Colombani Nicolo tesi

Colombani Nicolo tesi
Alma Mater Studiorum – Università di Bologna
DOTTORATO DI RICERCA IN
Scienze ambientali: tutela e gestione delle risorse naturali
Ciclo XXVII
Settore Concorsuale di afferenza: 04/A1
Settore Scientifico disciplinare: GEO/08
TITOLO TESI
HYDROGEOCHEMICAL MONITORING AND MODELLING OF
SALTWATER INTRUSION IN LOWLANDS (FERRARA, IT)
Presentata da:
Nicolò Colombani
Coordinatore Dottorato
Relatori
Prof. Enrico Dinelli
Prof. Enrico Dinelli
Prof.ssa Maria Dolores Fidelibus
Esame finale anno 2014
Acknowledgments
Prof. Ramon Aravena, Prof. Enrico Dinelli and Prof. Maria Dolores Fidelibus are gratefully
acknowledged for their precious comments and suggestions.
Many thanks are also due to Dr. Dario Di Giuseppe, Dr. Renzo Tassinari and Dr. Umberto
Tessari for laboratory assistance and to Dr. Mariachiara Caschetto, Dr. Massimo Marchesi,
Dr. Enzo Salemi and Dr. Fabio Vincenzi for field assistance.
Beatrice Maria Sole Giambastiani and Micòl Mastrocicco are acknowledge for all the rest!
CONTENTS
THESIS CONTENTS
List of figures
iii
List of tables
viii
RIASSUNTO
Contesto e motivazione
Significato del sito di test selezionato
Obiettivi della tesi
Implicazioni e prospettive
ix
ix
xii
xiii
xvi
1. INTRODUCTION
1.1 Background and motivation
1.2 Significance of the selected test site
1.3 Objectives of the thesis
1
1
5
6
2. SITE DESCRIPTION
2.1 Study area
2.2 Morphological evolution during the last glacial period
2.3 Recent anthropogenic activity
2.4 Hydrogeological setting
9
9
10
12
13
3. MATERIALS AND METHODS
3.1 Field activities
3.1.1 Monitoring wells installation
3.1.2 Continuous monitoring of piezometric heads and surface waters levels
3.1.3 Depth profiles acquisition
3.1.4 Constant rate pumping tests
3.2 Analytical methods
3.2.1 Sedimentological characterization
3.2.2 Chemical analysis of water samples
3.2.3 Geochemical analysis of soil samples
3.3 Numerical modelling activities
3.3.1 Site wide steady-state simulation
3.3.2 Two dimensional transient simulation
3.3.3 Scenario modelling
16
16
16
17
17
19
21
21
22
22
24
24
28
32
i
CONTENTS
4. RESULTS AND DISCUSSION
4.1 Characterization of the lowland coastal aquifer of Ferrara: hydrology,
hydrochemistry and evolution of the system
4.1.1 Hydrostratigraphic characterization
4.1.2 Hydrochemical characterization
4.1.3 The origin of the salinization processes
4.2 The importance of data acquisition techniques in saltwater intrusion
monitoring
4.2.1 Open-Borehole Logging and Multi-Level Sampling
4.2.2 Integrated-Depth Sampling and Multi-Level Sampling
4.2.3 Discrepancy between an integral-depth and a depth-dependent
monitoring
4.3 Numerical modelling to predict salinization trends in a lowland coastal
aquifer
4.3.1 Three-dimensional model calibration and sensitivity analysis
4.3.2 Three-dimensional model results
4.3.3 Tidal forcing effects on groundwater fluctuation
4.3.4 Two-dimensional model calibration and validation
4.3.5 Future scenarios
4.4 Ammonium occurrence in a salinized lowland coastal aquifer
4.4.1 Water table elevation and ammonium mapping
4.4.2 Core samples depth profiles
4.4.3 Groundwater quality depth profiles
4.4.4 Groundwater mineralization processes
4.5 Trace elements mobility in a saline coastal aquifer of the Po river lowland
4.5.1 Sediment composition
4.5.2 Hydrochemical conditions of the shallow aquifer
4.5.3 Trace elements distribution in the shallow aquifer
4.5.4 Trace elements profiles in sediments and groundwater
4.5.5 Trace elements partitioning between groundwater and sediments
34
5. SUMMARY AND CONCLUSIONS
5.1 Synthesis
5.2 Conclusions
5.3 Recommendations
5.3.1 Practical implications
5.3.2 Remaining challenges
5.4 Related publications
87
87
87
90
90
91
92
6. REFERENCES
95
ii
34
34
39
43
47
48
53
55
58
59
62
63
65
68
71
72
73
75
77
78
78
81
81
83
85
LIST OF FIGURES
LIST OF FIGURES
Figure 1.1 – Wheat fields in the Ferrara lowland (www.estense.com).
Figure 2.1 - Ferrara Province topography map and location of the coastal lowland study area
(dashed rectangle).
Figure 2.2 – Upper panel: Location of the 13 monitoring wells, the sandy paleochannel belts,
the dunes and paleodunes in the study area. Lower panel: sketch of the study area showing
the location of the two boreholes P33, P34 and of the courses of the main rivers. Below is
shown the hydro-stratigraphyc reconstruction along the transect AA’, with geological logs and
piezocone resistance profiles.
Figure 2.3 – Waterworks in the Ferrara Province at the beginning of the XX century
(www.bonificaferrara.it).
Figure 2.4 – “Valle Lepri” dewatering station in the Ferrara Province lowland
(www.bonificaferrara.it).
Figure 2.5 - Excerpt of the coastal territory of the Duchy of Ferrara in 1648
(ilmuseodimirabello.com).
Figure 3.1 - Multilevel Sampler set up and installation.
Figure 3.2 - Grid discretization and boundary conditions applied to the model domain: no flow
cells (grey), constant heads cells (blue) drain cells (yellow) and river cells (cyan).
Figure 3.3 - Hydraulic conductivity (m/s) discretization of the model domain based on surface
geology and the decreasing hydraulic conductivity pattern found during the multi-level
pumping tests.
Figure 3.4 - Variogram fitted with a power law function (left plot), and distribution of
Specified Concentration in layer 10 derived from the interpolated measured TDS and pilot
points (right plot).
Figure 3.5 - Numerical grid, with the optimized kh discretization, in yellow the Drain cell and
in light blue the Time-Variant Specified-Head cells pertaining to the canal and to the pro-delta
silty clays connected with the Adriatic sea.
