1 Introduction

The stable isotopes of oxygen and hydrogen are useful tracers of the sources of water in aquifers (Epstein and Mayeda 1953; Mazor 2004). Indeed they have been used extensively as environmental tracers in the past decades to investigate water cycles, looking at residence time, recharge areas, climate variability (Kendall and McDonnell 1998; Schiavo et al. 2009; Longinelli and Selmo 2010). Among stable isotopes oxygen-18 (δ18O) and hydrogen-2 (δ2H) anomalies are useful to assess the recharge areas and water mixing (Nisi et al. 2016 with references). Therefore, it is worthwhile to carry out a oxygen-18 (δ18O) and hydrogen-2 (δ2H) groundwater study for the sites where the Environment Protection Agency of Campania Region (ARPAC) collects data for the chemical status evaluation of Groundwater Bodies. The aim is to improve the conceptual model of aquifers, according to the EU Water Framework Directive (EC 2000) and the Groundwater Directive (EC 2006). Both regulations require the identification of groundwater bodies (GWB), in order to carry out a long-term monitoring devoted to the classification of the chemical and quantitative status (Onorati et al. 2009). This goal is achieved following a stepwise approach, including geological characterization of aquifers, mapping of GWB, delineation of monitoring network, evaluation of relevant chemical substances, data collection, and classification of GWB (EC CIS for WFD 2009). After classification further research on the pollution sources is often needed. ARPAC in 2002 implemented the GWBs quality monitoring network aiming at the evaluation of the environmental status, measuring chemical substances (Adamo et al. 2007; Onorati et al. 2009). The monitoring started in autumn 2002 and is ongoing; it includes large springs and adduction wells, feeding drinking water aqueducts, as well as wells in alluvial plains (Onorati et al. 2009). The groundwater bodies were mapped according to literature (Celico et al. 2003 published in digital form by De Vita et al. 2018), taking only major aquifers into account (Fig. 1). As a whole the network consisted at the beginning of about 120 monitoring sites for 40 major regional GWBs. In 10 years, the number of the sampling points increased, reaching 183 points for 46 GWBs in 2012. Since 2012 widespread network revision started, aiming at monitoring further 34 GWBs. New sampling points have been added, increasing also the sampling points for GWB in bad chemical status, i.e., with average concentrations for the GWB above thresholds. In 2015, the Campania Region has been partitioned into 80 significant groundwater bodies (PGA 2016): 10% volcanic GWBs, 25% alluvial GWBs, 30% karst GWBs and 35% mixed, at low–moderate permeability. The new monitoring network consists of 320 sampling points, of which 200 active in 2015 (Ducci et al. 2019). However, most of the monitoring sites remained unchanged. Groundwater of these sites are sampled twice a year in order to calculate the annual chemical status classification, (D.lgs. 30/2009; D.M. Env. 6 July 2016). The sampling points guarantee the monitoring of the most important carbonate, volcanic and alluvial aquifers of Campania (Fig. 1), which are the principal drinking water sources, as well as areas affected by high environmental pressures. The carbonate aquifers show hydrogeological features which endanger groundwater resources during drought periods, while the alluvial aquifers represent peculiar pollution risk areas. During the classification of the GWBs and single monitoring point, variations in chemical composition occurred several times, which could not be explained looking only at chemical data. Therefore, a subset of sites for isotope analysis has been defined in order to evaluate recharge and mixing effects (Nisi et al. 2016; Schiavo et al. 2009). The GWB have been chosen also in order to take the geological and geographical variability of Campania Region into account. Since in this research no data on rainfall has been collected, the comparison between groundwater and rain isotope δ2H–δ18O is based on datasets and scientific papers. Rain water δs have been extensively studied in the literature (Craig 1961; Rozanski et al. 1993; Gat et al. 1996; Longinelli and Selmo 2003; Dotsika et al. 2010) in order to check the deviations from the MWL (Meteoric Water Line) aiming at the detection of recharge areas. Source areas and seasons of recharge have been detected in Apennine aquifers interpreting δ2H–δ18O data of surface water, groundwater and precipitation (Barbieri et al. 2005; Petitta et al. 2010; Mussi et al. 2015). In the last years, many papers faced the variability of δ2H–δ18O in precipitation along the Italian Peninsula and in Southern Italy, and defined rain depletion functions and maps (Paternoster et al. 2008; Sappa et al. 2012; Madonia et al. 2014; Giustini et al. 2016 and cited papers).

