Environ Earth Sci DOI 10.1007/s12665-013-2367-2
ORIGINAL ARTICLE
Arsenic anomalies in shallow Venetian Plain (Northeast Italy) groundwater A. Carraro • P. Fabbri • A. Giaretta L. Peruzzo • F. Tateo • F. Tellini
•
Received: 23 October 2012 / Accepted: 28 February 2013 Springer-Verlag Berlin Heidelberg 2013
Abstract A pilot area within the Venetian Plain was selected to assess the arsenic (As) contamination of groundwater. The area represents a typical residential, industrial and agricultural organization representative of most western countries, and is also devoid of lithologies with high or anomalous As content. Hydrogeological and chemical data have been collected, the latter spatialized by a geostatistical approach. The unconfined aquifer reservoir varies from a predominantly gravel composition in the north to a sandy and silt–clay composition further south, including peat layers. The hydrochemical features of the waters are rather homogeneous, featuring low mineral content and a Ca-bicarbonate signature. In contrast, the redox state is highly variable; oxidizing conditions are predominant in the northern and coarse parts of the aquifer, whereas reducing potentials prevail in the southern and silt–clay parts. Several well waters contain arsenic in excess of drinkable limits (=10 ppb), and most of these wells are located in the southern area. A large portion of the studied area has a high probability of containing nonpotable water (up to 150 ppb As). Remarkably, As ‘‘hot spots’’ (As [ 300 ppb, up to 431 ppb) were identified at the transition from gravel to silt–clay sediments. No industrial or agricultural source of As has been found.
A. Carraro (&) P. Fabbri A. Giaretta L. Peruzzo F. Tateo F. Tellini Institute of Geosciences and Earth Resources, National Research Council (CNR) of Italy, c/o Department of Geosciences, University of Padova, 35131 Padova, Italy e-mail:
[email protected] P. Fabbri Department of Geosciences, University of Padova, 35131 Padova, Italy
Keywords Arsenic Contamination Geostatistics Hydrochemistry Venetian Plain (Italy)
Introduction Arsenic concentration in the natural environment has always been considered a risk factor for human disease. The occurrence of arsenic (As) in water directly designed for drinking, agricultural uses and livestock breeding is of particular interest. Numerous studies have evaluated As anomalies on Southeast Asiatic Plains (e.g., Bangladesh), which have been well known for many decades (Welch and Stollenwerk 2003; Ahuja 2008; Ravenscroft et al. 2009). In Italy, there have been studies of groundwater As concentrations, but these studies have mainly considered volcanic reservoirs (Angelone et al. 2009). The consequences of As occurrence in drinking water are very severe, and some diseases are directly related to the concentration of this element in water (Tseng 1977; Chen et al. 1988; Tseng et al. 1996; Chiou et al. 1997; Wang et al. 2002, 2009). Interest in As drinking water contamination has increased over time (Alerts and Khouri 2004) due to two main factors: (1) a widespread decrease in the drinkable water resources as a consequence of increasing pollution of anthropogenic origin; and (2) a continuous lowering of legally allowable As concentrations in the West due to the hazard posed by As. In Europe and the USA, an As concentration of 10 micrograms per liter (ppb) is currently considered acceptable for drinking. In Italy, the Po and Venetian Plains represent a continuous region where industrial activities are concentrated and where millions of people live and work. There are occasional reports of As contamination in groundwater in different parts of the plains (Scialoja 2005). The Venetian
123
Environ Earth Sci
Plain is a highly populated area. There are no localized emissions of hydrothermal and volcanic As, but a variety of industrial and farming activities involving a significant amount of water for food uses (directly for drinking, agriculture and animal nutrition) are present. In the past, As anomalies in water and groundwater have been identified only sporadically. The first signal was given by results of a survey that began in 1992 (Baldantoni and Ferronato 1995), which showed the presence of an area with As concentrations exceeding 50 ppb, ranging as high as 480 ppb. However, these data are related to six different superimposed aquifers, ranging from zero to approximately 300 m in depth over a fairly wide area (about 20 km 9 15 km), that includes the area considered in this work. The main goal of this study is to evaluate the distribution of As in shallow groundwater within a pilot area of 70 km2, which was selected according to the following features: (1) the study area shows a residential and industrial composition typical of most western countries; (2) a sampling survey was performed that took the hydrostratigraphical situation into account; (3) samples were collected in a narrow time interval; (4) wells were located in a welldefined stratigraphical context; and (5) known pollutant sources of As were not present.
Sile River, 5–6 m3/s). This configuration is a characteristic structure that is present throughout the northern Po Plain (Cucchi et al. 2008; Pilla et al. 2006). The unconfined aquifer is 200–300-m thick and is intensively exploited for water supply. The relatively homogeneous, unconfined hydrogeological unit changes into a multi-layered confined or semi-confined aquifer system south of the ‘‘fontanili’’ belt. Impervious or semipervious layers, which may be more than 10-m thick, are interbedded with sand and gravel, as shown in Fig. 1. Many public wells of different depths (100–300 m) were drilled in the southern part of the high plain and screened in a multilayered aquifer system for water supply purposes. The unconfined aquifer is mainly recharged from irrigation, surface water infiltration and rainfall. These processes are only effective in the high part of the Venetian Plains, where surface water infiltration can reach the unconfined aquifer and the artesian aquifers linked to it. The study area is located in the central part of the Venetian Plain, on the Brenta megafan (on the right bank of the actual Brenta River), including the municipal territories of Carmignano di Brenta, Grantorto, Gazzo and Piazzola sul Brenta (Northern Padova district, Northeast Italy; Fig. 1). Hydrostratigraphy
Study area Hydrogeological framework The Bacchiglione, Brenta, Piave and Leogra Rivers are responsible for the deposition of a significant portion of the material, hundreds of meters thick, which forms the subsoil of the Venetian Plain. Along the piedmont belt of the plain, fans from adjacent rivers laterally penetrate gravelly alluvial fans. The result is entirely gravelly subsoil throughout the thickness of the high Venetian Plain. Because deeper fans often invade further areas of the high plain from the undifferentiated gravel cover, the terminal parts of the fans extend downstream for various distances, producing an alluvial cover that is no longer uniformly gravely, but is instead composed by alternating layers of gravel and silty clay of swampy, lagoon or marine origin (Bondesan and Meneghel 2004; Fontana et al. 2008). The stratigraphical framework produces a large, unconfined aquifer extending 15–20 km south, from the Pre-Alps to the southeast. The water table, in a band that is 2–5 km wide and several kilometers long, intersects the topographic surface, creating numerous characteristic plain artesian springs referred to as ‘‘fontanili’’ (Fig. 1; Vorlicek et al. 2004). This natural drainage system supplies numerous perennial streamflows (e.g., the
123