Figure 3.6 - In the upper left hand side plot, the actual cumulative transpiration (blue line) and
evaporation rates (black line) computed by Hydrus-1D are shown. In the upper right hand
iii
LIST OF FIGURES
side plot, the recharge rate computed by Hydrus-1D (blue line) and its discretization as input
in SEAWAT (red bars) are shown. In the lower left hand side plot, the modelled (blue line) and
observed (black line) canal water levels are shown. In the lower right hand side plot, the
modelled (blue line) and observed (black open circles) canal TDS values are shown.
Figure 4.1 - (a) Profiles of hydraulic conductivity measured by constant rate pumping tests in
the four monitoring wells and (b) Scatterplot of K values calculated by pedotransfer formula
(Kped) and K values resulting from permeability tests in P33 and P34.
Figure 4.2 - Transect AA’ across the coastal aquifer (trace in Figure 2.2) showing freshwater
head (hf in green color) and seepage (qz in red color). Note: positive value means flow from
the bottom toward the surface, while negative values indicate downward flow (recharge).
Figure 4.3 - Salinity (TDS in g/l) and concentration of the anions of major salts. The purple
line indicates the TDS concentration of the current Adriatic Sea water.
Figure 4.4 - Saturation index for gypsum, calcite and dolomite along the aquifer depth
calculated with PHREEQC.
Figure 4.5 - Plots of the concentration of Br-, Na+, SO4-, and Mg2+ with respect to Cl-. The purple
lines indicate the non-reactive mixing line between freshwater and modern seawater; the star
symbol indicates the concentration of the modern Adriatic Sea water.
Figure 4.6 - Variation of sample concentration of Na+, K+, Mg2+, and Ca2+ with respect to the
theoretical composition expected for a non-reactive mixing of Adriatic Sea water and
freshwater. Note: positive values indicate enrichment, while negative values indicate
depletion.
Figure 4.7 - TDS versus Depth, blue circles indicate the MLS samples and red crosses indicate
the IDS samples. Water quality classes are drawn according to Davis and De Wiest (1996).
Figure 4.8 - TDS distribution along the coastal aquifer. Dashed lines indicate TDS values of
monitoring campaign in November 2011.
Figure 4.9 - Plot of Cl-/Br- molar ratio versus Cl- (meq/l) for the groundwater samples (left
panel). Plot of Cl- versus Br- (meq/l) for the groundwater samples (right panel). Black lines
indicate the ideal mixing line between actual freshwater and seawater, while the grey shaded
areas represent the mixing area.
Figure 4.10 - Gibbs diagrams for cations (left panel) and anions (right panel).
iv
LIST OF FIGURES
Figure 4.11 - Plots of salinity with MLS versus OBL data for each monitoring well along
transect AA’; stratigraphy (core logs data) from figure 2.2 is displayed on each plot to assist
the interpretation.
Figure 4.12 - Plots of pH with MLS versus OBL data for each monitoring well along the
transect AA’; stratigraphy (core logs data) from figure 2.2 is displayed on each plot to assist
the interpretation.
Figure 4.13 - Plots of Eh with MLS versus OBL data for each monitoring well along the transect
AA’; stratigraphy (core logs data) from figure 2.2 is displayed on each plot to assist the
interpretation.
Figure 4.14 - Plots of O2 with MLS versus OBL data, for each monitoring well along the
transect AA’; stratigraphy (core logs data) from figure 2.2 is displayed on each plot to assist
the interpretation.
Figure 4.15 - TDS versus Depth, blue circles indicate the MLS samples and red crosses indicate
the IDS samples. Water quality classes are drawn according to Davis and De Wiest (1996).
Figure 4.16 - Eh versus depth for selected monitoring wells. Blue circles indicate the MLS
values and red lines indicate the IDS values.
Figure 4.17 - Eh versus pH plot. Grey shaded areas identify the main TEAP environments.
Figure 4.18 - Cl- distribution maps, according to Stuyfzand classification (1989), derived using
IDS data (left panel) and MLS data from the upper aquifer portion (mid panel) and from the
lower aquifer portion (right panel).
Figure 4.19 - Subdivision of the model in drainage basins.
Figure 4.20 - Scatter diagrams of calculated and measured hydraulic head (left plot) and of
salt concentrations (right plot).
Figure 4.21 - Piezometric map of the calculated heads (m a.s.l.).
Figure 4.22 – Three-dimensional isoconcentration map of the calculated salt concentrations,
along the transect AA’.
Figure 4.23 - Plots of continuous piezometric head monitoring of monitoring well P8 (black
line) and sea level (red line). The left plot shows the entire monitoring period, while the right
plot shows a zoomed timeframe of one week when the tidal oscillation is more evident. Note
that the y scales have different ranges.
v
LIST OF FIGURES
Figure 4.24 - Location of the study area and conceptual model of the transect across
monitoring well P8 and Canale della Gronda, selected for the two-dimensional numerical
model.
Figure 4.25 - In the left hand side plots: comparison of modelled hydraulic heads (blue lines)
versus observed hydraulic heads (grey lines) and their running average with daily period
(black lines), In the right hand side plots: comparison of modelled TDS (blue lines) versus
observed TDS (grey lines) and their running average with daily period (black lines). The
observed water levels and TDS in the drain ditch are plotted with black open circles.
Figure 4.26 - In the upper panels the simulated TDS contours for 12/12/2012 (left panel) and
for 26/02/2013 (right panel) are shown with the location of P8. Profiles in P8 of modelled
(black lines) versus observed (black open circles) TDS used to validate the numerical model
are shown in the lower plots.
Figure 4.27 - In the upper panels the actual average TDS contours for the end of the summer
(left panel) and for the end of the winter (right panel) are plotted, while the future scenarios
are all plotted for the end of the summer conditions. The projected LSLR are shown in the left
panels and HSLR are shown in the right panels. The black line represents the 2.5 g/L TDS
isoline.
Figure 4.28 - Location of the study area with the piezometric contour map (black lines);
circles represent the groundwater NH4+ concentration at 1 m below the water table in the 59
piezometer investigated; in black colour the four selected piezometers along a flow line
(dashed line corresponding to transect AA’ in Fig. 2.2).
Figure 4.29 - Distribution of NO3- and NH4+ concentrations for the 59 monitoring wells with
respect to redox values (Eh).
Figure 4.30 - Depth profiles of sediment TOC, TLN and NH4+; stratigraphy (core logs data) is
plotted on each graph to facilitate the interpretation.