Fig. 1
figure 1

Groundwater bodies and chemical status monitoring network of Environment Protection Agency of Campania Region (from Ducci et al. 2019). Acronyms refer to the studied areas

Full size image

2 Data and methods

In order to evaluate recharge areas and mixing processes in the GWBs of the Campania Region (Italy), 52 monitoring sites, 48 wells and 4 springs (Table 1), located at elevations ranging from 0 to 1070 m a.s.l., have been measured in the period 2010–2016. On the whole, δ2H–δ18O isotopic compositions have been measured for 416 samples in the Benevento Lab of ARPAC using a Thermo Mass Spectrometer. Several instruments are presently used for stable isotopes (Muccio and Jackson 2009). In this research, a GC–IRMS with purge and trap has been used (see also Muccio and Jackson 2009 for description and Session 2006 for analytical methods). A thermo-proprietary software controls all elements of the GC–IRMS system, allowing for a coherent management of the GC and the IRMS components, with the processing of chromatographic and isotopic data. The analysis in the GC uses a purge-and-trap injection method. For water environmental samples, where crowded chromatograms are common, separation of chromatographic peaks is the most important factor limiting the acquisition of accurate isotopic data; anyway an adequate length of capillary column is useful to minimize this problem. Another crucial point is the interface between GC and IRMS which has several important functions: it must convert substances to the molecular form needed for measurement, it removes undesirable reaction products, and it hinders large pressure fluctuations. The IRMS unit includes electron-impact ionization sources, a magnetic-sector analyzer, multiple Faraday detectors affording a highly stable detection of ion beam currents. Standardization has been performed using standards developed in the laboratory. The method used by ARPAC for groundwater sample collection is described in the guidance for planning and undertaking groundwater sampling to survey the quality of groundwater, provided by the International Standard ISO 5667-11 (ISO 2009). The stable isotope ratios are expressed in δ units in parts per mile representing the deviations of the ratio between heavy and light isotopes of a sample with respect to the same ratio in a reference material, which for waters is the Vienna-Standard Mean Ocean Water (V-SMOW) international standard (Craig 1961; Rozanski et al. 1993). The maximum analytical uncertainty for δ18O is ± 0.62‰, with an average of 0.45‰, while for δ2H the maximum is ± 1.35‰, with an average of 0.90‰. These values have been taken into account in the dataset evaluation. The collected data concern 13 GWB (Fig. 1, Table 2), monitored at 52 sites (Table 1) of which 5 are karst aquifers (Lattari Mts., Durazzano Mts., Taburno Mt., Moschiaturo Mt., Camposauro Mt.) 5 alluvial plains (Benevento Plain, Volturno Regi Lagni Plain, Eastern Naples Plain, Sarno Plain, Isclero Plain) 3 volcanic apparatus (Ischia, Campi Flegrei, Vesuvius). The average elevations range between 31 m (Volturno Regi Lagni Plain) and 815 m (Taburno Mt.), average infiltration ranges between 51 and 377 mm. Rain stable isotopes reference data on δ18O and δ2H have been downloaded and processed from the Global Network of Isotopes in Precipitation (GNIP) (Terzer et al. 2013; IAEA/WMO 2018) created following a geostatistical approach (Bowen and Revenaugh 2003; Van der Veer et al. 2009; West et al. 2014). From the global gridded values dataset, the values of the δ18O and δ2H of the five geographical coordinates centered on Campania have been extracted, mapped and plotted (Fig. 3), they show very similar pattern. Moreover, looking at the GNIP dataset with the aid of wiser software (IAEA/WMO 2018) the availability of rain data for Campania Region has been checked. The only data available are those from sites in neighboring regions: Zannone, Fogliano, Subiaco in Latium, Vulture in Basilicata, Bari in Apulia, Cosenza in Calabria. These datasets, collected in 1998–2003 have been analyzed. For Campania the dataset by Madonia et al. (2014) concerning Mt. Vesuvius has been taken into account. The δ18O map by Giustini et al. (2016) shows for Campania values ranging between − 6‰ along the coast and − 10‰ in mountain areas near Apulia.