About thirty lithostratigrafies from local and regional administrative technical authorities, were used in this study. The original stratigraphies were simplified to conform the data to stratigraphical reconstructions. The maximum depth of the lithostratigraphic investigation using private wells was approximately 200 m, and three-dimensional aquifer hydrostructures were constructed for six hydrostratigraphic profiles, both in the north–south and east–west directions (Fig. 1a). The hydrostratigraphical profiles are presented in Fig. 2. In Fig. 1, the position of two boreholes specifically drilled in this zone to a depth of 30 m (RB1 and PM1) is also shown along cross-section 1L. Note that the two boreholes demonstrate a more complex and variable lithostratigraphy with respect to the cross-sections reported in Fig. 2. In particular, the organic matter is concentrated in peat layers and scattered in clays, silts and sands, either as small fragments or as a thin and discontinuous lens. Most of the materials plotted as clay in the cross-sections are silt and clay with a minor sandy component (Fig. 3). Besides the hydrostructural aspects, particular attention was placed on the presence of peat deposits, considering their debated role in As concentration (see, for example, Yamazaki et al. 2003; Zaccone et al. 2008). There is a visible peat layer approximately 20–30 m below ground level (from -5
Environ Earth Sci
Fig. 1 Sketch of the study area. The inset map shows the location of the six cross-sections reported in Fig. 2 and the position of two boreholes. The upper limit of the ‘‘Fontanili belt’’ and the location of
the schematic geological profile are also shown (a); schematic geological profile of the Venetian alluvial plain from north to south (b)
to 5 m with respect to the sea level, Fig. 2). This layer is visible in cross-sections 1T, 2T, and 1L and partially visible in cross-sections 3T and 4T. In the area between 3T and 4T, the continuity of this layer is not clear. However, stratigraphical details (such as the peat layer) are sometimes lost due to the procedure used during drilling (most of the wells were finalized to groundwater exploitation). The RB1 borehole confirms the presence of this layer. Furthermore, at a depth of approximately 90–100 m (from -70 to -80 m with respect to the sea level, Fig. 2), the presence of a second peat layer is visible, but most of the cross-sections are not this
deep. Thus, it is impossible to speculate about the actual continuity of this deeper peat layer. This peat horizon is visible in cross-sections 4T and 1L. From a strictly hydrogeological perspective, analysis of lithostratigraphic cross-sections has identified one unconfined aquifer and three superimposed aquifers ranging from 0 m to more than 170 m in depth. The typical quantitative and qualitative characteristics of each aquifer are reported in Table 1 and are described as follows. Unconfined aquifer (0–24 m.b.g.l.). This aquifer is visible on all of the shown cross-sections, where a shallow
123
BRENTA RIVER
Fig. 2 Cross-sections of the study area (see Fig. 1)
123
BRENTA RIVER
BRENTA RIVER
BRENTA RIVER
BRENTA RIVER
BRENTA RIVER
Environ Earth Sci
Environ Earth Sci
Fig. 3 Pictures from the PM1 borehole: organic matter concentrated in thin peat layers (arrows) approximately 13.5 m below ground level (a); and disseminated as isolated small plaques (arrows) within clay–silt sediments approximately 17.5 m below ground level (b)
Table 1 Aquifer range of depth in meters below ground level (m.b.g.l.) Depth (m.b.g.l.)
Head (m.a.g.l.)
Flow ratemean (l/s)
Flow ratemax (l/s)
Cond. (lS/cm)
Temp. (C)
pH
Aquifer
Number of wells
Unconfined
43
0–24
/
/
/
600
16.2
7.3
I confined
27
25–60
1.3
0.08
0.6
314
14.0
8.0
II confined
8
65–85
1.3
0.20
0.6
400
16.3
7.5
III confined
24
140–180
1.5
0.17
2.1
450
17.0
8.3
The principal parameters for the unconfined and three confined aquifers are also reported (head data measured using a manometer, chemical– physical data acquired by portable instruments)
layer of gravel is observed with decreasing thickness from north to south. Cross-sections 1L and 2L exhibit a different lithological setting that consists of mainly gravel in the north and sand in the south. This aquifer is exploited for domestic, irrigation and farming purposes. The water table is near the surface. The unconfined aquifer hydrostratigraphy is comparable in the right and left river banks. First confined aquifer (25–60 m.b.g.l.). The first aquifer is highly visible on the 1T profile, in the western part of 4T, and in the longitudinal cross-section 1L. This aquifer is artesian, and the potentiometric level is above ground level (m.a.g.l) (Table 1). In this aquifer, the most ancient wells are present, but intense exploitation of this aquifer occurred during the second post-war period. This aquifer is intensely exploited, and our estimation of the total extraction in the
studied area is approximately 200 l/s, with an increase of 55 l/s in the last decade. This aquifer is also exploited for commercial purposes as mineral water. Along the right bank of the Brenta river, this aquifer reduces its hydraulic head and thickness. Second confined aquifer (65–85 m.b.g.l.). The threedimensional reconstruction of this aquifer is less known because the stratigraphies available in the left Brenta bank are rarely deeper than 50 m. The information contained in cross-sections 1T, 2T, 3T and 4T shows that the thickness of this aquifer is very variable. The water presents a higher electrical conductivity than the first aquifer (Table 1). This aquifer is poorly known in the left Brenta because only a few wells reach this depth, due to the good productivity and quality of the first aquifer. The exploitation of this
123
Environ Earth Sci
aquifer ended in the 1990s, and its only current use is ornamental (fountains). This aquifer shows high concentrations of ammonium, iron and manganese, and it is not potable (Mion et al. 2009). Most of the wells were drilled in the left Brenta. Third confined aquifer (140–180 m.b.g.l.). This aquifer is visible in cross-sections 2T and 3T. Unfortunately, in the middle part of the study area, only stratigraphies shallower than 50 m (first aquifer) are available; thus, the third aquifer is not geometrically defined. All of the wells were drilled in the right Brenta. All the hydrogeological characteristics of these artesian wells present consistent differences (Table 1). In our study, the exploitation of this third aquifer visibly increased from 1995 to 2000. In Table 1, the principal parameters of the aquifers are presented. These parameters include depth and thickness, the value of the potentiometric level (i.e., the height at which the water flows without a pump), the average and maximum flow rates (without a pump), the electrical conductivity of the water, the temperature and the pH.
Methods Hydrochemical methods All of the samples were collected between June 2010 and November 2010, which is during the dry season and before the beginning of the heavy autumn rains. A sample survey was planned to collect water samples with a homogeneous spatial distribution throughout the entire study area, which is elongated along the flow direction of the Brenta River, approximately from northwest to southeast. Most of the samples (44) were obtained from the unconfined aquifer, which is preferentially exploited in this area (Mion et al. 2009). Samples from deeper aquifers were also collected (16 from aquifer I, 3 from aquifer II and 2 from aquifers deeper than 180 m), representing artesian aquifers up to 330 m in depth. The boreholes were purged for 30 min before sampling. Water temperature, Eh, pH, dissolved oxygen (DO) and electrical conductivity parameters were measured in the field, collecting the samples in a beaker. The redox potential was measured after prolonged immersion of the electrode (at least 20 min) and assuring measurement stability. Other parameters that were measured in the field include alkalinity (titration with HCl using phenolphthalein and methyl orange and calculating the total alkalinity as HCO3-), sulfides (excess iodine titrated with thiosulfate), free CO2 (Na-carbonate titration), ammonium ion, fluoride, nitrite and sulfate (using an Hanna spectrophotometer).