Figure 4.31 - Transect along the selected flow line (refer to figure 4.28 for the location) with
concentration profiles of NH4+, NO3- and NO2-; unconfined aquifer thickness (shaded grey
area), water table (blue line) and core logs stratigraphy are also shown. Note that curves are
plotted at different horizontal scale to facilitate the graph interpretation.
Figure 4.32 - Transect along the selected flow line (refer to figure 4.28 for the location) with
concentrations profiles of NH4+ and TDS; unconfined aquifer thickness (shaded grey area),
water table (blue line) and core logs data are also shown. Note that curves are plotted at
different horizontal scale to facilitate the graph interpretation.
vi
LIST OF FIGURES
Figure 4.33 - Salinity versus Sr2+ concentration in deep (from -25 to -9 m a.s.l.), medium (from
-9 to -7 m a.s.l.), and shallow (from -7 to +1 m a.s.l.) samples (left graph); grey bars represent
salinity and Sr2+ concentration of the Adriatic seawater. NH4+ versus Sr2+ concentration in
deep (from -25 to -9 m a.s.l.), medium (from -9 to -7 m a.s.l.), and shallow (from -7 to +1 m
a.s.l.) samples (right graph); grey bars represent NH4+ and Sr2+ concentration of the Adriatic
seawater.
Figure 4.34 - Scatter plots of selected TE in sediments versus clay content and their relative
regression coefficients (R2) for linear fitting.
Figure 4.35 - Depth profiles of TDS, Eh and selected redox species in groundwater and
sediments; stratigraphy (core logs data) is reported on each plot to facilitate the
interpretation.
Figure 4.36 - Depth profiles of TE in groundwater and sediments; stratigraphy (core logs data)
is reported on each plot to facilitate the interpretation.
vii
LIST OF TABLES
LIST OF TABLES
Table 3.1 - Flow and transport parameters used for the simulations.
Table 4.1 - Core sample analyses of monitoring wells P33 and P34.
Table 4.2 - Values of measured total porosity (n), calculated total porosity by Vukovic and
Soro (1992) (nv), respective effective porosity (ne, nev) and pore-water velocity (v and vv), and
measured hydraulic conductivity (k) in monitoring wells P33 and P34.
Table 4.3 - Discrepancy between MLS and IDS values concerning the major ions, the symbol –
is used when concentrations are below detection limit.
Table 4.4 - Discrepancy between MLS and IDS values concerning the physical-chemical
parameters.
Table 4.5 - Comparison between the observed and calculated volume of water (m 3/s)
discharged to the sea by the 5 main dewatering stations.
Table 4.6 - Hydraulic heads composite scale sensitivity values calculated by PEST.
Table 4.7 - Absolute mean error between observed and calculated values of the calibrated and
validated models.
Table 4.8 - Scenario modelling of the predicted variations of freshwater maximum thickness
respect to the actual situation, with sub-scenarios of projected low sea level rise (LSLR) and
high sea level rise (HSLR). Positive values indicate thinning of freshwater lens, while negative
values indicate thickening.
Table 4.9 - Chemical composition of samples from P33 and P34. Depth are expressed as
relative to mean sea level.
Table 4.10 – Derived parameter values in the sediment samples collected from P33 and P34
(in italics). SOM and clay content reported from Mastrocicco et al. (2012). Depth are
expressed as relative to mean sea level.
Table 4.11 - TE concentrations in groundwater samples from P33 and P34. Depth are
expressed as relative to mean sea level.
viii
RIASSUNTO
RIASSUNTO
Contesto e motivazione
Le risorse idriche sotterranee delle aree costiere sono sottoposte ad una crescente attenzione
come fonte di approvvigionamento idrica grazie alla loro minor vulnerabilità
all'inquinamento rispetto alle acque superficiali ed alla elevata capacità di stoccaggio degli
acquiferi (Dillon, 2005). Tuttavia, in assenza di sistemi di gestione sostenibile delle risorse
idriche, un incontrollato uso del suolo ed il sovrasfruttamento delle falde, possono portare ad
un deterioramento significativo e di lunga durata delle risorse idriche e degli ecosistemi da
esse dipendenti (Candela et al., 2009; Grassi et al., 2007; Park et al., 2005).
Per queste ragioni, negli ultimi decenni la ricerca sugli acquiferi costiere si è intensificata
soprattutto per quanto riguarda i processi di salinizzazione (Barlow e Reichard, 2010;
Custodio, 2010; Post e Abarca, 2010; Werner et al., 2013). Tuttavia, l’individuazione delle
sorgenti di salinizzazione delle falde costiere non è semplice in quanto essa può derivare sia
da fonti naturali che antropiche e, in molti casi, più sorgenti possono agire
contemporaneamente in uno stesso acquifero (Cartwright et al., 2004; Fakir et al., 2002;
Lenahan e Bristow, 2010; Sanchez Martos et al., 2002). In alcune zone costiere, la questione
della salinizzazione delle acque sotterranee vede il coinvolgimento non solo di acqua marina
attuale ma anche di acque di mare relitte e acque ipersaline (Fass et al., 2007; Ridd e Stieglitz,
2002; Wood et al., 2003). In questi casi, la distinzione tra intrusione attuale e relitta è
fondamentale per costruire un modello concettuale robusto con il quale definire un
programma di gestione sostenibile (Werner e Gallagher, 2006). Qualunque sia la sorgente
salina, la sua identificazione è spesso complicata dalle interazioni suolo-acqua, per altro
amplificate dall’aumento della forza ionica dell'acqua (de Montety et al., 2008), e dai processi
di miscelazione spesso accelerati dalle attività antropiche come l'eccessivo pompaggio o
l’escavazione di canali artificiali (Barlow e Reichard, 2010; Custodio, 2010; Kass et al., 2005;
Yamanaka e Kumagai, 2006).