Table 1 Location of the isotopes monitoring sites in the groundwater bodies of Campania Region
Full size table
Table 2 Key hydrogeological features of the 13 groundwater bodies of Campania Region with isotope measurements
Full size table

3 Statistical analysis

In order to evaluate the influence of spatial position and time for many measured δ, statistical analysis is a useful tool. Data processing has been performed using the STATISTICA software ver. 7.1. The values measured in groundwater (Table 3) range between − 58.52 and − 6.30‰ for δ2H and between − 9.32 and – 1.32‰ for δ18O with a huge variability. The frequency distribution of data is on the whole normal with mean, mode and median with similar values and a low skewness, however the kurtosis shows the effect of outliers. The differences between δ18O and δ2H distributions (Fig. 2a, b) is clear looking at the frequency diagrams: the first is less dispersed, however both show a secondary group with high values. The secondary groups concern less than 5% of the global populations. The normalized standard deviation is the largest for Is1 and Bvr 29 and very low for many Volturno plain wells (Table 4). The dataset evidences the larger variability for δ2H with respect to δ18O as known in the literature (Craig 1961). Two sites: Is1 (Forio thermal well) and Bvr29 (near a coastal marsh) show peculiar values and pattern, with δ2H larger than − 20‰ and δ18O of about − 2‰. Therefore, these values have been further investigated in detail. The lowest values are those of Mt. Moschiaturo spring (mean 53.4‰ δ2H and 9.1‰ δ18O). For the other sites the δs are very stable. The groundwater data set has been compared with up-to-date rain data. The most relevant statistics concerns the correlation between δ2H and δ18O values and MWL curves (Figs. 4, 5, 6). In the graphs the MWL lines GMWL (Rozanski et al. 1993) EMWL (Gat et al. 1996) SIMWL (Giustini et al. 2016) Vesuvius MWL (Madonia et al. 2014) Pontina Plain MWL (Sappa et al. 2012) are plotted against Campania GWB monitoring sites data. The groundwater data taken into account are both the whole samples data set (Figs. 4, 5) and the averages for each site (Fig. 6, Table 4). The relationships between GWB elevations and δ2H–δ18O have been also investigated looking at the surface mean elevation of the GWB (Fig. 7a, b). For the sites with more than 10 samples the temporal variability has been checked, since time-dependent variations have been observed, average yearly δ have been also calculated (Fig. 8a, b). Furthermore for the GWB Monti Lattari, with many monitoring points a comparison among sites has been plotted showing local trends (Fig. 9). In order to compare the characterizing parameters of the sample points (Table 4), the cluster analysis is a useful tool for grouping the sites (Kendall 1975). Hence a dendrogram has been built-up (Fig. 10) choosing the Ward method (Ward 1963) with Euclidean distances, and the results have been mapped (Fig. 11).

Table 3 Descriptive statistics of the δ2H and δ18O values of the samples measured in the GWBs of Campania Region (416 analytical determinations, 52 points, 13 GWB) reported in ‰ against the V-SMOW standard
Full size table
Fig. 2
figure 2

a, b Frequency distribution of the 416 samples of δ18O‰ (a) and δ2H‰ (acronyms) (b) (concentrations vs. V-SMOW). The diagrams show the presence of outliers with higher values

Full size image
Table 4 Statistical parameters of δ2H and r δ18O reported in ‰ against the V-SMOW standard for each sampling point (lacking data for single values)
Full size table

