Range-resolved retrieval of aerosol particle size distribution during Saharan dust intrusions over the Bavarian Alps using multiwavelength lidar observations

Posyniak, Michal Andrzej; Sitarek, Stefan; Speidel, Johannes; Stacewicz, Tadeusz; Vogelmann, Hannes

doi:10.5194/egusphere-2026-1060

Preprints

https://doi.org/10.5194/egusphere-2026-1060

Preprints

24 Jun 2026

| 24 Jun 2026

Range-resolved retrieval of aerosol particle size distribution during Saharan dust intrusions over the Bavarian Alps using multiwavelength lidar observations

Michal Andrzej Posyniak, Stefan Sitarek, Johannes Speidel, Tadeusz Stacewicz, and Hannes Vogelmann

Abstract. We present ground-based remote sensing observations of Saharan dust over the Bavarian Alps during three events on 29 February, 29 March, and 29 April 2024, with emphasis on aerosol particle size distribution (APSD) retrieval from multiwavelength lidar measurements. Raman–depolarization lidar observations from Garmisch-Partenkirchen were combined with sun-photometer data from the summit of Zugspitze. Long-range transport from North Africa was confirmed using back-trajectory analysis and aerosol forecasts. Elevated dust layers extending up to 6–7 km a.s.l. were observed, frequently structured into multiple layers.

APSDs were retrieved by applying an inversion method that directly substitutes predefined (bi-modal log-normal) size distributions into the lidar equations, reducing the inversion to the estimation of a limited set of microphysical parameters. This approach enables range-resolved retrievals of effective particle radius, yielding values up to 3–4 µm within dust layers and decreasing with altitude. Comparison with column-integrated sun-photometer retrievals shows consistent coarse-mode effective radii (1.2–1.4 µm). The results highlight both the strengths and limitations of APSD retrieval from lidar, particularly regarding assumptions on particle shape and refractive index.

Received: 24 Feb 2026 – Discussion started: 24 Jun 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

Michal Andrzej Posyniak, Stefan Sitarek, Johannes Speidel, Tadeusz Stacewicz, and Hannes Vogelmann

Status: final response (author comments only)

RC1: 'Comment on egusphere-2026-1060', Anonymous Referee #2, 31 Jul 2026

The manuscript addresses an important topic: retrieving range-resolved aerosol particle size distributions during Saharan dust outbreaks using multiwavelength lidar observations. The combination of lidar, AERONET observations, trajectory analysis, and operational models provides a valuable observational framework. The paper demonstrates that the APSD inversion software can be successfully applied to real dust cases over Central Europe and presents interesting first results. I really like that the authors openly acknowledge the limitations associated with spherical particle assumptions and fixed refractive indices.
However, some improvements of the current manuscript are still needed, and should focus on strengthening the discussion of retrieval uncertainties, clarifying methodological assumptions, and refining the presentation. There is no need in additional measurements or a different analysis. Specifically:
1. The manuscript presents retrieved effective radii but provides almost no quantitative uncertainty estimates. Since aerosol microphysical retrievals are highly sensitive to measurement errors and model assumptions, uncertainty bars or sensitivity analyses can be very helpful.
2. For mineral dust, Mie theory is a considerable simplification. Although the authors acknowledge this issue, the manuscript would benefit from a more quantitative discussion of how particle nonsphericity could bias the retrieved effective radius.
3. The inversion is based on a predefined bimodal lognormal distribution with several fixed parameters. Such constrained inversions may produce stable solutions but not necessarily unique ones. The paper should discuss more explicitly how strongly the retrieved APSDs depend on these assumptions.
4. Limited validation: The retrieved effective radius is compared only with AERONET coarse-mode values. Additional validation against established lidar inversion products, GRASP retrievals, or previous dust studies would strengthen confidence in the method.
5. The conclusions are based on three individual dust events. The presented cases demonstrate the applicability of the APSD retrieval method to Saharan dust observations. However, since only three events are analyzed, I recommend emphasizing that the reported effective-radius values and vertical structures are representative of the investigated cases rather than of Saharan dust transport in general. The broader applicability of the method could be highlighted as a perspective for future studies involving a larger statistical dataset. For example, in the Conclusions the manuscript currently states: "These observations provide compelling evidence for the frequent and significant impact of Saharan dust on atmospheric composition over Central Europe...", I would recommend something like: "The analyzed events illustrate the significant impact that Saharan dust intrusions can have on atmospheric composition over Central Europe." Similarly, instead of: "Effective radius profiles consistently showed large particle sizes..." I would suggest: "In the analyzed events, the retrieved effective-radius profiles showed large particle sizes within the identified dust layers..." Authors shall consider what is better and decide revision of the text.
6. Broader implications should be emphasized. maybe even in a separate section. The conclusions mainly summarize the observations. The manuscript would have greater impact if it discussed how this retrieval approach could contribute to future aerosol monitoring within ACTRIS, satellite validation, aerosol data assimilation, or climate model evaluation.
7. The retrieved effective radii should be discussed more extensively alongside previous lidar studies of Saharan dust, particularly those from SAMUM, EARLINET, and long-range transport experiments, to better position the present results within existing knowledge. I was missing these additional studies and literature background.
8. Finally, I also find that the manuscript doesn't explains much why particle size changes with altitude or transport history. I would suggest providing a bit deeper physical interpretation when possible.