123
In the laboratory, the untreated water samples were stored in the refrigerator and analyzed within 24 h for nitrate, chloride and phosphate with a Perkin Elmer UV–VIS spectrophotometer. All of the other elements considered in this study (Si, Na, K, Mg, Ca, Fe and As) were determined on untreated waters (as for all the other parameters), but on a subsample that was acidified with HNO3 immediately after sampling, to get pH \ 2. A Perkin-Elmer 5000 atomic absorption spectrometer (AAS) was used for the Na and K measurements, and a Jobin–Yvon JY 38Plus inductively coupled plasma optical emission spectrometer (ICP-OES) was used to measure the other elements. The total dissolved solids (TDS) were measured by evaporating the sample in a platinum crucible at 180 C (Hem 1985). Geostatistical method The data concerning the As concentration in the groundwater from the unconfined aquifer (up to 24 m in depth) were used to study its statistical and spatial distribution. The dataset is represented by original data collected by CNR in 2010 and by additional data collected from 1992 to 2005 by ARPAV, the Environmental Protection Agency for the Veneto Region (a total of 445 data points, from the two datasets). Geostatistical analyses were conducted by R code (R Development Core Team 2012) using the Gstat package in particular (Pebesma 2004). Many different spatial continuity indices and graphical tools can be used to quantify spatial continuity. A classical (semi-) variogram (Isaaks and Srivastava 1989) of the arsenic values was used, which is given by: n 1 X c ð hÞ ¼ ½zðxi Þ zðxi þ hÞ2 ð1Þ 2N ðhÞ i¼1 The experimental variogram demonstrates how the mean squared differences between pairs of samples z(xi) and z(xi ? h) varies according to h (modulus and direction). The variographic analysis can also be used to analyze the differences in spatial variability of an indicator variable I(xi; z) with respect to a particular threshold, z. In this study, the arsenic potability limit will be used as z: 1 if zðxi Þ z z ¼ cutoff I ðxi ; zÞ ¼ ð2Þ 0 otherwise The indicator variograms obtained with this procedure can be also interpreted as a frequency of transition from one state (above the threshold) to another (below the threshold) and can thus inform the spatial clustering of values above or below the chosen threshold.
Environ Earth Sci
Although experimental variograms are useful exploratory tools, variogram structure identification is important but not sufficient to achieve this study’s goals. The next step is the inference process, which is necessary for every chosen kriging procedure. To synthesize a spatial continuity structure, it is necessary to infer a mathematical function called a ‘‘theoretical variogram’’, which can synthetically represent an experimental variogram. During the inference process, in which the theoretical variogram model and its parameters are chosen, hydrogeological information should be taken into account to support the data analysis. Moreover, to conduct prediction using both ordinary and indicator kriging (Journel 1983), the theoretical variograms must satisfy a definite-positive condition (Chiles and Delfiner 1999).
Hydrochemistry results
Fig. 4 Piper diagrams showing the main hydrochemical composition of the groundwater under consideration
Hydrochemical characteristics Most of the samples (88 %) have low mineral content (average TDS of approximately 400 mg/l, conductivity of 0.6 mS/cm), and only a few (12 %) of the samples have TDS higher than 500 mg/l. The mineral content is always below the limit for waters rich in mineral salts (Council Directive 80/777/EEC 1980). The Piper’s diagram reported in Fig. 4 summarizes the major element chemistry of the waters under consideration. Table 2 reports the average, standard deviation, maximum and minimum values of all analyzed parameters. The entire dataset shows a rather monotonous chemistry, with 2 situations (4 samples) representing remarkable exceptions. Most of the samples fall near the Ca-bicarbonate corner of the Piper’s diagram, which corresponds with the carbonate abundance in the aquifers found in Brenta sediments (Jobstraibizer and Malesani 1973; Venzo and Gajo 1998; Monegato et al. 2010). Two samples that had high Na contents are located far from the others in the plot (black squares in Fig. 4) but still have a bicarbonate anionic signature. These samples represent the deepest waters of the entire dataset (200 and 330 m), and their composition is consistent with the evolution of Ca-bicarbonate water undergoing cation exchange reactions (Ca–Na), as quantitatively shown by Cheng (1988). Two other samples are outside the main population in the plot (grey squares in Fig. 4). These samples come from two neighboring shallow wells (\10 m in depth and 1,200 m apart) that are lower in Ca and Mg and slightly higher in Na; thus, their shifted position in the Piper’s diagram likely reflects local conditions. The variability of the chemical data was evaluated according to two parameters: (1) the depth (Fig. 5); and (2)
the geographic position of the wells (Fig. 6). The well positions are expressed by the distance between the wells and the apex of the Brenta megafan (i.e., the town of Marostica). This distance approximates the position of the wells along the flow direction of groundwater within the study area (from northwest to southeast) and is indicated by the parameter DW (measured in meters). The Brenta megafan, as defined by Fontana et al. (2008), covers the entire study area. Figure 5 shows some parameters that are more sensitive to variations in depth. The conductivity ranges from an average value of 603 lS/cm in the unconfined aquifer to an average value of 437 lS/cm in the first confined aquifer. As expected, the conductivity and TDS parameters are strongly correlated each other (r = 0.987, 52 observations). The decrease in conductivity with depth is coupled by an increase in pH, which is rather variable in the unconfined aquifer and becomes slightly alkaline below it. Furthermore, the Ca–Mg ratio shows significant variations between the unconfined aquifer and first confined aquifer, up to a depth of 67 m. In Fig. 5, two historical datasets are plotted with the original data (2010 survey) to increase the number of samples. The trends described for the 2010 survey are supported by historical data, even though the older databases were collected over a considerably larger time interval (from 1987 to 1993 and from 1999 to 2008). The more scattered data points in the external dataset can be accounted for by data collection spanning several hydrologic years. Historical datasets allow for the appreciation of the higher NO3- content in the north (Carmignano), as well as its decrease in depth.
123
Environ Earth Sci Table 2 Summary of depth, chemical and physical analyses of well water n
Average
Standard deviation
Min
Max
Depth (m)
65
31
47
6
330
T (C)
65
16
2
13
24
0.1
TDS (mg/l)
52
408
117
205
630
1 1
Cond (lS/cm)
65
556
155
253
873
pH
65
7.3
0.3
6.7
8.1 203
Detection limits
Eh (mV)
65
-44
149
-285
DO (ppm)
65
2.1
1.8
0.0
6.0
\0.01
CO2 (mg/l)
64
42
28
7
114
\0.2
HCO3- (mg/l)
65
396
114
171
773
1
Cl (mg/l)
65
11.9
13.4
\0.3
59.9
0.3
SO42- (mg/l) NO3- (mg/l)
65 65
21 8.7
21 10.6
\5 \0.5
95 66.1
5 0.5
-
PO43- (mg/l)
65
0.18
0.21
0.01
1.00
\0.01
F- (mg/l)
65
0.23
0.14
0.02
0.65
\0.01
?