Un tema chiave per comprendere i processi di salinizzazione è quello della corretta
caratterizzazione verticale delle proprietà fisico-chimiche degli acquiferi (de Louw et al.,
2010; 2011; Netzer et al., 2011). La variabilità nella composizione chimica verticale delle
acque sotterranee è stata descritta sia tra diverse unità idrogeologiche della medesima falda
(Nativ e Weisbrod, 1994) che su piccole sezioni di una data unità idrologica (Dhar et al.,
2008). Gli acquiferi costieri sono spesso costituiti da sequenze di strati con differenti
proprietà idrauliche (Bear et al., 1999), che possono complicare l’andamento dell'interfaccia
acqua dolce/acqua salata (Kim et al., 2008) e quindi i pozzi di monitoraggio delle zone
costiere possono agire come cortocircuiti verticali fra strati (Werner et al., 2013) inducendo
un flusso verticale locale (Elci et al., 2003) che può causare il trasporto dei soluti (Konikow e
Hornberger, 2006) compromettendo le misure concernenti la distribuzione della salinità
(Shalev et al ., 2009).
ix
RIASSUNTO
Le diagrafie (OBL) sono spesso impiegate per individuare l‘interfacce acqua dolce/acqua
salata negli acquiferi costieri e per caratterizzarne le zone redox (Jorstad et al., 2004; Kim et
al., 2008). Tuttavia, anche se questo metodo di monitoraggio è semplice, economico e veloce,
la tecnica OBL ha alcune limitazioni, in particolare nelle zone costiere dove i valori misurati
(di salinità o di qualsiasi altra proprietà dell'acqua) potrebbero essere rappresentativi della
colonna d'acqua stratificatasi all’interno del pozzo, ma non necessariamente del mezzo poroso
circostante (Balugani e Antonellini, 2010; Kurtzman et al., 2011; Shalev et al., 2009).
Allo stesso modo, il campionamento integrato (IDS) fornisce campioni d'acqua che
rappresentano un'integrazione indefinita della qualità delle acque sotterranee per l'intera
lunghezza del pozzo (Robbins e Hayden-Martin, 1991) ed è quindi in gran parte carente ai fini
dello studio dei processi di salinizzazione, dal momento che in un acquifero costiero la
stratificazione delle acque in funzione della loro densità è la regola piuttosto che l'eccezione.
Al contrario, il campionamento multilivello (MLS) della falda acquifera è in grado di fornire
una distribuzione tridimensionale della qualità delle acque sotterranee all'interno di un
sistema acquifero costiero (Mastrocicco et al., 2012c). La tecnica MLS è abitualmente applicata
in siti contaminati per mappare la diffusione dei contaminanti e quantificare i percorsi di
reazione biogeochimici (Colombani et al., 2009; Henderson et al., 2009; Mastrocicco et al.,
2012b; Prommer et al., 2006), ma è meno frequentemente utilizzata nel monitoraggio della
salinizzazione nelle zone costiere a causa dei costi elevati e dei tempi lunghi (Einarson e
Cherry, 2002). La spesa, tuttavia, non cancella la necessità di una caratterizzazione in funzione
della profondità per stimare correttamente la composizione chimica delle acque sotterranee e
migliorare quindi la comprensione dei processi che agisco nelle zone costiere.
La discretizzazione verticale negli acquiferi è fondamentale non solo per comprendere i
processi di salinizzazione ma anche nello studio della distribuzione e la speciazione di alcuni
composti maggiori ed in tracce che, in seguito all’aumentata della forza ionica delle acque
sotterranee, potrebbero essere rimobilizzati. Dalla diffusione di contaminanti geogenici o di
origine antropica disciolti nelle acque sotterranee, derivano potenziali rischi non solo per le
acque superficiali (Li et al., 2014) ma anche per le colture (Li et al., 2010). Ad esempio, un
aumento delle concentrazioni di azoto reattivo (N) può essere responsabile di pericoli per la
salute (Gelberg et al., 1999; Guliš et al., 2002), dell'eutrofizzazione delle acque superficiali
(Jayasingha et al., 2011) e della crescita di alghe nelle acque marino costiere con conseguenti
danni non solo ambientali ma anche alle attività turistiche (Linderfelt e Turner, 2001).
L’aumento del contenuto di N nelle acque superficiali e sotterranee è stato comunemente
attribuito a fonti antropiche, quali la presenza di scarichi civili o di attività agricole e/o
industriali (Rodvang e Simpkins, 2001). Minore attenzione è stata data alle sorgenti naturali
di N nelle acque sotterranee, quali ad esempio un arricchimento dovuto al contributo di
processi di interazione acqua/sedimento, la mineralizzazione della sostanza organica, (LoriteHerrera et al., 2009) e il rilascio di NH4+ per scambio cationico in condizioni anossiche
(Appelo e Postma, 2005; Lewandowski e Nützmann, 2010; Seitzinger et al., 1991).
Un altro argomento in cui la discretizzazione verticale è inevitabile è lo studio della
distribuzione degli elementi in traccia (TE) nelle acque sotterranee con salinità variabile e nei
sedimenti associati. I TE posso derivare da fonti antropiche quali attività agricole, industriali e
x
RIASSUNTO
minerarie (Buggy e Tobin, 2008; Du Laing et al., 2009), ma possono anche essere legati alla
naturale composizione delle formazioni da cui derivano i sedimenti erosi o alle reazioni di
scambio promosse dall'intrusione di acque saline (Teuchies et al., 2013). La contaminazione
da metalli geogenici è infatti possibile soprattutto nelle zone costiere dove la forza ionica delle
acque sotterranee può essere molto elevata, favorendo così la consistente mobilitazione di TE
(Acosta et al., 2011; Wang et al., 2012; Cidu et al., 2013; Delaune et al., 2008. Dévai et al., 2005;
Du Laing et al., 2008).
Un caso particolare è rappresentato dalle zone costiere che si trovano sotto il livello del mare,
che in futuro, con l'aumento del livello del mare previsto dai cambiamenti climatici (Werner et
al., 2013), potrebbero sperimentare un aumento della risalita delle acque saline presenti alle
base degli acquiferi, grazie ai flussi verticali indotti dalle attività di bonifica indispensabili per
mantenere asciutti i terreni delle pianure alluvionali utilizzati per l'agricoltura (Louw et al.,
2013; Giambastiani et al., 2013; Mollema et al., 2013; Hinsby et al., 2008; Vandecasteele et al.,
2006).
Con il recente perfezionamento delle proiezioni dei cambiamenti climatici, è aumentato il
numero degli studi che valutano gli effetti delle crescenti pressioni sugli acquiferi costiere
(Ranjan et al., 2006; Werner et al., 2013). Diversi studi hanno cercato di quantificare l'effetto
dell’innalzamento del livello marino (SLR) sull’intrusione di acque marine in acquiferi costieri.
