*3.3. Influence of Cloud Microphysics at PUY*

The air mass history can influence solute concentration by scavenging aerosol particles and gaseous species (as discussed in Section 3.2). This strongly depends on the CCN concentration related to the physicochemical properties of aerosol particles (size distribution and chemical composition) and on the gas phase chemical composition and corresponding phase equilibria. Microphysical cloud conditions such as liquid water content (LWC) and effective droplets radius (re) can also perturb solution concentration variability, as well as chemical reactions occurring within cloud waters. This section is devoted to the possible relationships between LWC and chemical variables. *Atmosphere* **2020**, *11*, x FOR PEER REVIEW 14 of 21conditions such as liquid water content (LWC) and effective droplets radius (re) can also perturb solution concentration variability, as well as chemical reactions occurring within cloud waters. This section is devoted to the possible relationships between LWC and chemical variables.

For this, a PLS analysis was performed with the LWC and the re, as the matrix of the explanatory variables (the "Xs"), and the chemical matrix (the "Ys"). The PLS chart is presented in Figure 8, and Table S6 reports the correlation matrix between these variables. There are weak correlations between LWC and ion concentrations. The strongest anti-correlation is between NH<sup>4</sup> <sup>+</sup> and re, i.e., r(NH<sup>4</sup> <sup>+</sup>, re) <sup>=</sup> <sup>−</sup>0.37. This analysis clearly demonstrates that at PUY, microphysical properties of the sampled clouds are almost not correlated with their chemical composition. This could be explained by the type of clouds that are collected; the majority are frontal clouds which were formed well before their arrival at the top of the mountain and which present a low variability in their microphysical properties. For this, a PLS analysis was performed with the LWC and the re, as the matrix of the explanatory variables (the "Xs"), and the chemical matrix (the "Ys"). The PLS chart is presented in Figure 8, and Table S6 reports the correlation matrix between these variables. There are weak correlations between LWC and ion concentrations. The strongest anti-correlation is between NH<sup>4</sup> <sup>+</sup> and re, i.e., r(NH<sup>4</sup> + ,re) = −0.37. This analysis clearly demonstrates that at PUY, microphysical properties of the sampled clouds are almost not correlated with their chemical composition. This could be explained by the type of clouds that are collected; the majority are frontal clouds which were formed well before their arrival at the top of the mountain and which present a low variability in their microphysical properties.

**Figure 8.** Chemistry/Microphysics PLS chart of cloud samples with t component on axes t1 and t2. Correlations map allows superimposing the "Xs" and the "Ys" (observations are removed for clarity). PLS performed on 73 cloud events. The dependent variables from the chemical matrix are displayed in blue, the explanatory variables from the microphysics matrix in red. **Figure 8.** Chemistry/Microphysics PLS chart of cloud samples with t component on axes t1 and t2. Correlations map allows superimposing the "Xs" and the "Ys" (observations are removed for clarity). PLS performed on 73 cloud events. The dependent variables from the chemical matrix are displayed in blue, the explanatory variables from the microphysics matrix in red.

A supplementary analysis (PCA) has also been performed (Table S7), demonstrating that correlations between microphysical variables and air mass history parameters (Section 3.2) are negligible. Thus, the influence of air mass history on the chemical composition of clouds cannot be attributed to the variability of microphysical parameters. In other words, there is an influence of microphysics, but it is statistically identical, whatever the zones or the sectors crossed by the air mass. A supplementary analysis (PCA) has also been performed (Table S7), demonstrating that correlations between microphysical variables and air mass history parameters (Section 3.2) are negligible. Thus, the influence of air mass history on the chemical composition of clouds cannot be attributed to the variability of microphysical parameters. In other words, there is an influence of microphysics, but it is statistically identical, whatever the zones or the sectors crossed by the air mass.

