by the electron beam. It is not “X-ray fluorescence”. The manuscript itself uses “EDS” in Section 2.2.3. Please correct it in the whole manuscript.

Line 9: The sentence “Intensity instead separates the classes” is not clear. Please rephrase it.

Lines 85–89: The representativeness of the deposition rates measured by the flat-plate sampler for natural surfaces, especially the sea surface, should be discussed. This is important because biogeochemical implications are discussed later. The collection efficiency as a function of particle size and its dependence on wind speed are also relevant for the size-resolved fluxes and the pollution roses. The results of Waza et al. (2019) can be briefly summarized here.

Lines 102–106: The particle mass is estimated from the perimeter-based volume-equivalent diameter with two fixed densities. Please explain how the particle thickness is treated or influence the results. In addition, 2 fixed density values are used to estimate the mass, please perform a sensitivity test for varied density.

Lines 350-352: It is said that the Ca-rich proportion is “relative to total mineral mass”, but Fig. 5 is based on number. Please clarify which metric (mass or number) gives the values of 3.6%, 39%, 18%, 41%.

Lines 412-444 and Fig. 11: The cumulative totals (mg m^-2) are compared between the two sites, but the sampled durations (392 vs. 425 days) and the seasonal coverage are different. When comparing the two sites, the mean deposition rates (mg m^-2 d^-1) for the overlapping periods should be given. Please also explain why the duration of sea salt at Lecce (388.1 days) is different from that of dust and sulfate (392.1 days). Besides, the terms “deposition flux (mg m^-2)” and “deposition rate (mg m^-2 d^-1)” are not usual, because flux usually means per unit time. Please define them clearly at the beginning, or use “cumulative deposition” for mg m^-2.

It would be better to show some SEM images of particles in different classes in supplementary information.

Please tone down the conclusions based on Pearson correlation. A high correlation between the deposition rates of different classes may simply reflect common driving factors such as wind speed, and does not prove a common origin. Similarly, a moderate correlation does not prove independent sources.
Citation: https://doi.org/10.5194/egusphere-2026-5232-RC1
RC2: 'Comment on egusphere-2026-5232', Anonymous Referee #2, 31 Oct 2026