NH4 (mg/l)
61
2.8
4.6
0.01
29.0
0.01
SiO2 (mg/l)
65
16.2
7.1
5.9
32.1
0.02
Na? (mg/l)
65
14
27
2
166
0.03
K? (mg/l)
65
3.1
5.0
0.6
29.4
0.05
Mg2? (mg/l)
65
23
8
12
40
0.001
2?
Ca
65
82
29
23
147
0.01
Fe (lg/l)
(mg/l)
65
842
1,768
2
[13,000
0.1
As (lg/l)
65
35
86
\0.4
431
0.4
A few values (less than 25 %) below the instrumental detection limits have been substituted by half of the detection limit (Farnham et al. 2002). TDS total dissolved solids, Cond conductivity, DO dissolved oxygen. The sensitivity of pH and Eh is \0.01 and 0.1, respectively
In addition to the variations in depth described above, many chemical parameters are observed to change from northwest to southeast. Nevertheless, such variations can only be clearly identified for the unconfined aquifer because most of the wells are very shallow in the southern part of the study area. In Fig. 6, the measured parameters are plotted against the DW values, to emphasize that several analytical parameters are controlled by the position of the wells (unconfined aquifer) along the approximately 20 km extension of the study area. The relationships between the DW parameter and water chemistry (Fig. 6) are mainly expressed by changes in the slope of linear trends (Na?, Mg2? and pH) or by a more-or-less sharp transition from a clear linear relationship in the north to more scattered values towards the south (conductivity, HCO3- and SiO2). Other parameters (NH4? and Fe) change when Eh becomes negative, shifting from almost undetectable to appreciably high values. Arsenic in groundwater The concentration of arsenic in the water samples varies significantly, ranging from \0.4 ppb to 431 ppb (Table 2).
123
With respect to the quality of the water resource, the most relevant feature is the high concentration of this element in the groundwater. In particular, the limit for drinkable water (10 ppb, Council Directive 98/83/EC 1998) is far below the average value measured in the area (35 ppb), even if the median is considerably lower (5 ppb). When considering the variability of As concentrations, the number of wells that pass the drinkable limit can be considered instead of the average and of the median concentration. In this case, considering that 32 % of the samples are above the drinkable limit, the occurrence of a significant As anomaly is confirmed. According to both the original data and older data from ARPAV (Environmental Protection Agency for the Veneto Region, sampling from 1992 to 2005), significant variability in As concentration characterizes both the vertical profile and spatial distribution of the element (Fig. 7a, c). In general, low As concentrations can be found irrespective of the well depth. However, it is very interesting to note that the As anomaly decreases from shallow to greater depths (dashed line in Fig. 7a). In Fig. 7b, data from Bangladesh (BGS and DPHE 2001) are reported for comparison purposes. Naturally occurring As in groundwater (up to 2,500 ppb) is well
Environ Earth Sci
0
400
800
1200
NO3- (mg/l)
pH
Conductivity (µS/cm) 6.5
7
7.5
8
8.5
0
20
40
60
Ca/Mg 80
1
2
3
4
5
6
0
50
100
150
200 CNR survey 2010
250
CNR survey 2010
CNR survey 2010 Piazzola dataset (1987-1993)
300
Carmignano dataset (1999-2008)
Depth (m) 350
Fig. 5 Chemical variation of selected parameters with depth. The dataset provided by the municipalities of Carmignano and Piazzola is also reported (small empty symbols). The dashed line indicates the
transition from the unconfined to confined aquifer approximately 24 m below ground level
documented in a wide Bangladeshi area of approximately 150 000 km2, where the relevant aquifers consist of alluvial/deltaic sediments with abundant organic matter. It is evident that samples from Bangladesh show a behavior similar to that of the samples from the Venetian Plain, with As anomalies that also decrease with depth. Another similarity is the positive relationship between As and NH4? observed in Bangladesh (BGS and DPHE 2001) and the Venetian Plain (Pearson correlation coefficient = 0.50, 50 observations). Such similarities allow to consider analogue mechanisms responsible for elevated groundwater As concentrations. Because data are available for a continuous DW range, the distribution of As values versus the DW parameter has been considered for the unconfined aquifer. The distribution of As concentrations along DW distance follows a rule: concentrations are rather low for DW distances less than 17,000 m but become higher and more variable with increasing DW (Fig. 7c). A remarkable anomaly can be detected at approximately 21,500 m DW, where 3 samples (2010 sampling survey) show very high As concentrations of up to 431 ppb. Such a small cluster of samples can be considered a ‘‘hot spot’’ (see Smedley 2006) that is characterized by the highest As contents (2010 survey) and by a strong threshold with respect to adjacent points (Fig. 8a). At the same DW distance, another well demonstrated similar features (330 ppb As in 1993, 24 m depth), and one of the ‘‘hot spots’’ recognized in the 2010 survey was sampled previously in 1995, yielding very similar results (318 ppb As).
Geostatistical results A preliminary statistical analysis, reported in Fig. 8b, shows a lognormal distribution of As in the unconfined aquifer, with a skewness value of 6.5. Such an As distribution is often observed in datasets (Angelone et al. 2009; Hosono et al. 2011), and a data transformation is frequently conducted before geostatistical analysis, mainly when a gaussian kriging (Digger and Ribeiro 2007) is used. In this case, it is preferable not to transform the As data because, when confronted with a more difficult variogram interpretation, the actual As concentration previsions are obtained directly in As concentrations. Before calculating the experimental variograms, it is useful to analyze the variogram cloud (Ploner 1999). The variogram cloud is an extremely helpful tool for checking spatial continuity and detecting outliers. Figure 9 illustrates how the variogram cloud increases with h and the presence of outliers, as suggested by the initial statistical analysis. According to Isaaks and Srivastava (1989), a convenient tool to visualize experimental variograms is to also analyze the variogram simultaneously in all directions with a surface variogram. The surface variogram in Fig. 10a reveals an anisotropy in the continuity structure of arsenic in the study area. In particular, the experimental surface variogram detects an anisotropy with a maximum continuity at 55. Thus, directional experimental variograms related to the arsenic along and orthogonal to the principal anisotropy (along 10, 55, 100 and 145) are shown in Fig. 10b. Analysis of Fig. 10b confirms the anisotropy visible in the experimental surface variogram of Fig. 10a.
123
Environ Earth Sci
Fig. 6 Variations in the main analytical parameters as a function of DW values for the unconfined aquifer (DW well distances respect to the apex of Brenta megafan, expressed in meters)
The variographic analysis can also be used to analyze the differences in spatial variability with respect to a particular threshold, which is the guideline value for As in this case. This type of elaboration allows for the identification of the local probability that the arsenic concentration in shallow groundwater exceeds the potability threshold of 10 ppb. With this in mind, the surface experimental indicator variogram is shown in Fig. 10c, where an anisotropy at approximately 90 is notable. Thus, to complete the spatial anisotropy analyses, directional experimental indicator variograms along the direction of maximum continuity and its perpendicular are shown in Fig. 10d. The indicator variogram analysis also demonstrates the maximum continuity at 90.
