the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Drivers of ice-nucleating particle variability at urban and high-mountain environments of the Western Mediterranean
Abstract. This study presents a phenomenological analysis of immersion freezing ice-nucleating particles (INPs) at two different environments, urban (UGR) and high-mountain (SNS) stations, in Southern Spain. Measurements were conducted from June 2024 until June 2026, including INP concentration spectra and aerosol properties. The INP concentrations at UGR and SNS were consistent with previous measurements in the Mediterranean, showing a seasonal pattern. Spring shows the highest INP concentration values at high temperatures, which is likely related to the presence of biological particles. In contrast, during summer and autumn INP concentration values were higher at lower temperatures, which can be linked to the presence of Saharan dust. A correlation analysis between INP concentrations and aerosol optical properties (AOPs) revealed that at UGR there is negative correlation with the single scattering albedo (SSA), whereas for SNS there is positive correlation between INP concentrations and aerosol load proxies, such as the scattering and absorption coefficients. After calculating permutation feature importances between the INP concentrations and the AOPs, we developed parameterizations to predict INP concentrations. At UGR the parameterization is based on SSA, whereas at SNS the parameterization is based on aerosol scattering coefficient. For both parameterizations, 97 % percent of the data fall within a factor 10, and around 90 % of the data is within a factor 5, which reveals a good agreement between predicted and observed INP concentrations. These prediction schemes based on AOPs open new possibilities to extend INP estimations to observatories that routinely measure AOPs, extending the spatial and temporal coverage of INPs.
- Preprint
(17460 KB) - Metadata XML
-
Supplement
(10367 KB) - BibTeX
- EndNote
Status: open (until 26 Nov 2026)
- RC1: 'Comment on egusphere-2026-6200', Anonymous Referee #1, 29 Oct 2026 reply
-
RC2: 'Comment on egusphere-2026-6200', Anonymous Referee #2, 05 Nov 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-6200/egusphere-2026-6200-RC2-supplement.pdf
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 364 | 129 | 34 | 527 | 86 | 35 | 38 |
- HTML: 364
- PDF: 129
- XML: 34
- Total: 527
- Supplement: 86
- BibTeX: 35
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “Drivers of ice-nucleating particle variability at urban and high-mountain environments of the Western Mediterranean” by Elena Bazo et al. (2026), submitted to EGUsphere.
This paper by Elena Bazo et al. reports the relationship between optical properties and INP analysis through a long-term field campaign in two different environments (urban and high mountain) of the Western Mediterranean. The authors present an impressive off-line data set of immersion freezing INP analysis for a long-term period (two years), which, in my opinion, is worth publishing by itself. The idea of using scattering and absorption coefficients as INP proxies, by analogy with CCN studies, is novel (to my knowledge, as the authors state) and potentially useful. However, there are four remarkable points to improve before publication, specifically regarding experimental details which could be included in the text to improve data interpretation. Additionally, I found some conflicting statements that, in my opinion, should be double-checked before publishing, as outlined below.
Major comments:
1) Complementary information could help the reader to better understand the limitations and weirdly observed behaviors in the experimental data. There is a clear variation in INP concentration and ns in the first part of the campaign, which is, in my view, more related to the sampler changes than to the sample dilutions mentioned in lines 215f. I recommend adding a table with columns for the number of samples, size cut, type of filter and sampler, typical sample time and volume, methodological details (like dilutions or more details described in Bazo et al. (2026), https://doi.org/10.5194/amt-19-2695-2026), and typical INP concentration for each season (including the number of days with simultaneous INP, APS, σsp and σap data would be awesome). In addition, should the LOD reported in Fig. 3 be stated as a function of the sampling method? The LOD for GRAINS depends directly on sample loading and particle size; I personally think that the LOD should change between 2024 and the rest of the campaign.
Following the same thought, more information about the background considered for the INP analysis should be discussed in the text. The authors report FFs only for 2025 and 2026 blanks in Fig. S3 for both sites but, as I understand from lines 134f, the INP concentration was corrected using the FF of ultrapure water. The authors should discuss (i) the lack of blanks for the 2024 sampling period, (ii) the clear variation in the T50 among SNS blanks in comparison with UGR blanks, and (iii) how these blanks are included and how they could be a limitation for the INP results reported in this work (again, the changes in LOD should be considered). Additionally, it would be awesome to have some average FFs for real samples by season.
2) In my view, there is an important unsupported statement throughout the text and conclusions. The authors acknowledge at lines 186f that “the results of the model are clearly station-dependent”, and later (lines 405–408, and lines 433–436) they establish that optical properties could be used to estimate INP concentrations in similar environments. Although the prediction schemes (based on AOPs) presented in this work are solid, at least for the SNS site, it is necessary to present the previous statement as a hypothesis, not as a result, since no third site and no independent period are tested. Moreover, for future tests, it is necessary to include more details about the optical instrumentation conditions and discuss them properly, like RH (as ACTRIS recommends an RH below 40 % for in-situ optical measurements), detection limits and propagated uncertainties for SSA, AAE and SAE, and show how the AE33 and AE36s records were harmonized (considering that some values are low and noisy, especially in winter).