Considerando l'Europa, la vulnerabilità delle aree costiere al SLR è stata indagata da Lebbe et
al. (2008) e da Vandenbohede et al. (2008) per la pianura costiera del Belgio, da Oude Essink
(2010) per i Paesi Bassi, da Feseker (2007) per il Nord-ovest della Germania, da Fujinawa et
al. (2009) per la regione orientale della Turchia, da Giambastiani et al. (2007) e da Colombani
et al. (2015a) per le falde acquifere costiere di Ravenna e Ferrara. La maggior parte di questi
studi sottolineano la salinizzazione dell’acquifero costiero spesso mette a rischio anche le
risorse idriche superficiali. Recenti studi hanno esaminato le interazioni quantitative a grande
scala tra acque superficiali e sotterranee in prossimità della costa (Oude Essink et al., 2010;
Vandenbohede et al., 2008) ma meno è stato fatto per chiarire le interazioni a piccola scala tra
acque sotterranee e canali di irrigazione e drenaggio. Solo di recente sono stati pubblicati
alcuni studi dettagliati in cui i processi di flusso e trasporto del sale in pianure agricole
bonificate sono stati quantificati e modellati (de Louw et al., 2010; 2011; 2013).
Una volta che un modello concettuale affidabile è stato costruito, l'implementazione di modelli
numerici di flusso e trasporto densità-dipendenti, come SEAWAT (Guo e Langevin, 2002;
Langevin et al., 2007) e MODFLOW (Harbaugh et al., 2005; McDonald e Harbaugh, 1988),
costituisce un valido approccio per capire: (i) i meccanismi di salinizzazione (da intrusione di
acqua di mare attuale e per risalita di acque saline relitte), (ii) il loro effetto sulla
mobilitazione di contaminanti geogenici e (iii) gli impatti che i diversi scenari di cambiamenti
climatici avranno su questi fenomeni. L’utilizzo di tali strumenti, vincolati ai rilievi di campo,
consentirà migliorare la gestione dell’interazione tra acque superficiali e sotterranee e di
individuare eventuali misure di mitigazione degli impatti per evitare il deterioramento delle
risorse idriche.
xi
RIASSUNTO
Significatività dell’area di studio prescelta
Le aree soggette a deterioramento delle acque sotterranee sono prevalentemente pianure
costiere densamente popolate, che più di altre sono esposte alle perturbazioni antropiche e
alle conseguenze dei cambiamenti climatici. Tra i problemi che queste aree devono affrontare,
la salinizzazione delle risorse sia di superficie che sotterranee sembra essere il più
preoccupante, soprattutto in pianure intensamente coltivate.
Per questo motivo, l‘acquifero costiero non confinato della Provincia di Ferrara, situato nella
pianura del delta del fiume Po, è stato scelto come area di studio per esaminare i meccanismi
che regolano il processo di salinizzazione delle acque sotterranee.
Questo territorio è stato ampiamente studiato negli ultimi decenni e un gran numero di
contributi provenienti da discipline diverse permettono di ottenere una solida conoscenza
dell'evoluzione geologica e geomorfologica della falda costiera della Provincia di Ferrara.
D'altra parte, essendo le acque sotterranee poco sfruttate a causa delle abbondanti risorse
idriche superficiali, gli studi sulle falde acquifere locali sono limitati.
Per migliorare la comprensione degli aspetti idrologici e idrochimici dell'acquifero costiero
sono stati utilizzati, 11 dei 30 pozzi di monitoraggio delle rete del Servizio Geologico della
Regione Emilia Romagna insieme con 38 pozzi di monitoraggio pre-esistenti della Provincia di
Ferrara.
Oltre al buon livello di dettaglio con cui questa zona è stato indagata, altri fattori la rendono
ideale per condurre lo studio oggetto di questa tesi: (i) la zona costiera di Ferrara è
intensamente coltivata sin dalla sua bonifica più recente (metà del 1900), (ii) ancora oggi il
sistema di drenaggio è necessario per abbassare la falda e garantire la coltivabilità dei terreni,
(iii) l'irrigidimento della rete idrografica causata dalla bonifica ha aumentato la subsidenza
del terreno, già naturalmente presente, (iv) l'aumento della differenza tra la quota del piano
campagna ed il livello del mare ha favorito l’ingressione marina nella rete idrografica, (v) la
salinizzazione del suolo danneggia la produttività agricola, (vi) la diminuzione della
disponibilità delle risorse idriche, specialmente nel periodo estivo, rappresenta un rischio per
lo sviluppo delle attività turistiche.
L'analisi dei risultati aiuterà le autorità locali a gestire correttamente le risorse idriche
costiere e permetterà di individuare possibili azioni di mitigazione del fenomeno della
salinizzazione delle acque sotterranee e superficiali.
Va comunque sottolineato che lo studio qui presentato si è focalizzato esclusivamente sugli
aspetti fisico-chimici del problema della salinizzazione senza tener conto degli aspetti
economici di tale problematica, che sono altrettanto importanti per l'identificazione e
l'attuazione delle azioni di mitigazione. Tuttavia, il protocollo di caratterizzazione impiegato
in questo studio potrebbe essere applicato con successo in molte altre zone costiere con
problemi analoghi a quelli descritti per l’area costiera ferrarese.
xii
RIASSUNTO
Obiettivi della tesi
A causa dell'impatto potenzialmente irreversibile del deterioramento della qualità delle
risorse idriche , l'obiettivo primario di questa tesi è stato quello di ottenere una comprensione
quantitativa dei processi che hanno causato la salinizzazione delle acque sotterranee nella
falda acquifera costiera ferrarese. In secondo luogo si è voluto indagare l’impatto che i
cambiamenti climatici potranno avere sul processo di salinizzazione in atto e le eventuali
conseguenze che un aumento di salinità possono avere sulla mobilizzazione di contaminanti
sia maggiori che in tracce. Alla luce di quanto detto sin qui, lo studio in oggetto si propone di
raggiungere i seguenti obiettivi:
Caratterizzazione della falda acquifera costiera della pianura di Ferrara: idrologia,
idrochimica ed evoluzione del sistema
La caratterizzazione idrodinamica e idrogeochimica della falda acquifera costiera ferrarese,
soggetta a profondi cambiamenti naturali e antropici dall'ultima glaciazione ad oggi, ha
permesso di comprendere le cause che hanno portato alla salinizzazione delle acque
sotterranee (vedi capitolo 4.1).