123
The variogram of arsenic (Fig. 10b) and the indicator variogram related to the arsenic threshold of 10 ppb (Fig. 10d) are fitted using exponential models. The comparison between experimental and theoretical variograms suggests a good reproduction of the spatial structure of As data (Fig. 10b and d). According to Davis (1987), the cross-validation procedure cannot confirm that a particular variogram model is or is not the optimum. Then, the resulting theoretical variograms will be used while applying different kriging procedures. The first variogram related to As values, reported as a continuous line in Fig. 10b is: 3h cðhÞ ¼ 100 þ 700 1 exp ð3Þ 1300
Environ Earth Sci
Fig. 7 Arsenic concentration with respect to well depth (a study area, b Bangladesh, BGS and DPHE 2001) and the DW parameter (c, unconfined aquifer only)
Fig. 8 Geographic distribution of the As concentration in the groundwater of the unconfined aquifer (a). Map coordinates are shown in metric units WGS84. Lines showing DW distances of 17,000 and 21,500 m are reported. The histogram showing the lognormal distribution of As data is also reported (b)
123
Environ Earth Sci
Fig. 9 Variogram cloud. The initial statistical analysis indicates an increase with h and the presence of outliers
where 100 is the ‘‘nugget effect’’ (12.5 %), 700 is the ‘‘sill’’ (87.5 %) and 1,300 is the ‘‘range’’ of the variogram, with a geometrical anisotropy of 0.46 at 55. The second variogram, related to the potability threshold and reported as a continuous line in Fig. 10d is: 3h cðhÞ ¼ 0:009 þ 0:035 1 exp ð4Þ 1000 where 0.009 is the ‘‘nugget effect’’ (20.5 %), 0.035 is the ‘‘sill’’ (79.5 %) and 1,000 is the ‘‘range’’. This model presents a geometrical anisotropy (0.5) at 90. After completing a detailed variogram analysis and the inference process, the next step is to apply kriging predictions. One of the main objectives of this process is the spatial prediction of arsenic distribution in the study area using an ordinary kriging. An additional result is the spatial prediction of the probability that the arsenic concentration in shallow groundwater exceeds 10 ppb (the arsenic potability threshold), using an indicator kriging. In Fig. 11a, the predicted distribution of arsenic in shallow groundwater is shown. The inset shows the prediction error variance in more detail, clearly showing that the reliability of the prediction varies throughout the study area. A preliminary analysis of Fig. 11a reveals a wide zone in the northwest part of the study area with low arsenic values and high kriging result reliability (Fig. 11b), accordingly to the high datapoint density. This zone is roughly delimited by a DW parameter of 17,000 m (which has been established according to several geochemical parameters). In the central part (DW between 17,000 and
123
21,500 m), the kriging result is more heterogeneous because domains with low and intermediate As values are side by side but can still be predicted (Fig. 11b). In contrast, southeast of the 21,500 m DW line, the kriging prediction shows domains with very high and very low As concentrations. Nevertheless, most of these low As domains are characterized by lower reliability due to the lack of wells. Moreover, the domains characterized by the highest As anomaly show a lenticular shape that matches the anisotropy at 55 (Fig. 10a, b). In Fig. 11c, the probability that arsenic concentration in shallow groundwater exceeds 10 ppb is shown, and its prediction error variance is reported in the related inset (Fig. 11d). The southeastern part of the studied area is characterized by a high probability ([51 %) of encountering non-potable water. However, within this southeastern part, the small zones with a low probability of exceeding the potability threshold must still be interpreted with caution if they are located in areas where the well density is very low (very low reliability). A different condition is typical of the band along the right bank of the Brenta River in the southeast, where the As concentration is low and the prediction reliability is good.
Discussion The widespread Ca-bicarbonate composition of groundwater in the study area is in agreement with the subsurface geology and sediment composition (Monegato et al. 2010), which is characterized by abundant amounts of calcite and dolomite. The unconfined aquifer shows a larger variation of selected parameters, such as conductivity, pH, NO3- and the Ca–Mg ratio, as compared to the deeper aquifers. In particular, the high nitrate content in the northern part of the unconfined aquifer was already known to be attributed to anthropogenic actions (Boscolo and Mion 2008). In confined aquifers ([24 m.b.g.l.), the effect of anthropogenic activity (NO3-) is minimized, and the Ca–Mg ratio decreases to as low as 2, which was almost reached by the deepest samples. The variation of the Ca–Mg ratio has been evaluated using a geochemical modeling approach (Visual MINTEQ; Gustafsson 2012) along the transition from the unconfined aquifer to the first confined aquifer. The main forcing variables are pH, the progressive attenuation of the atmospheric buffer for CO2 and the continuous presence of calcite and dolomite in the aquifers. These conditions cause the water from the unconfined aquifer to become more oversaturated in calcite approaching the first confined aquifer, thus promoting calcium carbonate precipitation (that is, lowering of the Ca–Mg ratio, as observed). Geochemical modeling predicts a value of 2.2 for the Ca–Mg ratio at equilibrium, which is
Environ Earth Sci
Fig. 10 Experimental surface variograms and related directional experimental variograms for the unconfined aquifer: experimental surface variogram for As concentration (a); directional experimental variograms for As concentration (b); experimental surface indicator
variogram for the drinkable limit threshold of 10 ppb As (c); and directional experimental indicator variograms for the threshold of 10 ppb As (d)
in agreement with the measured trend and also accounts for the lowering of TDS and conductivity. Despite the homogeneous Ca-bicarbonate composition of waters, strong variations in redox conditions occur from north to south and such redox variability is quite remarkable, ranging from 172 mV (in the north) to -272 mV (in the south), and is accompanied by changes of other ‘‘master’’ variables (such as pH and conductivity; Fig. 6). Approaching the distal areas of the Brenta depositional megafan, along the flow direction (DW parameter; Fig. 6), a stronger importance of the organic matter decay is suggested by increasingly negative Eh values, the increase in
NH4? and the decrease in pH (both aerobic or anaerobic decay determine a decrease in pH; Langmuir 1997). The effect of lowering the Eh and pH is an enhanced dissolution of carbonates, oxides and hydroxides, shifting the TDS and conductivity to higher values, as demonstrated by the increased DW parameter (Table 2; Fig. 6). Such a chemical evolution of groundwater from north to south agrees with stratigraphical data that indicate the disappearance of gravel and the presence of thick clay deposits towards the south (Fig. 2). Owing to the formation of peat in wet depressions, fine-grained sedimentation in the Venetian alluvial plains is characterized by abundant
123
Environ Earth Sci
Fig. 11 Predicted distribution of As (a) and related error variance (b); probability that As concentration in shallow groundwater exceeds 10 ppb (c); and related error variance (d). Map coordinates are shown in kilometric units WGS84 (OK ordinary kriging, IK indicator kriging)
123
Environ Earth Sci