4 Discussion

The most important factors influencing 18oxygen (δ18O) and 2hydrogen (δ2H) compositions are altitude and continental effect (Longinelli and Selmo 2003; Giustini et al. 2016). The latter, however, seems to be negligible for coastal regions like Campania, and significant only for GWBs located eastward of the Apennine axis. Moreover, temperature has also an effect on isotopes enrichment, as well as rainfall regime. Indeed seasonal variations of rainfall stable isotopes are large (Fig. 3a, b, and GNIP dataset extracted with wiser software IAEA/WMO 2018) and must be taken into account for detailed annual recharge studies. The comparison between groundwater samples and MWL lines (Figs. 4, 5) shows that the Campania line drawn from GWB data is crosscutting all the MWL lines, due to a group of samples with higher δ2H–δ18O. Indeed two sites show peculiar characteristics, with δ near to V-SMOW (Fig. 5, Table 4). For these sites, a contamination of sea water probably occurs since the sites are located at Ischia (Is1) and near Patria Lake (Bvr29) less than 1 km from the Tyrrhenian Sea. In order to check for sea water influence versus rain stable isotope fingerprint, reference rain values have been taken into account together with seawater of Mediterranean Sea (Bigg and Rohling 2000; Cox et al. 2011) (Table 5). The rainfall patterns suitable for this check are those of Vesuvius (Madonia et al. 2014), Zannone Island and the IAEA model (IAEA/WMO 2018) which are the datasets at the lowest distance from the Ischia and Patria lake sites and show also coherent rainfall δ. A simple mass balance evaluation has been performed with coherent results for δ2H–δ18O (Table 5), therefore the groundwater stable isotopes measured values of Is1 and Bvr29 should be due to water mixing, the proportion is in the order of 50% seawater and 50% rainfall, the best estimates seem those from IAEA Model and Zannone Island weighted averages. In the comparison of groundwater stable isotopes with rainfall and altitude, Is1 and Bv29 data have been hence discarded. After the withdrawal of samples from Is1 and Bvr29 outliers, the Campania groundwater line falls within the rainfall MWLs (Fig. 5), almost overlapping the SIMWL (Giustini et al. 2016) and the correlation, for a dataset encompassing several GWB with different elevation and geology, is high, with R2 above 0.75. This result shows that rainfall infiltration is the prevailing recharge process. Looking in further detail at average values of each site (Fig. 6), the δ2H–δ18O data show an even better correlation (R2 = 0.857), with the most negative values for Moschiaturo Mt. For this site values below − 50‰ δ2H and − 8.5‰ δ18O are the witness of an important influence of precipitation in higher recharge areas, with δ18O of − 8 to − 9 (Giustini et al. 2016). In Italy typically the altitude effect for precipitation ranges from − 0.1 to − 0.4‰ of δ18O/100 m (Giustini et al. 2016). For Southern Italy the effect is approximately of − 0.2‰ δ18O/100 m, Campania literature data are in the range of − 0.11‰ to − 0.27‰ near Molise (Longinelli and Selmo 2003, 2010), in Petitta et al. (2010) the reported gradient for an area lying 40 km N of Campania is − 0.19‰ δ18O/100 m, in Giustini et al. (2016) a review of gradients is reported, with altitude effect for precipitation ranging from − 0.1 to − 0.6‰ of δ18O/100 m. For Campania groundwater samples, the relationships of elevations versus δ2H and δ18O show, with respect to the cited rainfall data, lower gradients for δ18O: − 0.1‰/100 m (Fig. 7a), while for δ2H of − 1.0‰/100 m (Fig. 7b) the literature is poor due to weaker elevation/δ2H correlations. The low Campania groundwater δ18O versus elevation gradients should be due to the fact that most of the cited research on rainfall/elevation deal with the Apennine chain, while the GWBs of this study lie nearby the coast. The high correlations between δs and GWB average elevations is related to the prevailing recharge of aquifers from rainfall, while the Ischia outlier and the Bvr (Regi Lagni Volturno Basin) anomalous value testify the mixing of different waters: rainfall recharge water and sea water, as discussed above. For Campania the groundwater δ2H and δ18O data set is based on 2 to 4 samples each year, not sufficient for robust seasonal studies, so annual averages are considered; however, part of the variability of data is certainly related to seasonality, and the monthly variations are much larger than uncertainty (Figs. 3, 8a, b). Only a part of the monitoring sites (Table 2) has been measured for the whole period 2010–2016 and the results for Mt. Lattari GWB sites with long series are shown in Fig. 9. Indeed the “short” time series 2010–2016 are not adequate for trend analysis, however they show a coherent pattern, with low δ2H and δ18O values measured in 2013 and high values in 2015 (Fig. 8a); this pattern is even more evident looking at the data normalized with respect to average values (Fig. 8b). The annual pattern is also apparent looking at single time series (Fig. 9). Indeed rainfall in winter was above average in the Campania region in 2013, with lower δs as typical for this season as shown in Fig. 3. However, to explain the yearly variations of isotope values, further comparison between groundwater and climatologic data is needed. In order to check for spatial variations of isotope values, with cluster analysis three major groups of monitoring sites have been detected (Fig. 10): the groundwater influenced by seawater intrusion, the GWBs located in the alluvial plains and volcanic edifices (in red), including most of the wells which show rather constant values, probably due to the large recharge areas, smoothing local effects. The third group encompasses the points at higher elevation, as well as some sites which should be further investigated (in magenta) in order to check for contamination with less depleted waters. The spatial distribution of the classes is shown in a map (Fig. 11). The detection of outliers by means of cluster analysis and mapping proved to be a useful tool for the identification of sites with anomalous δ2H and δ18O, which needs further local investigations.