To remove any influence of LWC variation, cloud water loadings (CWLs) are commonly calculated to evaluate the solute content per volume of air. The statistical analysis, in this study, were conducted on solute concentrations in cloud water to get more robust results, because microphysical parameters are not always available, especially under winter conditions. However, LWC and r<sup>e</sup> variations are not highly variable (Figure S2) at PUY, and CWL patterns resemble those of solute concentrations in cloud waters (Figure S3). This suggests that for clouds sampled at PUY, the air mass history can better explain the variability of cloud water solute concentrations than LWC variations. To remove any influence of LWC variation, cloud water loadings (CWLs) are commonly calculated to evaluate the solute content per volume of air. The statistical analysis, in this study, were conducted on solute concentrations in cloud water to get more robust results, because microphysical parameters are not always available, especially under winter conditions. However, LWC and r<sup>e</sup> variations are not highly variable (Figure S2) at PUY, and CWL patterns resemble those of solute concentrations in cloud waters (Figure S3). This suggests that for clouds sampled at PUY, the air mass history can better explain the variability of cloud water solute concentrations than LWC variations.

A decrease of the solute concentrations from continental origin (NO<sup>3</sup> − , SO<sup>4</sup> 2− , and NH<sup>4</sup> + ) was observed between the periods 2001–2011 and 2012–2018, as mentioned in Section 3.1.3. In addition, a low decrease of its mean value (from 0.31 to 0.27 g·m−3) suggests that the CWLs for these species also A decrease of the solute concentrations from continental origin (NO<sup>3</sup> <sup>−</sup>, SO<sup>4</sup> <sup>2</sup>−, and NH<sup>4</sup> <sup>+</sup>) was observed between the periods 2001–2011 and 2012–2018, as mentioned in Section 3.1.3. In addition, a low decrease of its mean value (from 0.31 to 0.27 g·m−<sup>3</sup> ) suggests that the CWLs for these species also

significantly decreased. This trend could be explained by the aerological evolution highlighted by the

CAT model and requires further investigation.

Previous field studies have investigated the dependency of cloud chemical composition with

significantly decreased. This trend could be explained by the aerological evolution highlighted by the CAT model and requires further investigation.

Previous field studies have investigated the dependency of cloud chemical composition with microphysical parameters [16,80–82]. It has been shown, for sites more exposed to anthropogenic emissions, that LWC could modulate solute concentrations. For example, clouds freshly formed by orography can have their chemical composition modulated by cloud microphysics [7,19,73]. Anthropic mean cloud water concentrations at PUY are notably low ([NH<sup>4</sup> <sup>+</sup>] = 91 µM, [NO<sup>3</sup> −] = 70 µM, and [SO<sup>4</sup> <sup>2</sup>−] = 27 µM, see Table S2) as compared with the literature data [19]. Polluted events are exceptionally observed at PUY and most likely originate from afar (northeast France, several hundred kilometers away). If we consider the synoptic scale, we should see higher concentrations of NH<sup>4</sup> +, NO<sup>3</sup> <sup>−</sup>, and SO<sup>4</sup> <sup>2</sup>−; however, these ions are involved in chemical and photochemical reactions [83–86]. Hence, more information on the chemical aging of the air masses is needed (chemical characterization in progress). Moreover, such a long-term monitoring, with varied air masses, smooths the microphysics (LWC and re) influence (Figure S2). Cloud water is a complex matrix resulting from the interaction of many factors. Nevertheless, it appears that the air mass history, despite reduced correlations, remains the prevailing parameter, with either western and oceanic clouds or northeastern anthropogenic clouds.

### **4. Conclusions**

In this study, statistical analyses (AHC and PCA) were carried out on 208 cloud samples collected at the Puy de Dôme station (France) between 2001 and 2018, which resulted in clustering the cloud samples according to their chemical properties (concentrations of inorganic ions from marine and continental origins) into four categories as follows: "highly marine", "marine", "continental", and "polluted". Despite an evolution of the statistical treatment to classify the clouds samples, this work confirms those established in a previous study by Deguillaume et al. [46] for clouds sampled between 2001 and 2011. A change between the relative proportions of categories is however noticed and attributed to a significant decrease in the NH<sup>4</sup> <sup>+</sup>, NO<sup>3</sup> <sup>−</sup>, and SO<sup>4</sup> <sup>2</sup><sup>−</sup> concentrations during the second period (2012–2018) of cloud sampling.