organic matter (Miola et al. 2006; Fontana et al. 2008; Zecchin et al. 2011). Both peat deposits and disperse organic matter are detectable in boreholes (Fig. 3). As suggested by the positive relationship between As and NH4?, the role of organic matter in the study area has direct effects on the specific problem of As occurrence in groundwater. Similar conditions are well known worldwide (Varsanyi et al. 1991; BGS and DPHE 2001; Dowling et al. 2002; Ahmed et al. 2004; Dangic and Dangic 2007; Berg et al. 2008; Deng et al. 2009; Reza et al. 2010; Winkel et al. 2011; see also several examples shown by Ravenscroft et al. 2009 and the review by Smedley 2006). Another similarity between the Venetian Plain and other areas affected by As groundwater contamination (such as Bangladesh; BGS and DPHE 2001) is the decrease in the maximum As content at certain depths (Fig. 7). In particular, at intermediate depths (100–200 m), the variation in As content is less pronounced, and maximum values are significantly lower (average in Bangladesh: As = 45 ± 49 ppb; see Fig. 7 for the Venetian Plain). Below 200 m.b.g.l., the groundwater As content is still lower (average in Bangladesh: As = 4 ± 10 ppb; see Fig. 7 for the Venetian Plain), implying an increasing probability of finding water with acceptable As content. Furthermore, this feature can be explained by considering the role of labile organic matter, which loses its reducing capability with aging (BGS and DPHE 2001; Ravenscroft et al. 2009), as well as the effect of flushing (more prolonged in deeper aquifers), which has been identified as a key factor (BGS and DPHE 2001; Ravenscroft et al. 2005). According to the chemical data reported in Figs. 6 and 7c (including the master variables pH and Eh) and the As distribution (see ordinary kriging predictions in Fig. 11), an important change in water chemistry occurs from northwest to southeast, between a DW parameter of 17,000 and 21,500 m. This trend represents a transitional discontinuity in the area and marks a northern and southern domain featuring different likelihoods of finding potable water with respect to As concentration (Fig. 11c). In particular, the southern domain is clearly characterized by an ‘‘anomalous background’’ identified by diffuse high As values up to 150 ppb, which is considerably higher as compared to the drinkable limit of 10 ppb (Figs. 7c, 8a and 11c). Nevertheless, the main environmental and health concern goes beyond such diffuse ‘‘anomalous background’’ and is represented by the As ‘‘hot spots’’ (As [ 300 ppb) punctuating the distance reference of 21,500 m DW (Fig. 8a). A speculative explanation for the peculiar locations of the ‘‘hot spots’’ considers the abrupt increment of the retardation factor (Rd) for As. The increased As concentration is consistent with lithological variations (from coarser to finer sediments) that determines an increase in Rd due to changes in sediment texture and the distribution coefficient
of As for clays. A similar mechanism accounts for the occurrence of As ‘‘hot spots’’ in South and East Asian countries (The World Bank 2005) and also as a general rule (Smedley 2006). In the study area, a lithological variation appropriate to account for the occurrence of these ‘‘hot spots’’ corresponds to 21,500 m DW. Such lithological discontinuity of the unconfined aquifer is emphasized by matching the DW parameter (that represents a simple geometric expression of the sediment dispersion) with the kriging arsenic predictions. In further lithostratigraphic detail, cross-sections 1L and 2L (Fig. 2) show that the thinning of gravel deposits towards the south (unconfined aquifer) gives rise to a fine-grained aquifer passing through the 21,500 m DW. The abundance of clay-silt materials in the southern part of the unconfined aquifer appears to be a main constraint for water quality with respect to As concentration. In particular, the probability of exceeding the threshold limit of 10 ppb (Fig. 11c) reflects this condition. An example is the southern part of the study area, where the probability is often very high but where the easternmost side is characterized by a low probability. The lithostratigraphic arrangement is represented by cross-section 4T (Fig. 2), where the unconfined aquifer (0–24 m.b.g.l.) is mainly sandy west of the Brenta River for approximately 2 km and where the probability of encountering non-potable water is low. In contrast, the aquifer becomes silty-clay to the West, where it contains peat material and where the probability increases by over 50 %. A final consideration relates to the origin of As anomalies and their stationary or stable areal distribution. No significant variation can be found between the most recent analytical survey (2010) and historical data covering the last 20 years. In particular, both the ‘‘anomalous background’’ in the southern area and the ‘‘hot spots’’ appear to be very stable in space and intensity. Because the distribution of industrial plants does not match the distribution of As in groundwater, an anthropogenic origin should be considered cautiously. Furthermore, the possible input of As salts as pesticides before 1976 (which was the ban year in Italy; Celli 1995) does not agree with the traditional agricultural practice of the area (Bittante et al. 1985; Celetti 2008; Varini 2008). The alternative hypothesis, a natural origin of As in groundwater, cannot be completely assured but is consistent with the geology of the area. In particular, the soils developed on the Brenta fan (including the study area) and Adige fan have As contents much higher than the adjacent depositional systems (such as the Piave and Tagliamento), which are similar in age and tectono-stratigraphic setting (average values for topsoils: Brenta = 21 ppb, Adige = 19 ppb, Piave = 8 ppb, Tagliamento = 10 ppb; Giandon et al. 2011). The high As concentration in the Brenta and
123
Environ Earth Sci
Adige soils (in a large area extending over thousands of square kilometers) can be accounted for, at least qualitatively, by the occurrence of sulfide ore deposits in the catchments, several tens of kilometers from the study area (see the compilation by De Vivo et al. 2009). A prevailing geogenic origin of As in the Brenta soils has been identified by Ungaro et al. (2008), who studied an area in the southern Brenta fan.
minimized by the provision of potable water to the public. However, the pipe network is not as pervasive as it needs to be. The health effects of As pollution in groundwater are generally very severe, including chronic exposures occurring at relatively low concentrations (see the review by EFSA 2009), particularly among the elderly (see, for example, Tseng 1977). Accordingly, geochemical, lithostratigraphical and geostatistical analyses are powerful tools for the study of the environment and healthcare.
Summary and conclusions
Acknowledgments This study has been funded by National Research Council (CNR) of Italy, Veneto Region, the Brenta ‘‘Autorita` d’Ambito Territoriale Ottimale’’ (AATO) and Carmignano municipality. The ‘‘Agenzia Regionale per la Prevenzione e Protezione Ambientale’’ (ARPAV), Carmignano and Piazzola municipalities are greatly appreciated for providing historical datasets. Anonymous reviewers are acknowledged for improving the manuscript.
The study area represents a typical northern Italian plain area where industrial, agricultural and economic activities have been well developed for a long period of time. The social and economic wellness reaches high standards respect to the western countries and rural activities are also prosperous. This condition requires the use of local groundwater for private purposes (low volumes), social and economic activities (large volumes). In this context, the results of the present study are as follows: •
•
•
•
•
several water wells (32 %) from the most exploited aquifer (the unconfined one) have been found to contain As in excess of drinkable limits; the kriging arsenic prediction map (Fig. 11a) shows that most of these high As values are located in the southern part of the study area, where reducing groundwater conditions occur; the indicator kriging prediction map (Fig. 11c) demonstrates the high probability of encountering non-potable water in a large part of the unconfined aquifer, which is the most frequently exploited aquifer in the area; the most remarkable groundwater As anomalies ([300 ppb) are found at the transition from a gravel to a silty-clay reservoir in the unconfined aquifer. These anomalies have the characteristics of so-called ‘‘hotspots,’’ which are observed worldwide and are characterized by very high As concentrations and a strong threshold with respect to adjacent wells from the same aquifer; a rapid decrease in As concentration with depth was observed, allowing for the prediction that more favorable water quality (at least with regard to As) can be found below a depth of 150–200 m.