Citation: https://doi.org/10.5194/egusphere-2026-1060-RC1
RC2:
'Comment on egusphere-2026-1060', Anonymous Referee #1, 15 Sep 2026
Posyniak et al. present multiwavelength Raman–depolarization lidar observations of three Saharan dust intrusion events over the Bavarian Alps in February, March, and April 2024, together with AERONET sun-photometer observations from Zugspitze and supporting HYSPLIT and ICAP-MME analyses. The central objective is to retrieve range-resolved aerosol particle size distribution and effective radius from the lidar observations using a direct-fitting inversion algorithm based on a predefined bimodal lognormal particle size distribution. The topic is interesting and potentially useful because range-resolved retrieval of coarse-mode aerosol microphysical properties remains challenging, particularly for nonspherical mineral dust. The combination of multiwavelength lidar and sun-photometer observations can also be powerful when used carefully. However, in its current form, I do not think the manuscript provides sufficient validation, uncertainty analysis, or physical justification for the APSD/effective-radius retrieval, which is the main technical result of the study. The retrieved size profiles are interpreted rather strongly despite substantial assumptions made regarding physical properties. The manuscript also does not fully utilize the multiwavelength Raman–depolarization capability of the instrument to independently support the aerosol typing and microphysical interpretation.
I therefore recommend major revision. My main comments are given below.
The introduction needs substantial revision. At present it is very broad, in several places vague, and does not establish clearly enough what the actual scientific problem is. The paper is specifically about range-resolved microphysical retrieval of transported Saharan dust from multiwavelength lidar, yet much of the introduction discusses aerosol–climate interactions in very general terms.

Some examples are:
lines 24–31, statements such as “the interplay between aerosols and climate,” “regional ramifications,” “dynamics of aerosol distribution,” and “dispersion patterns” are too general. The authors should be more specific about the physical processes relevant to this study, particularly transport, vertical layering, optical evolution, particle nonsphericity, and the challenges these present for lidar-based microphysical retrieval.

Around lines 32–40, the discussion of global desert dust sources and particle size needs stronger and more recent literature support. The description of particles with as the “smallest particles range” is misleading. This radius range includes particles that are already relatively large compared with the accumulation mode. The authors should describe the relevant dust size modes more accurately and use modern observational literature on source-region and transported dust size distributions.

The statement that dust particle size and shape are strongly controlled by wind speed is too simplified. Wind speed affects emission and lofting, but observed transported size distributions are also controlled by source mineralogy, aggregation, gravitational settling, turbulent mixing, cloud processing, and transport time. A more careful literature review is needed.

lines 48–50, the manuscript states that lidar observations revealed desert dust transport. The authors should specify which lidar observables were used in the cited studies, for example depolarization ratio, backscatter coefficient, extinction coefficient, lidar ratio, or spectral dependence.

lines 60–75, the discussion of lidar and sun-photometer capabilities needs more precision. Sun photometers and lidars provide optical and microphysical constraints, but they do not directly measure aerosol chemical composition. Aerosol type may be inferred from combinations of optical properties, but “composition” should not be presented as a direct retrieval.

Similarly, depolarization provides information on particle nonsphericity but does not uniquely determine particle morphology. The wording “particle shape can be concluded” should therefore be revised.

The statement that lidar measurements typically provide spatial resolution “of the order of 10 m” is possible in principle, but the manuscript should state the actual range resolution and smoothing used for this system rather than use a generic value.

The discussion of multiwavelength lidar is one of the most important parts of the introduction, but the manuscript does not sufficiently explain what additional information is gained from a 3β+2α+δ configuration. This needs to be linked directly to the APSD retrieval problem.

