Major element concentrations as well as trace element concentrations of rainwater analyzed for our study compare quite well with previously published data for the same studied sites [16,17]. The data therefore confirm that the chemical composition of these rainwater samples departs from seawater chemical composition (Fig. 2), implying the contribution to these rainwater samples of continental components of natural and/or human origin. For instance, the high amounts of NO3−, SO42− and NH4+ in Alsatian rainwater, as well as the high SO42−/Cl or NO3−/Cl ratios (Fig. 2) result from the incorporation into and/or reaction with rainwater of nitrogenous and sulfur compounds of industrial and automobile origin [17,32]. Similarly, the high trace metal enrichments measured in these samples especially for Cu, Zn and Cd (Table 3) can be related to the heavily industrialized character of western Europe, close to the observations for rainwater of industrialized zones of northeastern America [35]. These metals (Cu, Zn, Cd) were also reported in excess (relative to Al) in lichens from the Vosges Mountains [10]. Combined to Pb isotope data, an anthropogenic origin was suggested for these metals measured in lichens. The second information that our data confirm is that the different study sites have quite contrasted chemical characteristics. Thus, pH of rainwater is higher in Esch-sur-Alzette than in the two other sites, and the dominant cations and anions are different. In the Esch-sur-Alzette rainwater, the dominant cations and anions are Ca²⁺ and HCO32−, respectively, whereas NH4+ and, at a lower degree, Ca²⁺ are the dominant cations in Aubure and Strasbourg rainwater, and SO42− and NO3− the dominant anions. The origin of the high level of Ca²⁺ and HCO3− in Esch-sur-Alzette rainwater was discussed by Herckes [16]. It could result from the input of carbonates particles into rainwater from neighbouring industries, i.e. iron and steel works, cement works. Such carbonate inputs would also explain the relatively high concentrations of this rainwater in Sr and Ba [16]. Finally, the third information brought by our data is a confirmation that the U concentration of rainwater is still very low, much lower than the U concentration found in river water of studied geographical area [13,29], with concentration around a ppt or a few ppt in rainwater of the three sites (Table 2).

The Sr and U isotopic data confirm that rainwater from the three studied sites have geochemical characteristics different from seawater and different from one site to another, as illustrated in the plot of (²³⁴U/²³⁸U) activity ratio against ⁸⁷Sr/⁸⁶Sr isotope ratio, where the three sites define three very different variation domains (Fig. 3). The Aubure rainwater exhibits variable Sr isotope ratios (between 0.71080 and 0.71459) systematically higher than the seawater ratio, and quite constant U activity ratios slightly superior to seawater. By contrast, rainwater from the two other sites (Strasbourg and Esch-sur-Alzette) have much less variable Sr isotope ratios with values inferior to seawater and quite variable U activity ratios with USr variation trends different for the two sites. These results certainly outline the important role of local components to explain the geochemical characteristics of rainwater at a regional scale. The origin of the variation for each site is discussed below.

The previous study by Herckes [16] proposed that the Esch-sur-Alzette rainwater is marked by the dissolution of carbonate particles, a large part of them could be emitted by local industry such as cement works. The chemical data obtained in our work for the Esch-sur-Alzette rainwater is consistent with carbonate aerosol dissolution. The U and Sr isotope data can also be interpreted in this way. Indeed, the Sr isotope ratios of the three analysed rainwater are inferior to seawater value, which is consistent with an input into rainwater of Sr from dissolution of carbonate dusts coming from old marine sediments, as used by local industry (iron and steel works, cement works). Furthermore, such carbonate dust dissolution should not fractionate the ²³⁴U and ²³⁸U isotopes and will therefore induce a concomitant U flux with no or only very limited U disequilibrium – (²³⁴U/²³⁸U) activity ratios close to 1. This explains therefore the U activity ratios of the Esch-sur-Alzette rainwater lower than seawater value and also the USr covariation that seems to appear in the USr diagram (Fig. 3). Interpretation of this covariation in terms of mixing between an end-member with seawater U and Sr isotope ratio characteristics and a carbonate end-member close to carbonates found in the Paris Basin allows us to infer that around 10% of Sr and between 30 to 60% of U of these waters would be controlled by dissolution of the carbonate dusts (insert in Fig. 3).