Le acque saline e ipersaline sono per lo più associate a sedimenti fini di retro-barriera e
ambienti lagunari, legati alla trasgressione Flandriana, mentre le acque dolci e salmastre sono
legate a sistemi di dune e spiagge relitte. Le acque ipersaline si trovano nella parte più
profonda della falda acquifera e non possono essere associate all’intrusione marina attuale ma
testimoniano piuttosto la presenza di acque marine relitte intrappolate in ambienti retrodunali e lagunari durante le fasi interglaciali dell’Olocene. Dati di campo mostrano come gli
elevati tenori di salinità (superiori all’acqua di mare) possano essere attribuiti
all’evaporazione diretta e/o alla traspirazione di acqua di mare prima della sua infiltrazione.
La parametrizzazione idrodinamica dell’acquifero ha permesso di individuare la presenza di
un gradiente idraulico verticale dovuto al drenaggio delle rete di canali di bonifica presenti
nell’area. Di conseguenza le acque saline precedentemente intrappolate nella parte più
profonda dell‘acquifero da lenti a bassa permeabilità, sono state trasportate verso la parte
superficiale della falda e addirittura scaricate nella rete di canali di irrigazione.
L'indagine idrogeochimica ha anche permesso di distinguere tra il ruolo della miscelazione tra
acque a diversa salinità e quello dell’interazione acqua/sedimento (scambio ionico,
dissoluzione/precipitazione di calcite e dolomite, ossidazione della materia organica,
riduzione dei solfati) nell'evoluzione delle acque sotterranee. Le acque più saline infatti sono
caratterizzate da scambio cationico e riduzione di SO42-, soprattutto dove è presente materia
organica fossile (depositi paleo-palustri e strati di torba). L'insieme di tali processi comporta
poi l'aumento del contenuto di Ca2+ e HCO 3- nelle acque sotterranee con conseguente
precipitazione di dolomite/calcite.
xiii
RIASSUNTO
Importanza di tecniche di monitoraggio nella comprensione dei processi di salinizzazione
L'esperienza in campo ha dimostrato che una dettagliata caratterizzazione verticale è l'unico
metodo che può restituire una ricostruzione affidabile della distribuzione delle specie
disciolte all'interno della falda freatica. Per questo motivo sono state applicate diverse
tecniche di monitoraggio per acquisire i principali parametri fisico-chimici in funzione della
profondità: diagrafie (OBL), campionamento integrato (IDS) e campionamento multilivello
(MLS). I risultati derivanti dall’applicazione delle diverse tecniche di monitoraggio sono stati
quindi confrontati per individuare la metodologia più adatta ed efficace (vedi capitolo 4.2).
Il confronto tra la tecnica OBL e la tecnica MLS mostra che, escludendo la distribuzione delle
acque saline e ipersaline alla base dell’acquifero, le due metodologie non presentano
differenze significative. Ne consegue che il metodo OBL può essere utilizzato con successo per
una determinazione preliminare della posizione dell’interfaccia acqua dolce/acqua salata.
Tuttavia, il metodo OBL ha fornito una caratterizzazione fuorviante della distribuzione degli
ambienti redox all'interno della falda poiché la distribuzione dei dati idrochimici (pH, Eh e O2)
raccolti con la tecnica OBL non concorda con quella della tecnica MLS. Ne consegue che uno
studio più dettagliato sui processi geochimici reattivi può essere intrapreso solo attraverso la
tecnica MLS, che è comunque preferibile anche quando è richiesto un elevato livello di
precisione nel descrivere la distribuzione delle acque ad elevata salinità.
Un confronto diretto tra l’affidabilità delle tecniche di monitoraggio IDS e MLS è stato
effettuato derivando due diversi modelli concettuali della distribuzione della salinità sulla
base dei monitoraggi effettuati con l’una e con l’altra tecnica.
Il disaccordo nell'interpretazione dei risultati ottenuti con le due tecniche di campionamento
è particolarmente evidente nell'analisi del TDS rispetto alla profondità, e del Eh rispetto al pH.
Il modello concettuale IDS richiama il fenomeno dell’intrusione marina attuale, poiché le
concentrazioni di Cl- non sono mai superiori a quella dell'acqua di mare e per l'assenza di
acque ipersaline in tutta l'area di studio. Al contrario, nel modello concettuale MLS la parte
inferiore della falda freatica presenta acque ipersaline rendendo necessario considerare
processi di salinizzazione diversi dall’intrusione marina attuale, come l'evoluzione geochimica
delle acque sotterranee causata dall’evapoconcentrazione o l’interazione acqua/sedimenti.
Le discrepanze tra il monitoraggio effettuato via MLS e IDS risultano particolarmente evidenti
specialmente nel caso di specie disciolte sensibili agli ambienti redox, confermando che nella
tecnica IDS la miscelazione artificiale all’interno del pozzo può maschere la stratificazione
verticale delle concentrazioni nella falda acquifera.
I risultati dimostrano non solo che la tecnica IDS è decisamente più imprecisa rispetto alla
tecnica MLS, ma addirittura che essa potrebbe dare rappresentazioni fuorvianti della realtà.
Pertanto, i modelli concettuali ottenute tramite la tecnica IDS non possono neppure essere
considerati uno studio preliminare in un acquifero costiero affetto da salinizzazione ma
dovrebbero essere evitati con fermezza, soprattutto quando il contrasto di densità tra acque
dolci e saline è particolarmente elevato.
xiv
RIASSUNTO
Previsione dei trend di salinizzazione della falda acquifera costiera
Attraverso l'utilizzo della modellazione numerica, basata sui rilievi di campagna e sul
conseguente modello concettuale, è stato possibile ricostruire e confermare l'ipotesi
sull'origine di salinizzazione (vedi capitolo 4.3). La salinizzazione della falda dovuta al
richiamo verso l'alto di acque ipersaline relitte ed il conseguente deterioramento della qualità
anche delle acque superficiali, sono stati simulati tramite l’implementazione e la calibrazione
di un modello di flusso tridimensionale (MODFLOW-2000) accoppiato ad un modello di
trasporto non reattivo a densità variabile (SEAWAT).
Le simulazioni hanno chiaramente indicato che il principale driver del processo di
salinizzazione è l’interazioni tra la falda freatica costiera e la rete di canali che richiamano
verso l’alto le acque salate relitte. Questo processo, che è il risultato di un drenaggio avvenuto
per secoli e che continua ancora oggi, influisce sulla qualità delle acque superficiali mettendo
a rischio il loro futuro utilizzo a scopo agricolo e sottoponendo ad una crescente pressione gli
ecosistemi costieri.