The paper should clearly establish the gap in previous work. In particular, what remains unresolved in current multiwavelength lidar inversions for coarse-mode, nonspherical Saharan dust, and how does the approach used here improve upon or differ from existing methods such as regularized inversion or synergistic GRASP retrievals?

lines 85–90, the manuscript states that the APSD retrieval software was previously applied to aerosol below cumulus clouds and over the Baltic Sea. Applying the same algorithm to aged Saharan mineral dust is a very different problem. The authors should explain why this algorithm is expected to remain valid under these conditions and what is novel about the present application.

2. The manuscript states that several Saharan dust events occurred during the first half of 2024, but only three dates were selected. This needs more justification.
lines 92–96, please specify how many dust events were observed between February and June, how many had usable lidar measurements, how many had valid multiwavelength data, and how many had simultaneous AERONET retrievals.

The event-selection criteria should be stated objectively. At present, the first and third events appear to have been selected largely because clear-sky conditions allowed lidar–photometer comparison, while 29 March was included because the dust event was considered unusually intense.

What defines “unusually intense” on 29 March? The manuscript should quantify the event intensity using AOD, modeled dust concentration, extinction, or another measurable parameter and place it in the context of previous Saharan dust events over Central Europe.

The CAMS dust mixing-ratio values reported around lines 124–129 need clearer formatting and interpretation. The units should be checked carefully, and the authors should explain whether these values refer to model aerosol mass mixing ratio integrated over a specific dust size class.

The vertical extent of each modeled dust layer should also be discussed more carefully. For example, what are the lower and upper boundaries of the dust layers, and how do these compare with the lidar-observed layers?

It would be useful to compare these layer heights and modeled concentrations with previously documented Saharan dust events over Germany or the Alps.

3. The TONI lidar is central to the manuscript, but the description of calibration and retrieval uncertainty is currently insufficient.
The authors state that full overlap occurs near 250 m. Please clarify whether this means measurements below approximately 250 m are excluded or corrected, and provide evidence or characterization of the overlap function.

The manuscript mentions ATLAS calibration and polarization crosstalk corrections, which is appropriate, but it does not provide quantitative uncertainties for the measured or retrieved quantities.

The following information should be provided, at least in the Supplement: signal-to-noise ratio for each wavelength; calibration uncertainty andsensitivity of the APSD result to signal noise and vertical averaging.

Since the paper relies heavily on range-resolved microphysical retrieval, uncertainty propagation from the measured lidar signals into the retrieved is essential.

The paper states that the system contains three elastic channels, two Raman channels, and two depolarization channels. However, many standard multiwavelength aerosol optical products are not shown in the Results. This is a missed opportunity. Spectral backscatter/extinction behavior, lidar ratio, Ångström exponents, and particle depolarization would provide independent constraints on aerosol type and particle size.

4. This is the most important part of the manuscript and, in my view, currently the weakest part in terms of validation.
The paper describes the APSD as “retrieved,” but the functional form is prescribed as a bimodal lognormal distribution. This distinction needs to be made clearer. The method retrieves parameters of a constrained two-mode distribution, not an arbitrary APSD.

There appears to be an internal inconsistency in the manuscript. In Section 2.2, the authors state that the number of modes and their widths are fixed a priori, whereas later the parameter set is written as , which implies that the mode widths are also optimized. This must be clarified.

The rationale for assuming a bimodal distribution should be demonstrated rather than simply imposed. Transported Saharan dust can exhibit broad and multimodal size distributions, and the inversion result may depend strongly on the assumed number and width of the modes.

The manuscript itself states that many parameter combinations yield comparable , after which the 300 solutions with the lowest are averaged. This indicates substantial non-uniqueness.

Why were 300 solutions selected? The authors should show whether the final APSD/effective-radius profile changes if, for example, 100, 300, or 500 near-optimal solutions are used.

The spread among those solutions could potentially be used to construct an uncertainty interval. At present this information is discarded even though it directly reflects inversion non-uniqueness.

The manuscript should show retrieval residuals or a closure test demonstrating how well the retrieved size distribution reproduces the measured signal ratios at each wavelength.

Without such a closure test, it is difficult to determine whether the profiles shown in Fig. 8 represent actual microphysical information or simply the best solution within a strongly constrained parameter space.

5. The spherical-particle assumption is a major issue for this application and should not be treated as a brief limitation.
The manuscript applies Mie theory to Saharan mineral dust while simultaneously using depolarization to identify the aerosol as nonspherical.

This directly affects the extinction and backscatter efficiencies used in the inversion kernel and therefore can affect the retrieved modal radius and effective radius.

The discussion around lines 230–238 acknowledges this limitation but does not quantify its effect.