An anthropogenic source of As appears to be unlikely in the study area. In fact, soils on silty-clay Brenta sediments have higher As concentrations than most common soils, but the positive relationship between As and NH4? suggests that sedimentary organic matter is the first candidate to determine As anomalies in groundwater. The likely natural origin of As contamination suggests that the problem goes well beyond the area considered in this study. In western countries, health effects are
123
References Ahmed KM, Bhattacharya P, Hasan MA, Akhter SH, Alam SMM, Bhuyian MAH, Imam MB, Khan AA, Sracek O (2004) Arsenic enrichment in groundwater of the alluvial aquifers in Bangladesh: an overview. Appl Geochem 19:181–200 Ahuja S (ed) (2008) Arsenic contamination of groundwater. Mechanism, analysis, and remediation. Wiley, USA Alerts GJ, Khouri N (2004) Arsenic contamination of groundwater: mitigation strategies and policies. Hydrogeol J 12:103–114 Angelone M, Cremisini C, Piscopo V, Proposito M, Spaziani F (2009) Influence of hydrostratigraphy and structural setting on the arsenic occurrence in groundwater of the Cimino-Vico volcanic area (central Italy). Hydrogeol J 17:901–914 Baldantoni E, Ferronato A (1995) Presenza di arsenico nelle acque di falda del Mediobrenta: aspetti ambientali e sanitari (Occurrence of arsenic in the Mediobrenta groundwater: environmental and health). Quad Geol Appl Supplement 2:421–427 Berg M, Trang PTK, Stengel C, Buschmann J, Viet PH, Van Dan N, Giger W, Stu¨ben D (2008) Hydrological and sedimentary controls leading to arsenic contamination of groundwater in the Hanoi area, Vietnam: the impact of iron-arsenic ratios, peat, river bank deposits, and excessive groundwater abstraction. Chem Geol 249:91–112 BGS (British Geological Survey), DPHE (Department of Public Health and Engineering) (2001) Arsenic contamination of groundwater in Bangladesh. In: Kinniburgh DG, Smedley PL (eds) British geologic survey report WC/00/19, vols 1–4. BGS, Keyworth Bittante G, Ziliotto U, Spanghero M, Bettini D, Santomaso R, Ferroli R (1985) Carta foraggera del Destra Brenta (Forage map of the Brenta right bank). Provincia di Padova, Associazione allevatori, Padova, pp 17–21 Bondesan A, Meneghel M (2004) Geomorfologia della provincia di Venezia (Geomorphology of Venice province). Esedra, Padova Boscolo C, Mion F (2008) Le acque sotterranee della pianura veneta. I risultati del progetto SAMPAS (Groundwater of the Venetian plain. Results from the SAMPAS project). ARPAV, Agenzia Regionale per la Prevenzione e Protezione Ambientale, Regione Veneto, Padova Celetti D (2008) Lo sviluppo economico e sociale (Economic and social development). In: Fontana GL (ed) Carmignano di Brenta nel Novecento (Carmignano di Brenta in the Nineteenth century). Cleup, Padova
Environ Earth Sci Celli G (1995) I composti dell’Arsenico come insetticidi. Atti del Convegno Nazionale ‘‘Arsenico, sı`-no’’ (Arsenic compounds as insecticide. Proceedings of the National Meeting ‘‘Arsenic, yesno’’). Codroipo, Udine, Italy, pp 61–69 Chen CJ, Kuo TL, Wu MM (1988) Arsenic and cancers. Lancet 331:414–415 Cheng S (1988) Trilinear diagram revisited: application, limitation, and an electronic spreadsheet program. Ground Water 26:505–510 Chiles JP, Delfiner P (1999) Geostatistics modeling spatial uncertainity. Wiley, New York Chiou HY, Huang WI, Su CL, Chang SF, Hsu YH, Chen CJ (1997) Dose-response relationship between prevalence of cerebrovascular disease and ingested inorganic arsenic. Stroke 28: 1717–1723 Council Directive 80/777/EEC (1980) On the approximation of the laws of the Member State relating to the exploitation and marketing of natural waters. Off J Eur Commun, L 229 Council Directive 98/83/EC (1998) On the quality of water intended for human consumption. Off J Eur Commun, L330 Cucchi F, Franceschini G, Zini L (2008) Hydrogeochemical investigations and groundwater provinces of the Friuli-Venezia Giulia aquifers, northeastern Italy. Environ Geol 55:985–989 Dangic A, Dangic J (2007) Arsenic in the soil environment of central Balkan Peninsula, southeastern Europe: occurrence, geochemistry, and impacts. In: Bhattacharya P, Mukherjee AB, Bundschuh J, Zevenhoven R, Loeppert RH (eds) Arsenic in soil and groundwater, trace metals and other contaminants in the environment, vol 9, pp 207–236 Davis BM (1987) Uses and abuses of cross-validation in geostatistics. Math Geol 19:241–248 De Vivo B, Lima A, Cicchella D, Frizzo P, Di Lella A, Raccagni L, Bove MA, Albanese S, Grezzi G, Sabatini G, Protano G, Riccobono F (2009) Geochemical environmental atlas of Italy. Aracne, Rome Deng Y, Wang Y, Ma T, Gan Y (2009) Speciation and enrichment of arsenic in strongly reducing shallow aquifers at western Hetao Plain, northern China. Environ Geol 56:1467–1477 Digger PJ, Ribeiro PJ (2007) Model-based geostatistics. Springer, Berlin Dowling CB, Poreda RJ, Basu AR, Peters SL, Aggarwal PK (2002) Geochemical study of arsenic release mechanisms in the Bengal Basin groundwater. Water Resour Res 38:1173 EFSA Panel on Contaminants in the Food Chain (CONTAM) (2009) Scientific Opinion on Arsenic in Food. EFSA J 7(10): 1351. doi: 10.2903/j.efsa.2009.1351, Available online: www.efsa.europa.eu Farnham IM, Singh AK, Stetzenbach KJ, Johannesson KH (2002) Treatment of nondetects in multivariate analysis of groundwater geochemistry data. Chemom Intell Lab 60:265–281 Fontana A, Mozzi P, Bondesan A (2008) Alluvial megafans in the Venetian–Friulian Plain (north-eastern Italy): evidence of sedimentary and erosive phases during Late Pleistocene and Holocene. Quat Int 189:71–90 Giandon P, Garlato A, Ragazzi F (2011) Metalli e metalloidi nei suoli del Veneto – Determinazione dei valori di fondo (Metals and metalloids in the Veneto soils – Background determination). ARPAV, Agenzia Regionale per la Prevenzione e Protezione Ambientale, Regione Veneto, Padova Gustafsson JP (2012) Visual MINTEQ version 3.0. A free equilibrium speciation model. Available at http://ww2.lwr.kth.se/english/ OurSoftware/Vminteq/index.htm. Accessed March 5, 2012 Hem JD (1985) Study and interpretation of the chemical characteristics of natural water (3rd edn). US Geological Survey WaterSupply Paper 2254 Hosono T, Nakano T, Shimizu Y, Onodera S, Taniguchi M (2011) Hydrogeological constraint on nitrate and arsenic contamination