Per comprendere meglio l’evoluzione di tale fenomeno in un prossimo futuro, quando anche
gli effetti dei cambiamenti climatici contribuiranno ad alterare i labili equilibri che governano
gli scambi idrici tra acque sotterranee ed acque superficiali nella bassa pianura ferrarese, è
stato implementato un modello numerico bidimensionale che tenesse conto di: (i) aumento
dell’evapotraspirazione indotto dall’aumento della temperatura; (ii) aumento della frequenza
di eventi di pioggia estremi; (iii) comparsa di condizioni di estrema siccità; (iv) e aumento del
livello medio marino.
I risultati mostrano che in primavera e estate i canali di irrigazione ricaricano la falda
diminuendo la falda salinità superficiale, anche nel caso di un aumento della temperatura (e
quindi dell’evapotraspirazione). Al contrario, durante la stagione fredda i canali non sono più
utilizzati per l'irrigazione e, rimanendo vuoti, portando ad una inversione di flusso che
provoca un aumento della salinità nella falda superficiale dovuta al movimento verso l'alto
dell’acqua ipersalina contenuta nei sedimenti di prodelta. Questo flusso verso l'alto salinizza
lentamente i corpi d’acqua superficiali connessi con la falda acquifera tanto più quanto più è
elevata la differenza di carico idraulico tra i due corpi idrici. Ne deriva che il previsto aumento
de livello del mare influirà negativamente sui processi di salinizzazione, molto più di quanto
non facciano l’aumento della temperatura o degli eventi di precipitazione estremi.
L’influenza della salinizzazione sui trend di concentrazione dell’ammonio e degli elementi
nell’acquifero costiero
Dato che i cambiamenti climatici possono aumentare il tasso di salinizzazione della falda, per
valutare se il cambiamento nelle attuali condizioni chimico-fisiche e idrodinamiche può anche
indurre la mobilizzazione di elementi geogenici, sono stati studiati il contributo
dell’interazione acqua/sedimento e l’influenza antropica sulla qualità delle acque sotterranee,
xv
RIASSUNTO
focalizzando l'attenzione sull’azoto (vedi capitolo 4.4) e sugli elementi in traccia (vedi capitolo
4.5).
Per capire se i processi di salinizzazione possono svolgere un ruolo nella mobilitazione delle
specie azotate, è stata condotta una caratterizzazione dettagliata riguardante la distribuzione
spaziale e le possibili fonti di nitrati (NO3-) e ammonio (NH4+) nell’acquifero costiero
ferrarese. L’NH4+ è la specie di azoto inorganico prevalente nelle acque sotterranee. Mentre
NO3- è presente nella porzione superiore dell’acquifero dove esistono condizioni ossidanti, la
concentrazione di NH4+ aumenta con la profondità e la salinità. Concentrazioni molto elevate
di NH4+ si trovano in coincidenza con sedimenti torbosi presenti alla base dell’acquifero dove
sussistono condizioni anossiche e nell’aquitard sottostante. I risultati di questo studio
mostrano che l’NH4+ deriva principalmente dalla mineralizzazione della sostanza organica
presente nei sedimenti fini depositati in ambienti paleo-palustri.
L'effetto dell’elevata salinità sulla mobilità degli elementi in traccia (TE), è stato studiato
mediante l’analisi sia di campioni di falda sia di carote di sedimento, a varie profondità. La
concentrazione totale di TE è maggiore nei sedimenti fini, ed in particolare nella frazione
argillosa, mentre le concentrazioni di TE in falda aumentano con la forza ionica della
soluzione e cioè con la profondità dal momento che la parte basale dell’acquifero presenta
acque a maggior tenore di salinità. L'elevato contenuto di materia organica impone condizioni
riducenti lungo tutto il profilo falda, cosa che promuove ulteriormente la mobilizzazione di
TE. Le uniche eccezioni sono V e As, probabilmente rilasciati dalla lisciviazione dei fertilizzanti
nella parte superiore della falda e da strati torbosi sepolti nella parte inferiore della falda
acquifera. Pertanto, i risultati di questo studio mettono in evidenza la necessità di distinguere
tra il rilascio di TE a causa dell‘alterazione dell'equilibrio acqua/suolo indotta dalla
salinizzazione delle acque sotterranee ed il rilascio di TE in falda a causa di attività
antropiche.
Implicazioni e prospettive
Il presente studio ha impiegato diverse discipline delle Scienze della Terra per indagare i
fenomeni di salinizzazione dell’acquifero costiero ferrarese. Nessuna delle metodologie
impiegate sarebbe stata in grado di garantire la completa comprensione del sistema
idrogeologico in esame se applicata separatamente.
Discriminare tra processi di salinizzazione passati e attuali è spesso difficile, se non
impossibile, soprattutto quando sono disponibili solamente i dati di salinità. La presenza di
acque ipersaline nella falda acquifera costiera ferrarese dimostra che l'attuale situazione
idrogeologica testimonia la presenza di acque relitte originatesi in un ambiente totalmente
diverso da quello odierno. Ne deriva che l'interpretazione delle osservazioni attuali deve
tener conto che la qualità dell'acqua continua ad evolversi dal passato, adattandosi alle
condizioni al contorno contemporanee. La conoscenza della storia geologica della zona
oggetto di indagine è quindi essenziale per riconoscere il potenziale contributo di processi
idrologici passati alla situazione attuale.
xvi
RIASSUNTO
Inoltre, la scelta di un metodo di monitoraggio e campionamento affidabile è risultata di vitale
importanza per la corretta identificazione delle proprietà fisiche dell’acquifero e della
distribuzione della qualità delle acque sotterranee. Contare su appropriate tecniche di
monitoraggio, come ad esempio il campionamento multilivello, è quindi importante tanto
quanto selezione le giuste tecniche di analisi e di modellazione. Il campionamento multilivello
non richiede competenze altamente specializzate o attrezzature particolarmente costose,
tuttavia ancora oggi fatica ad imporsi negli studi sui processi di salinizzazione.
Questo studio, dimostrando in modo tangibile, come gravemente fuorviante potrebbe essere
un modello concettuale più semplice, costruito sulla base di tecniche di monitoraggio
inaffidabili, vuole essere un monito per i gestori delle risorse idriche che devono fare scelte
difficili per la pianificazione della futura gestione e protezione delle acque sotterranee.