The reference to Bi et al. on hydrated salt particles is not sufficient to establish that errors caused by nonsphericity are small for irregular mineral dust. The optical behavior of sea-salt particles and mineral dust is not interchangeable.

The authors themselves cite Hoshyaripour et al. and Huang et al., which provide more relevant evidence that dust particle shape affects scattering properties.

The authors should therefore perform a sensitivity analysis using physically reasonable nonspherical dust optical properties, or at minimum provide an estimate of the expected retrieval bias introduced by the spherical assumption.

If this cannot be done, the effective-radius values should be presented more cautiously as model-dependent retrievals rather than direct particle-size measurements.

6. The refractive-index treatment is not sufficiently documented.
The authors state that an average refractive index of approximately 1.46 at 532 nm was derived from AERONET measurements.

However, the APSD inversion is multiwavelength, so the complete wavelength-dependent complex refractive index used at 355, 532, and 1064 nm should be reported.

AERONET standard inversion products do not provide refractive index directly at 532 nm, so the authors should explain whether this value was interpolated or otherwise derived.

The imaginary component of the refractive index is not discussed, although it can influence scattering and absorption efficiencies.

The refractive-index retrieval from AERONET also becomes less well constrained under low aerosol loading, which is relevant here because the observed AOD values are relatively low.

A sensitivity analysis over realistic refractive-index ranges for transported Saharan dust should therefore be included.

7. In Figure 5, I do not think the interpretation should rely so strongly on the 1064-nm range-corrected signal.
A range-corrected signal is useful for identifying atmospheric structure, but it is not in itself an aerosol-type measurement.

Clouds, aerosols, attenuation, and instrumental response can all affect signal intensity.

Since the instrument provides backscatter, Raman extinction, and depolarization capability, the paper should present combinations of actual retrieved optical quantities rather than primarily the logarithm of signal intensity.

I recommend showing backscatter coefficient, extinction coefficient, volume or preferably particle depolarization ratio, lidar ratio where available, and a spectral color ratio or Ångström exponent.

This would also provide a much stronger independent test of whether the layers selected for APSD retrieval are actually dominated by mineral dust.

8. Depolarization interpretation needs more careful treatment.
The manuscript presents volume depolarization ratio, not particle depolarization ratio.

The values in Fig. 5 appear to be generally around or below ~0.2 in several layers.

These values should not be directly compared with literature values for pure-dust particle depolarization, because molecular scattering lowers the volume depolarization ratio.

The authors should therefore calculate particle depolarization where possible or clearly explain why that conversion cannot be performed.

If particle depolarization cannot be retrieved, the manuscript should be more cautious when describing the aerosol as “pure” or predominantly mineral dust based only on volume depolarization.

The paper should also discuss whether the observed values may indicate aerosol mixtures, rather than simply treating all enhanced depolarization as equivalent dust.

9. The 29 March event needs additional justification.
AERONET observations were unavailable because of cloud cover.

This does not necessarily invalidate the lidar retrieval, but the authors should demonstrate that the specific lidar interval used for APSD retrieval was free from cloud contamination and not affected by significant cloud attenuation.

The manuscript should provide an objective cloud-screening criterion for this case.

Since this event lacks independent photometer comparison, it is particularly important to show lidar optical products and retrieval uncertainty.

10. The AERONET AOD and SSA interpretation is currently too weak.
The observed AOD at 500 nm is only approximately 0.1-0.25 during the two days shown.

AERONET inversion products such as SSA and refractive index become much less constrained at low AOD.

The manuscript should state the AERONET product level/version and the quality flag of the SSA retrievals.

The authors should also discuss the applicable uncertainty under these conditions.

The conclusion that the aerosol on 29 April represents “pure mineral dust” and that 29 February represents a mixture of anthropogenic aerosol and dust cannot be supported confidently from SSA alone at these AOD levels.

The statement in the Conclusions that the February event presumably included residential wood-burning emissions is particularly speculative and should be removed unless independent evidence is presented.

AERONET fine- and coarse-mode AOD or other inversion parameters could provide useful additional information on aerosol type.

The relatively low AOD does not necessarily contradict a significant dust event because the photometer is located at Zugspitze, approximately 2962 m a.s.l., while the lidar is in the valley at 734 m a.s.l. The authors should explicitly discuss this difference in sampling geometry when comparing the two datasets.

11. Effective-radius retrieval in Figure 8 is the central figure of the manuscript, but I do not think the current presentation is sufficient to establish confidence in the result.
The profiles show substantial altitude-to-altitude variability and abrupt changes in effective radius.

Some of this may represent real layer structure, but without uncertainty bounds it is impossible to distinguish physical variability from inversion instability.