The Sr isotope ratios of the Aubure rainwater are characterized by radiogenic values higher than the seawater value. These rainwater data, collected in 1998 and 1999, are in the range of the Sr data published by Aubert et al. [3] for a similar collection period (1997–1998), but at a higher altitude. On the other hand, they are higher than the Sr data reported by Probst et al. [27] for rainwater collected between 1989 and 1992 (Fig. 4a). All these samples have nevertheless Sr values superior to the seawater value, which indicates the contribution to the Aubure rainwater of a radiogenic Sr source. The simplest explanation would be the contribution of a Sr input of dusts from local granitic bedrocks and soils that are radiogenic (e.g., [3]). Nevertheless, the lack of a simple relationship between the Sr isotope ratios and the rainwater Al concentrations or the Al/Sr ratios (which, in first approximation, can be used as an index of dust concentrations) questions this first interpretation. Moreover, the high Sr isotope ratios of our dataset is associated with quite high Ca/Sr and K/Sr ratios, which are different from the ratios found in the Aubure soils and granite, but much closer to the value classically found in plants and throughfall, as well illustrated in Fig. 4a and b. Such observations are certainly strong arguments to consider radiogenic Sr in Aubure rainwater as a result of recycling of local Sr through plants by a mechanism that remains to be defined, but that could certainly be plant exudation. The fact that for our samples the highest Sr isotope ratios are found in the rainwater samples collected in summer and fall reinforce this point. The observation that the rainwater data obtained by Probst et al. [27] for samples collected between 1989 and 1992, that is at a time when the trees were much less developed at the top of the site than today, have less radiogenic Sr values (Fig. 4a), agrees also with such an interpretation. At least, variations of Ca isotope data from Aubure rainwater are also consistent with this scenario [33]. Nevertheless, when all Aubure rainwater samples are considered, it appears that the Sr from the Aubure rainwater cannot be only explained by a simple binary mixing between a seawater-like end-member and a ‘plant’ end-member, but that a third end-member with low Sr isotope value and high Ca/Sr ratios has to be involved (Fig. 4a and b). This third end-member has geochemical characteristics quite close to sedimentary carbonates, as observed in the Esch-sur-Alzette samples, and can therefore be assimilated to a Sr flux from sedimentary carbonate dusts, which must come from outside the granitic Aubure watershed. The Sr isotopic data of Aubure rainwater confirm therefore the multiple origins of chemical components carried by rainwater with a local component introduced into the rain through the vegetation, and two other end-members of more regional origin, which are external to the Aubure watershed.

Compared to the Sr isotope data, the U activity ratios of Aubure rainwater is constant, within the analytical precision of 2 to 4%, implying either a unique U flux into the rainwater or several fluxes with very similar signatures. On the basis of our study, it is difficult to choose between these two possibilities; nevertheless, the fact that U concentrations of the Aubure rainwater, even if very low, decrease when the Sr isotopic ratios or the Ca/Sr ratios increase would suggest that the plant contribution is very low in the U budget of Aubure rainwater. The U activity ratios of the Aubure rainwater would therefore correspond to a regional geochemical background, whose mean value could indicate that it is slightly different from a purely seawater origin.

Compared to Aubure rainwater, the Strasbourg precipitations are marked by much more variable (²³⁴U/²³⁸U) activity ratios and much less in ⁸⁷Sr/⁸⁶Sr isotope ratios, with U values between 1.1±0.03 and 1.4±0.03, and ⁸⁷Sr/⁸⁶Sr isotope ratios between 0.70860 and 0.70904 (Fig. 3). Similarly to the two other localities, the geochemical characteristics of the Strasbourg rainwater cannot be explained by the contribution of only one geochemical end-member. The scatter of the Strasbourg data points in the ⁸⁷Sr/⁸⁶Sr vs. Na/Sr or Ca/Sr diagrams implies the contribution of at least three different Sr sources to the Strasbourg rainwater. If we assume the contribution of a Sr seawater end-member in the Sr budget of this rainwater, as proposed by the works by Herckes [16] and Sanusi et al. [32], then two other end-members with lower Sr isotope ratios have to be invoked (Fig. 5a). From their Sr isotope ratios, these two end-members could correspond to an input of aerosol of carbonate origin. The U activity ratios suggest nevertheless that this aerosol contribution cannot be identified with dissolution of carbonate aerosols, since in this case we should expect, as observed in the Esch rainwater samples, a decrease in the U activity ratios with an increase in the carbonate dissolution. Such a covariation is not observed with our data. On the contrary, the samples with high Ba/U ratios are marked by high U activity ratios (Fig. 5b). Determination of the origin of this material is not obvious on the basis of the lone data. It would be tempting to consider these end-members (or one of them) as representing chemical fluxes linked to human activity (automobile emission, particles from local incinerators, etc.), especially because samples like EPS6 or EPS1 exhibit quite high enrichment factors in some metals (Cd, Mn) (Table 3). Nevertheless, we are conscious that the low number of data does not allow us, at this time, to give a clear answer to this point, and that a detail study of rainwater at the scale of the Alsace plain is now required to resolve this question. Such a study should also help to understand the processes accounting for U activity ratios in rainwater higher than in seawater.