This paper by Morán et al. investigates the chemical composition, particle size distribution, temporal variations and potential source regions of atmospheric dry deposition particles at two sites in the central Mediterranean (Lecce in southern Italy and St Lawrence in Malta) between 2017 and 2018. By combining flux-based dry deposition observations, automated scanning electron microscopy–X-ray microanalysis (SEM–X-ray microanalysis), single-particle classification and FLEXPART-based backward transport modelling, the authors sought to establish a link between atmospheric transport pathways and the composition of deposited aerosols.
This study explores an important yet relatively under-researched area within atmospheric aerosol research. In particular, the combination of single-particle characterization with deposition flux observations is of significant value, as total aerosol mass concentration alone cannot fully elucidate the compositional diversity of atmospheric particulate matter or its deposition behavior as a function of particle size.
Overall, I believe this manuscript falls within the scope of EGUsphere and may be considered for publication following minor revisions. My specific comments are as follows:
Major comments
1. Page 6, lines 129–130: The text describes the detection limit as the upper bound of the central 90 per cent confidence interval, assuming α = 0.95. There may be inconsistencies in these statistical descriptions. Please clarify whether the authors intended to use a 90 per cent or 95 per cent confidence interval, and explain the method used to calculate the threshold.
2. Page 6, lines 152–158: The authors classify particulate matter using a compositional similarity approach based on normalized elemental contributions within a 21-dimensional elemental space. Whilst this method provides a quantitative classification framework, several important issues remain to be resolved. However, this classification scheme appears to rely primarily on elemental composition. Atmospheric particulate matter may contain complex internal structures and multiple mineral phases, which may not be distinguishable based solely on average elemental composition. I suggest that the authors provide a more detailed explanation of the classification thresholds.
3. Page 10, lines 250–257: The authors utilized the FLEXPART backward modelling method, employing an inert air tracer, to calculate the air mass sensitivity field over a five-day period. Please explain how the five-day backward modelling period was determined, and whether this period is sufficient to infer long-range transport pathways.
4. Page 15, lines 366–369: The explanation regarding atmospheric ageing has not yet been fully supported by single-particle observations. The authors suggest that the maximum abundances of silicate–sulphate and silicate–sea-salt mixtures occur after the peak in mineral dust deposition, indicating that atmospheric processing is gradual during the dissipation phase of the Saharan dust intrusion. This is an interesting observation. However, the authors should consider factors such as the relative contributions of different source regions, particle size distributions and size-dependent dry deposition rates, local meteorological conditions, and boundary layer dynamics during the evolution of the dust intrusion. Unless additional chemical evidence is available, it is recommended that this mechanistic interpretation be treated with caution.
5. Page 16, lines 373–377: The authors report that, across all particle size fractions, the proportion of iron atoms during peak dust deposition events was higher than under background conditions. However, the statistical robustness of this finding has not been adequately explained. My questions are: how many individual particles are included in each particle size fraction and deposition state? Is the difference between peak and background conditions statistically significant? Can differences in mineral composition or source regions explain the observed iron enrichment?
6. Page 31, lines 615–627: The authors note that dust particles deposited during the dissipation phase exhibit higher atomic abundances of iron across all size fractions. They subsequently note that atmospheric processing increases the solubility of iron and the bioavailability of phosphorus, thereby enhancing the transport of available nutrients to Mediterranean surface waters. Although this analysis provides the relative elemental abundance of iron in the analyzed particles, it does not provide measurements of iron solubility, dissolved iron concentration or bioavailable iron. How is the relative atomic contribution of iron linked to bioavailable iron? Please clarify this point.
7. Page 33: In the section ‘4.3.3 Distinguishing Local Carbonate Sources from Saharan Dust using Ca-rich Particles’, the authors suggest that the abnormally high levels of calcium-rich sediments observed in the St Lawrence region during winter are partly derived from local carbonate sources, in particular nearby limestone quarries. Calcium-rich particles may originate from a variety of sources, including Saharan carbonate minerals, local soil resuspension, construction activities, quarry emissions and biogenic marine carbonates. Elemental composition alone is insufficient to clearly distinguish between these sources. The authors should therefore provide a more cautious interpretation.
Minor comments
1. Page 3, in “2.1 Field Campaign”: Please provide the total number of samples collected at each sampling point and the number of samples retained following quality control.
2. Page 11, Lines 275–289: Please provide the complete MA-CWT mathematical formulation and clarify whether the standardised DCWT field is applied independently to each particle class and sampling interval.
3. Page 13: Please clarify whether the relative contributions shown in the figure are based on particle count or mass.
4. Page 17, Figure 8: It would be more helpful for understanding if the figure could show the number of particles for each particle size class and clearly label the uncertainty represented by the error bars.
5. Page 24, in the section of “3.8 General classes deposition-based trends (MA-CWT)”: The term "source regions" should be used cautiously when describing CWT-derived maps. Consider consistently using "potential source regions" or "transport-associated regions."

Citation: https://doi.org/10.5194/egusphere-2026-5232-RC2

Marcos Morán, Kilian Schneiders, Melanie Eknayan, Ferdinando De Tomasi, Pierina Ielpo, Mark Scerri, Michael Nolle, and Konrad Kandler

Supplement

https://doi.org/10.5194/egusphere-2026-5232-supplement

Data sets

Single-particle composition, mixing state and dry deposition fluxes from scanning electron microscopy at Lecce (Italy) and San Lawrenz (Malta), 2017-2018, with PM10 and surface wind observations from nearby monitoring stations M. Morán et al. https://doi.org/10.5281/zenodo.22146888

Marcos Morán, Kilian Schneiders, Melanie Eknayan, Ferdinando De Tomasi, Pierina Ielpo, Mark Scerri, Michael Nolle, and Konrad Kandler

Viewed

Total article views: 809 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
519	214	76	809	134	82	68

HTML: 519
PDF: 214
XML: 76
Total: 809
Supplement: 134
BibTeX: 82
EndNote: 68

Views and downloads (calculated since 11 Sep 2026)

Month	HTML	PDF	XML	Total
Sep 2026	274	110	30	414
Oct 2026	189	87	40	316
Nov 2026	56	17	6	79

Cumulative views and downloads (calculated since 11 Sep 2026)

Month	HTML	PDF	XML	Total
Sep 2026	274	110	30	414
Oct 2026	189	87	40	316
Nov 2026	56	17	6	79

Viewed (geographical distribution)

Total article views: 809 (including HTML, PDF, and XML) Thereof 809 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 11 Nov 2026

Short summary

Dust, sea spray and pollution particles settle out of the air, but few measurements exist for the central Mediterranean. Over two years we collected what settled at two sites, in southern Italy and on Malta, and examined the particles one by one under an electron microscope. Although the sites lie only about 600 kilometres apart, what settled differed strongly between them, changing with the season and with the type of particle, showing that a single site cannot represent the whole region.


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