Please show uncertainty envelopes derived from signal uncertainties and/or the spread of the near-optimal solutions.

It would also help to plot together with one or more independent lidar quantities such as backscatter, extinction, depolarization, or fit residual.

The manuscript states that particle size generally decreases with altitude. This is not obvious from all four profiles and should be demonstrated quantitatively rather than inferred visually.

The authors should calculate layer-averaged with uncertainty for each identified dust layer and test whether any vertical trend is statistically meaningful.

The manuscript contains an inconsistency in the description of the number of retrieval cases. Section 3.2 states that three cases were chosen, but Fig. 8 shows four profiles: two from 29 February, one from 29 March, and one from 29 April. This needs correction.

12. The extinction-weighted comparison with AERONET effective radius is a useful attempt, but I do not think it can yet be described as validation.
AERONET coarse-mode effective radius and the lidar-derived quantity in Eq. (12) are not necessarily the same mathematical quantity.

AERONET effective radius is derived from moments of the retrieved coarse-mode size distribution.

In contrast, the lidar value appears to be calculated from the full bimodal APSD at each height and subsequently averaged vertically using extinction weighting.

The authors should preferably calculate a coarse-mode-only lidar effective radius and integrate the coarse-mode distribution over the same atmospheric column before comparing it with the AERONET coarse-mode product.

The extinction weighting used in Eq. (12) also depends on a Klett–Fernald retrieval with an assumed constant lidar ratio of 50 sr.

Since the system includes Raman channels, the authors should state whether Raman extinction measurements were available during these daytime periods. If they were available with sufficient SNR, they would provide a more independent extinction constraint than assuming a fixed lidar ratio.

Agreement between AERONET and lidar inversion products is useful as a consistency check, but both retrievals depend on assumptions regarding particle size, refractive index, and shape. It therefore should not be presented as independent validation.

The local lidar-derived values of 3–4 µm remain considerably larger than the AERONET coarse-mode effective radius. The manuscript should demonstrate that this difference is physically plausible using the vertically resolved size distributions, rather than simply attributing it to column integration.

13. The paper repeatedly emphasizes that TONI is a multiwavelength Raman–depolarization lidar, but the Results do not make full use of this capability. The authors should add spectral backscatter coefficients; spectral extinction coefficients where available; lidar ratios; particle depolarization and color ratios. These would substantially strengthen aerosol typing and provide an independent physical check on the APSD results. At present, the paper discusses the advantages of multiwavelength lidar much more strongly than it actually demonstrates them in the Results.
14. In the Conclusions:
The statement that the APSD inversion “proved effective” is too strong without a quantitative uncertainty analysis and independent validation.

The statement that lidar–photometer synergy “significantly enhanced the reliability” of aerosol classification and quantification should be supported quantitatively or rephrased.

The conclusion that the retrieved particle size decreases with altitude requires stronger evidence.

The attribution of the February aerosol mixture to residential wood-burning emissions is unsupported by the observations shown and should be removed.

The paper states that the findings improve understanding of “radiative properties” and aerosol–radiation–cloud interactions, but no radiative forcing or aerosol–cloud interaction analysis is performed. These broader implications can be mentioned as motivation or future relevance, but not as demonstrated outcomes of this study.
Citation: https://doi.org/10.5194/egusphere-2026-1060-RC2

Michal Andrzej Posyniak, Stefan Sitarek, Johannes Speidel, Tadeusz Stacewicz, and Hannes Vogelmann

Viewed

Total article views: 189 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
73	98	18	189	12	12

HTML: 73
PDF: 98
XML: 18
Total: 189
BibTeX: 12
EndNote: 12

Views and downloads (calculated since 24 Jun 2026)

Month	HTML	PDF	XML	Total
Jul 2026	31	34	6	71
Aug 2026	26	45	2	73
Sep 2026	16	19	10	45

Cumulative views and downloads (calculated since 24 Jun 2026)

Month	HTML	PDF	XML	Total
Jul 2026	31	34	6	71
Aug 2026	26	45	2	73
Sep 2026	16	19	10	45

Viewed (geographical distribution)

Total article views: 159 (including HTML, PDF, and XML) Thereof 159 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 18 Sep 2026

Short summary

Saharan dust from North Africa regularly reaches Central Europe and affects air quality and climate. We studied three major dust events over the Bavarian Alps in 2024 using laser-based atmospheric measurements and sun observations. Our analysis reveals how high the dust layers reached and how large the particles were. The results improve our understanding of long-distance dust transport and help refine atmospheric monitoring and climate models in Europe.


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