Fig. 3
figure 3

a, b Model estimation of monthly and yearly δ18O‰ (a) and δ2H‰ (acronyms) (b) (concentrations vs. V-SMOW) for the grid points of reported Lat and Long coordinates (values in geographical degrees) extracted from IAEA site (IEAEA/WMO 2013)

Full size image
Fig. 4
figure 4

The 5 MWL lines have been redrawn from cited datasets

Relationships between δ2H‰ and δ18O‰ (concentrations vs. V-SMOW) (52 sites, 416 samples) with ellipses are outlined the samples with marine water contamination (upper left corner of graph), and the sites at higher elevation and those with relevant winter recharge (lower right corner of graph).

Full size image
Fig. 5
figure 5

The 5 MWL lines have been redrawn from cited datasets

Relationships between δ2H‰ and δ18O‰ (concentrations vs. V-SMOW) for samples without relevant marine contamination (50 sites 396 samples) with regression equation and R2 for the Campania GWB line.

Full size image
Table 5 Mass balance for Is1 and Bvr29 groundwater
Full size table
Fig. 6
figure 6

The 5 MWL lines have been redrawn from cited datasets

Relationships between the mean of δ2H‰ and δ18O‰ (concentrations vs. V-SMOW) (50 relevant sites) not included the samples with marine water contamination, with regression equation and R2 for the Campania GWB line.

Full size image
Fig. 7
figure 7

a, b Comparison between mean δ2H‰ and δ18O‰ (concentrations vs. V-SMOW) and mean elevation of the groundwater bodies (13 GWBs). Ischia outlier (star) has not been included in the statistics, while Bvr has been taken into account although including a site with marine ingression

Full size image
Fig. 8
figure 8

a, b Comparison between mean δ2H‰ and δ18O‰ (concentrations vs. V-SMOW) yearly mean values: a mean values showing a negative anomaly in 2013, b normalized values with a marked 2013 negative anomaly for both δ2H‰ and δ18O‰

Full size image
Fig. 9
figure 9

δ18O‰ (concentrations vs. V-SMOW) 2010–2016 time series in the 5 sites of Lattari Mts. The lowest values are those of winter 2013

Full size image
Fig. 10
figure 10

Dendrogram created according the Ward method with Euclidean distances, data set is reported on Table 4. Missing cases have been discarded, 44 samples have been classified. With ellipses are outlined three groups: the samples with marine water contamination (a), those in alluvial plains and volcanic edifices (b), the sites at higher elevation or with relevant winter recharge (c)

Full size image
Fig. 11
figure 11

Geographic location of the classes of the dendrogram created according the Ward method with Euclidean distances. The blue dots with a × correspond to the samples with marine water contamination (Class A), in magenta with a\ are the sites at higher elevation and those with relevant winter recharge (Class C), the red dots refer to the other points, mostly in alluvial plains and volcanic edifices (Class B)

Full size image

5 Conclusions

The ARPAC δ2H–δ18O archive is a robust data set (416 samples, 52 sites, 13 GWB, period 2010–2016) filling a gap in reference knowledge of Campania Region stable isotopes hydrogeochemistry. In the first evaluation of the data set performed in this study, the pattern of δ2H and δ18O data set has been described, with the aid of statistics. Some peculiar sites with marine intrusion have been detected, whereas for the other coastal monitoring point no major saltwater intrusion has been observed. Springs and wells with recharge areas at high elevation (above 700 m a.s.l.) show a more “cold and continental” stable isotope finger tip. For most of the sites in the alluvial plains local variability is low. The comparison between groundwater δ2H–δ18O (excluding Is1 and Bvr29 seawater contaminated samples) and MWL redrawn from literature show that the samples, the sites and the GWB of Campania fall within de SIMWL (Giustini et al. 2016) with a very slight deviation toward δ18O (Fig. 6), hence direct rainfall recharge of aquifers is the prevailing mechanism. The vertical gradients for groundwater samples of δ2H–δ18O, fall within the values reported in literature for rainfall, however the 0.1‰/100 m δ18O gradient is low, anyway in most of the papers on this topic (see Giustini et al. 2016 for citations) the gradients are evaluated for small watersheds, whereas in this study the GWBs area stretches for about 100 × 100 km creating a smoothing effect. The cluster analysis proved to be a useful tool in detecting groups of sites with similar characteristics, associated with analogous groundwater recharge sources. This approach can be further applied to each GWB in detailed studies concerning single hydrostructures. A detailed comparison of the stable isotope measurements with meteorological and water quality data as well as geostatistics of data (Van der Veer et al. 2009) is the next step to improve the interpretation of GWB recharge in the Campania Region.