3) It is necessary to perform statistical tests to support the seasonal differences. As described in major comment 1, low-temperature data from 2024 are not directly comparable with the rest of the campaign. The authors report a seasonal contrast, shown in Fig. 5, in which data apparently mix years and methods. Clarify what data are used in Figs. 5, S6 and S7 (again with the help of a detailed sampling table, see above), discuss methodological variations in the text, and add a statistical test (e.g. Kruskal–Wallis with post-hoc Dunn) that should support or dismiss the mentioned variation. Independently of the statistical analysis, interannual variability should be considered in the seasonal variation discussion, since each season had at most two values (from only two years).
Following the same idea of intercomparison, I think that the urban/high-mountain design is not completely exploited. Maybe a paired analysis (same-day UGR versus SNS for INP and ns, or the SNS/UGR ratio as a function of PBL height, mentioning upslope transport and dust events) would help support the principal ideas described in the conclusions (lines 419f) and, somehow, the seasonal contrast more clearly.
4) The source attribution should be speculative since no direct measurements were made in this study (no heat treatment, no pollen counts, no dust-event identification and no back-trajectories are used). To turn this into a supported statement I suggest (i) comparing INP spectra (and ns) on identified dust versus non-dust days at each site (Bazo et al., 2025; https://doi.org/10.5194/acp-25-6325-2025), (ii) relating INP at −10 to −15 °C to daily pollen counts (Cariñanos et al., 2025; https://doi.org/10.1016/J.ATMOSENV.2024.120917), and (iii) complementing with HYSPLIT back-trajectories (https://www.ready.noaa.gov/hypub-bin/trajtype.pl). Following the same idea, the autumn SNS hypothesis of “a different type of biological particles” (lines 263–265) cannot be tested with the available data and should be marked as speculation or removed. Statements such as “most likely caused by pollen or subpollen particles” (lines 240–242) and “related to biogenic emissions” (lines 417f) should be softened accordingly. The mixing of dust with BC invoked for the lower autumn INP at UGR (lines 245–250) is likewise not supported by any data shown.
Specific comments:
1) Abstract, lines 14-16: Some statements are stronger than the results (see major comments 1 to 4). Statements should improve accordingly.
2) lines 118–119: The measurement period is given as 10 June 2024 to 7 June 2026. The cronograms (Figs. S1–S2) and Fig. 3 suggest that the first INP samples at UGR date from about late July 2024 and that some series end in May 2026. Please check and give the exact start and end of each station’s INP series (see major comment 1).
3) line 124: “Field blanks were taken approximately every month”: Fig. S3 shows blanks only from March/May 2025 onwards (see major comment 1). Please state which blanks exist for 2024 and which filter type they refer to.
4) Fig. 2: the literature data are plotted at their own measured temperatures and with different methods; consider a clearer legend and a note about the INP-sampling methods.
5) lines 205–209: The comparisons with Beijing, Mexico City, Mount Helmos and the Rocky Mountains are qualitative. Please provide a quantitative comparison (e.g. median INP at fixed temperatures such as −10, −15 and −20 °C in a table) and note methodological differences.
6) lines 217–219: The seasonal description at UGR (“lower INP during winter and higher during spring and summer”) should be supported by statistical tests (see major comment 3), and it appears to be at odds with Fig. S6, in which UGR winter medians are close to those of the other seasons over most of the range. For SNS the description (increase towards the end of summer, decrease during autumn) is not obviously consistent with the later statement that INP at SNS is higher in spring, summer and autumn than in winter (lines 257–258). Please harmonize.
7) lines 230–235: The data-coverage criteria for the boxplots (30 % and 50 % of the 46 expected samples per season) are reasonable, but the actual numbers of samples per season and temperature bin should be given, especially for the low-temperature bins that are only available from 2025 onwards and for winter at SNS.
8) Fig. 5: please add the number of samples per box (or per season).
9) Fig. 6: the labelled categories overlap with the data; you could show the median INP per cell or try to move the labels to a clearer space in the cell.
10) lines 403–408: The paragraph on applicability to other stations should be shortened and turned into an outlook (see major comment 2).
11) lines 414–420: The conclusion that clear seasonal variability was observed “at both sites, with higher INP in spring at warm temperatures” is inconsistent with the SNS description in Section 4.1 (higher INP in spring, summer and autumn than in winter; maximum at the end of summer) and with Fig. S6. “INP concentrations systematically higher at UGR than at SNS” should be tested statistically, since Section 4.1 states that the two stations are similar at the temperature extremes (see major comment 3).
12) Fig. S3 should show the blank spectra together with the sample spectra (or the blank-equivalent INP concentration), so that readers can judge the background directly.
13) Figs. S1–S2 would benefit from showing the number of samples per month (or from adding the table suggested in major comment 1).
Technical corrections:
1) line 13 (abstract): “97 % percent” → “97 %”.
2) line 86: “advention” → “advection”.
3) line 225: “similar value as those” → “similar values to those”.
4) line 353: “can be visualize also” → “can also be visualized”.
5) Eq. 7 and text: σscat versus σsp is used interchangeably; use a single notation throughout.
6) Fig. 4 caption: “(3.2)” → “(Section 3.2)”.
7) Fig. 6 versus Fig. S8: “hourly” and “daily” averages.
8) Supplement title page: “Lucas Alados, Arboledas” → “Lucas Alados-Arboledas”.
9) Table S1: the stray superscript in “Mt. Helmos (HAC)²”; check the units/format of coordinates across Tables S1 and S2 (some give elevation, some do not, and Table S2 mixes “a.g.l.” and “m” without a stated reference).