| 07 Nov 2025
Cloud condensation nuclei phenomenology: predictions based on aerosol chemical and optical properties
Abstract. This study presents a comprehensive phenomenological analysis of cloud condensation nuclei (CCN) and aerosol properties — including activation properties, microphysical characteristics, chemical composition, and optical properties — across ten surface sites in different environments. Aerosol properties vary widely, reflecting the diverse environments, and controlling the CCN activation characteristics. Despite their critical role in aerosol–cloud interactions, CCN observations remain sparse and unevenly distributed, limiting global assessments of activation behavior. To address this gap, this study presents CCN predictive methods based on chemical composition combined with particle number size distribution (PNSD) data, and aerosol optical properties (AOPs). The chemical composition driven predictions are tested using three hygroscopicity schemes. All schemes overpredict the CCN concentrations (median relative bias; MRB=13–15 %), although the two composition-derived CCN concentrations are markedly better predictors than the fixed-κchem assumption (MRB=24 %). The AOPs-derived CCN prediction is based on two approaches: an extended empirical parameterization of Shen et al. (2019) (hereafter S2019) to 13 stations, which reduces bias from - 27 % to - 8 % and improves CCN agreement; and second, a random forest model that infers Twomey activation parameters (C and k) using both the S2019 variables and all the available AOPs. Including all AOPs reduces MRB from 19 % to 15 % and highlights the role of absorption in predicting CCN activation. These findings demonstrate that both chemical and optical measurements can provide a reasonable estimate of CCN concentrations when direct measurements are unavailable. These results enable retrospective analyses of long-term aerosol time series to investigate aerosol–cloud interactions.
Competing interests: One of the co-authors, Anna Gannet Hallar, is a member of the Editorial Board of Atmospheric Chemistry and Physics. The remaining authors declare that they have no competing interests.
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.
This manuscript evaluates several different cloud condensation nuclei (CCN) prediction methods, including a couple of new ones derived here, against long-term measurements conducted at 10 different continental sites. The topic of the conducted research is a very important one. Both the technical approach and scientific conclusions made from the data appear robust. Overall, the paper is very well written and properly organized. I recommend accepting the paper for publication after a few, relatively minor revisions.
Detailed comments
lines 327-329: It is mentioned that the 3 mountain sites considered here have low activated fraction (AF) compared to other high-mountain sites. Does the “other sites” refer to all other sites for which such information is available, or some sub-set of sites in earlier studies? Do the authors have some idea why AF is particularly low at these 3 sites compared with other sites?
lines 362-365: I am not able to follow the logic here. While the bimodalily of Dgeo distribution admittedly suggests a mixture of two sources with very different particle size characteristics influencing the site, how would this bimodality by itself tell anything about the hygroscopicity of particles from these two sources (even when combined with Drit and kappa distributions)?
line 431: Associated with the first statement on this line, I would add reference to both Fig. 4a and Table 1, so that the reader can easily confirm the stated fact.
lines 435-439 (and 474-479): The authors correctly point out that different particle size ranges covered by CCNC and ACSM probably influence the comparability between kappaChem and kappaCCN. They should bring up more explicitly the fact that kappaCCN is influenced mainly by the hygroscopicity of particles having sizes close to Drit, while kappaChem is determined by some sort of bulk or “mass-average) hygroscopicity of all particles measured by the ACSM. This implies simply that if particles close to Drit are less (more) hygroscopic than the larger particles making most of sub-micron mass, then kappaCCN is expected to smaller (larger) than kappaChem. The systematically lower kappaCCN at these sites might simply indicate that organic fraction of particles increases with decreasing particle size when approaching Drit. For the same reason outlined above, I think that it is irrelevant to explain differences between kappaChem and kappaCCN by whether Dcrit drops between the lower size cut of ACSM or not (lines 474-479): Dcrit at 0.4% supersaturation is anyhow well below that mass-mean diameter of particles measured by ACSM, so it is not expected that these 2 kappas are the same. There are, of course, many other things (as discussed in literature and to some extend also in section 4 here) that might affect the comparability between kappaChem and kappa CCN, but I feel that this “size-issue” should be mentioned already here.
Finally, the authors could explain, or mention, somewhere why all the calculations presented in the paper correspond to the supersaturation (SS) of 0.4%. This choice is fine, but in the literature also many other values of SS spanning from about 0.1 to 1 % have been reported.