CAT models the history of the air masses arriving at PUY, providing for each air mass the time spent above the eight cardinal sectors and above continental or sea surfaces. The CAT model specifies whether the air mass is in a free troposphere or in an atmospheric boundary layer. From these in silico zone and sector matrices and in situ chemical characteristics, PLS analysis highlights two main relationships between air mass origins and ion concentrations. A type of air mass comes predominantly from western sectors and from the "sea surface" (> ABLH) zone, with the highest concentrations of sea salts (Cl−, Mg2+, and Na+). A total of 31 cloud samples are gathered in the "highly marine" AHC category, which are characteristic of this air mass. Slightly linked to the latter, the "marine" AHC category, which is named for its air mass history and its low ion concentrations, is the most important (113 cloud samples) and the most "homogeneous". The second main air mass type arrives from the northeast sector and from the "continental surface" (> ABLH) zone, with the highest concentrations of potentially anthropogenic ions (NH<sup>4</sup> <sup>+</sup>, NO<sup>3</sup> <sup>−</sup>, and SO<sup>4</sup> <sup>2</sup>−). Only nine cloud samples are grouped in the "polluted" AHC category, characteristic of this air mass. With less extreme values and 55 cloud samples, the "continental" category represents the body of this set.

Finally, the influence of cloud microphysical properties (LWC and re) on the cloud water composition is investigated using PLS analysis in a similar way. This indicates no robust statistical correlations between cloud microphysics and cloud water chemical composition. This suggests that cloud chemical composition at PUY is influenced by air mass history which includes several physicochemical processes (CCN physical and chemical processes, mass transfer of soluble species, multiphase reactivity, etc.).

Clearly, this study highlights parameters that could drive the chemical composition of clouds at PUY; this statement cannot be generalized to other observation sites presenting different environmental scenarios. However, in a remote site, it appears that without major and immediate urban or marine influence, an air mass coming from the ocean or from a polluted area would be observed more or less loaded, according to complex biophysicochemical processes. In addition, much of the oceanic influence (i.e., Cl−, Mg2+, and Na<sup>+</sup> concentrations) seems to decrease quickly (78% of the clouds coming from the ocean appear "cleaned"), and much of the anthropic influence seems more persistent (NH<sup>4</sup> <sup>+</sup>, NO3, and SO<sup>4</sup> <sup>2</sup><sup>−</sup> concentrations) which remain significant.

The PUY site is a reference European station for the study of gases, aerosols, and clouds. International field campaigns have been conducted there in the past and future campaigns would especially target cloud biophysicochemical processes. Cloud waters collected at PUY for various air mass histories also serve for laboratory investigations that consider the following: (1) characterizing the complex chemical composition and its environmental variability by innovative analytical methods, and (2) quantifying photochemical and biological transformations occurring in this complex liquid medium. Cloud field investigations performed at PUY also help to build relevant chemical scenarios that help to better constrain cloud chemistry models [87]. For the dynamical frame, the CAT model makes it possible to give an overview of the air mass history; this helps to constrain cloud chemistry models but also makes it possible to compare the PUY station to other observatories where cloud studies are conducted.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4433/11/7/732/s1, Figure S1: Category distribution within each zone (a) and sector (b), Figure S2: LWC distribution at PUY, Figure S3: Comparison of normalized concentrations and normalized CWLs, Table S1: PuyCloud data ("TableS1.xlsm"), Table S2: Ion concentrations of the categories; Table S3: Spearman chemical correlation matrix of PCA, Table S4: Squared cosines of the variables, Table S5: Squared cosines of the variables (sectors, zones, and chemistry), Table S6: PLS correlation matrix between microphysics and chemistry, Table S7: PCA correlation matrix between microphysics and parameters related to air mass history.

**Author Contributions:** Conceptualization L.D. and P.R.; statistical analysis P.R.; dynamical analysis J.-L.B. and P.R.; writing—original draft preparation P.R. and L.D.; writing—review and editing, P.R., L.D., A.B., J.-L.B., A.-M.D. and M.B. All authors agreed with the submission of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This work on the long-term analysis of the cloud water chemical composition was supported by the French Ministry and CNRS-INSU. Authors acknowledge additional financial support from the Observatoire de Physique du globe de Clermont-Ferrand (OPGC), from the Regional Council of Auvergne, and from the Fédération de Recherche en Environnement through the CPER Environnement founded by Région Auvergne−Rhône-Alpes, the French ministry, and the FEDER from the European community. This work was supported by CEA/CNRS contracts. The authors thank Jean-Marc Pichon and Mickaël Ribeiro for technical support and discussions.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