in Asian metropolitan groundwater. Hydrol Process 25: 2742–2754 Isaaks MH, Srivastava MR (1989) An introduction to applied geostatistics. Oxford University Press, Oxford Jobstraibizer PG, Malesani P-I (1973) Sedimenti dei fiumi veneti (Sediments of Veneto rivers). Mem Soc Geol Ital 12:411–452 Journel AG (1983) Nonparametric estimation of spatial distribution. Math Geol 12:285–303 Langmuir D (1997) Aqueous environmental geochemistry. PrenticeHall, Upper Saddle River Miola A, Bondesan A, Corani L, Favaretto S, Mozzi P, Piovan S, Sostizzo I (2006) Wetlands in the Venetian Po Plain (northeastern Italy) during the Last Glacial Maximum: interplay between vegetation, hydrology and sedimentary environment. Rev Palaeobot Palyno 141:53–81 Mion F, Boscolo C, Stecca L (2009) Progetto Mo.Sp.As. Monitoraggio sperimentale dello ione arsenico nelle acque sotterranee della media e bassa pianura veneta (Measure of As ion in groundwater of the middle and low Venetian plain). Relazione tecnicoscientifica n. 17/2009, Direzione Tecnica, Servizio Acque Interne (ARPAV), Padova Monegato G, Stefani C, Zattin M (2010) From present rivers to old terrigenous sediments: the drainage system in the eastern Southern Alps. Terra Nova 22:218–226 Pebesma EJ (2004) Multivariable geostatistics in S: the gstat package. Comput Geosci 30:683–691 Pilla G, Sacchi E, Zuppi G, Braga G, Ciancetti G (2006) Hydrochemistry and isotope geochemistry as tools for groundwater hydrodynamic investigation in multilayer aquifers: a case study from Lomellina, Po plain, South-Western Lombardy, Italy. Hydrogeol J 14:795–808 Ploner A (1999) The use of variogram cloud in geostatistical modeling. Environmetrics 10:413–437 R Development Core Team (2012) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna Ravenscroft P, Burgess WG, Ahmed KM, Burren M, Perrin J (2005) Arsenic in groundwater of the Bengal Basin, Bangladesh: distribution, field relations, and hydrogeological setting. Hydrogeol J 13:727–751 Ravenscroft P, Brammer H, Richards K (2009) Arsenic pollution: a global synthesis. RGS-IBG book series. Wiley, West Sussex Reza AHMS, Jean JS, Lee MK, Yang HJ, Liu C–C (2010) Arsenic enrichment and mobilization in the Holocene alluvial aquifers of the Chapai-Nawabganj district, Bangladesh: a geochemical and statistical study. Appl Geochem 25:1280–1289 Scialoja MG (ed) (2005) Presenza e diffusione dell’arsenico nel sottosuolo e nelle risorse idriche italiane: nuovi strumenti di valutazione delle dinamiche di mobilizzazione (Presence and diffusion of As in subsoil and italian water resources). ARPA Emilia-Romagna, Bologna Smedley PL (2006) Sources and distribution of arsenic in groundwater and aquifers. In: Appelo T (ed) Arsenic in groundwater—a world problem. Proceedings Seminar Utrecht The World Bank (2005) Towards a more effective operational response. Arsenic contamination of groundwater in South and East Asian Countries. Vol. II, Technical Report n 31303, Environment and Social Unit—South Asia Region, Water and Sanitary Program (WSP)—South and East Asia, Washington Tseng WP (1977) Effects and dose-response relationships of skin cancer and black foot disease with arsenic. Environ Health Perspect 19:109–119 Tseng CH, Chong CK, Chen CJ, Tai TY (1996) Dose-response relationship between peripheral vascular disease and ingested inorganic arsenic among residents in blackfoot disease endemic villages in Taiwan. Atherosclerosis 120:125–133
123
Environ Earth Sci Ungaro F, Ragazzi F, Cappellin R, Giandon P (2008) Arsenic concentration in the soils of the Brenta Plain (Northern Italy): mapping the probability of exceeding contamination thresholds. J Geochem Explor 96:117–131 Varini S (2008) La montagna che vive in pianura (The mountain that lives in the plain). Bozzetto Edizioni, Cartigliano Varsanyi I, Fodre Z, Bartha A (1991) Arsenic in drinking water and mortality in the southern Great Plain, Hungary. Environ Geochem Health 13:14–22 Venzo GA, Gajo A (1998) The sands of the Brenta River, northeastern Italy. Atti Istituto Veneto di Scienze, Lettere ed Arti, Classe di Scienze Fisiche, Matematiche e Naturali 156:39–66 Vorlicek P, Antonelli R, Fabbri P, Rausch R (2004) Quantitative hydrogeological studies of Treviso alluvial plain (north east of Italy). Q J Eng Geol Hydrogeol 37:23–29 Wang CH, Jeng JS, Yip PK, Chen CL, Hsu LI, Hsueh YM, Chiou HY, Wu MM, Chen CJ (2002) Biological gradient between long-term arsenic exposure and carotid atherosclerosis. Circulation 105: 1804–1809 Wang CH, Chen CL, Hsiao CK, Chiang FT, Hsu LI, Chiou HY, Hsueh YM, Wu MM, Chen CJ (2009) Increased risk of QT
123
prolongation associated with atherosclerotic diseases in arseniasis-endemic area in southwestern coast of Taiwan. Toxicol Appl Pharmacol 239:320–324 Welch AH, Stollenwerk KG (eds) (2003) Arsenic in ground water. Kluwer Academic Publishers, New York Winkel LHE, Trang PTK, Lan VM, Stengel C, Amini M, Ha NT, Viet PH, Berg M (2011) Arsenic pollution of groundwater in Vietnam exacerbated by deep aquifer exploitation for more than a century. Proc Natl Acad Sci USA 108:246–1251 Yamazaki C, Ishiga H, Ahmed F, Itoh K, Suyama K, Yamamoto H (2003) Vertical distribution of arsenic in Ganges delta sediments in Deuli Village, Bangladesh. Soil Sci Plant Nutr 49:567–574 Zaccone C, Cocozza C, Cheburkin AK, Shotyk W, Miano TM (2008) Distribution of As, Cr, Ni, Rb, Ti and Zr between peat and its humic fraction along an undisturbed ombrotrophic bog profile (NW Switzerland). Appl Geochem 23:25–33 Zecchin M, Caffau M, Tosi L (2011) Relationship between peat bed formation and climate changes during the last glacial in the Venice area. Sediment Geol 238:172–180