Infine, anche se i dati di campo sono indispensabili per comprendere il sistema idrogeologico
in esame, i modelli numerici possono fornire un valido aiuto per testare le ipotesi del modello
concettuale. Con il giusto approccio, l'uso congiunto di attività di campo e simulazioni può
portare ad una migliore comprensione dei fenomeni ed a nuove intuizioni. Ad esempio, gli
scenari numerici permettono di quantificare i possibili effetti dei cambiamenti climatici sulla
salinizzazione delle acque sotterranee e le sue conseguenze sulla qualità delle risorse idriche,
e potrebbero essere utili per trovare strategie di mitigazione per il futuro.
xvii
INTRODUCTION
1.
INTRODUCTION
1.1 Background and motivation
In coastal areas groundwater has gained increasing attention as a source of water supply,
owing to its relatively low vulnerability to pollution in comparison to surface water as well as
its relatively large storage capacity (Dillon, 2005). However, in the absence of sustainable
water resource management schemes, uncontrolled land-use activities and over-exploitations
can lead to significant and long-lasting deteriorations of water resources and groundwaterdependent ecosystems (Candela et al., 2009; Grassi et al., 2007; Park et al., 2005). For these
reasons, in the last decades the research on coastal aquifers has intensified especially with
regards to salinization processes (Barlow and Reichard, 2010; Custodio, 2010; Post and
Abarca, 2010; Werner et al., 2013). In these areas the saltwater intrusion is a well-known and
studied problem, like the Polders in the Netherlands (Oude Essink et al., 2010) or in the southeastern Florida (Langevin and Zygnerski, 2013) and the Po river lowland (Giambastiani et al.,
2007).
The recognition of groundwater salinization origin in coastal aquifers is not straightforward.
Salinization into coastal aquifers may occur from both natural and anthropogenic sources and
in many cases salinity problems arise from more than one salt source (Cartwright et al., 2004;
Fakir et al., 2002; Lenahan and Bristow, 2010; Sanchez Martos et al., 2002). In some coastal
area, the issue of saline groundwater origins involves, beyond actual seawater, relic seawater
and hyper-saline waters (Fass et al., 2007; Ridd and Stieglitz, 2002; Wood et al., 2003). In
these cases, it is vital to distinguish between modern and ancient saltwater intrusion to build
a robust conceptual model of groundwater salinization through which a sustainable
management program can be designed (Werner and Gallagher, 2006). The most relevant salt
sources/processes recognized worldwide are: evaporation, evaporite leaching, mobilization
of salts stored in the unsaturated zone, infiltration of non-marine polluted surface waters,
slow-moving saline/salt waters of marine origin, highly mineralised waters from geothermal
fields (Fidelibus et al., 2011), sea spray, hyper-filtration, agricultural practices (return flow,
use of fertilizers and irrigation with treated wastewater), cycling wetting and drying.
Whatever the salt source, its identification is often puzzled by soil-water interactions
amplified by the increased water ionic strength (de Montety et al., 2008) or by mixing
processes enhanced by human activities as over-exploitation, land reclamation and
construction of navigation channels along rivers or artificial canals (Barlow and Reichard,
2010; Custodio, 2010; Kass et al., 2005; Yamanaka and Kumagai, 2006).
A key issue to understand salinization processes is the correct characterization of vertical
variability of aquifer properties governing groundwater quality (de Louw et al., 2010; 2011;
Netzer et al., 2011). A variability in vertical chemical composition of groundwater has been
described both among different hydrological units of the same aquifer (Nativ and Weisbrod,
1994) and over small sections of a given hydrological unit (Dhar et al., 2008). Coastal aquifers
often consist of layered sequences with varying hydraulic properties (Bear et al., 1999) that
1
INTRODUCTION
can complicate monitoring of fresh–saltwater interface because of heterogeneity of the
aquifer matrix (Kim et al., 2008). Therefore, coastal monitoring wells may act as vertical short
circuits between layers (Werner et al., 2013), inducing local vertical flow due to connection
between higher and lower hydraulic head zones (Elci et al., 2003). A common consequence is
that ambient flow within long-screen wells may cause solute transport (Konikow and
Hornberger, 2006), even in homogeneous aquifers with small vertical hydraulic gradients
(Britt, 2005) and it may also compromise hydraulic head measurements (Shalev et al., 2009).
Open borehole logging (OBL) is often employed to identify fresh-saltwater interfaces in
coastal aquifers (Kim et al., 2008) and to characterize the redox zones within contaminant
plumes (Jorstad et al., 2004). Even if this method of obtaining depth dependent groundwater
quality data is plain, inexpensive and fast, the OBL technique has some limitations, most
notably in coastal zones where measured values (salinity or any other water property) may
be well representative of the stratified water column accumulated within the monitoring well,
but not necessarily of the surrounding porous media (Balugani and Antonellini, 2010;
Kurtzman et al., 2011; Shalev et al., 2009). Similarly, Integrated Depth Sampling (IDS)
provides water samples representing an undefined integration of groundwater quality over
the entire screen length (Robbins and Hayden-Martin, 1991): thus, it is largely deficient for
the purposes of understanding salinization processes, since in coastal aquifer groundwater
stratification is the rule rather than the exception.
On the contrary, Multi-Level Sampling (MLS) is an invaluable method that, by sampling at
multiple intervals of an aquifer, allows the reconstruction of three-dimensional distribution of
groundwater quality within a coastal aquifer system (Mastrocicco et al., 2012c). MLS is
routinely applied in contaminated sites to map contaminant spreading and to quantify
biogeochemical reaction pathways (Colombani et al., 2009; Henderson et al., 2009;
Mastrocicco et al., 2011b; Prommer et al., 2006), but it is less frequently used in monitoring
salinization in coastal areas because of high costs and time involved (Einarson and Cherry,
2002). The costs, however, cannot obscure the need for a dedicated depth-dependent
groundwater monitoring, which is necessary to properly estimate the chemical composition
of groundwater in coastal areas and to enhance the predictability of regional scale
management models that will inevitably be required in the near future.
Depth-dependent aquifer characterization is essential to indentify vertical flow, which, beside
salinizing the whole aquifer body, could enhance the mobility of geogenic and anthropogenic
contaminants dissolved in groundwater, posing risk to surface water (Li et al., 2014) and to
crop root uptake (Li et al., 2010). For instance, high-nitrogen loading in surface and ground
waters is getting an increasing attention in order to fulfil the water quality objectives required
by the Water Framework Directive (Directive 2000/60/EC, 2000). Nitrogen (N) enrichment is
believed to be responsible for health hazards (Gelberg et al., 1999; Gulis et al., 2002) and
deterioration of water quality, such as eutrophication (Jayasingha et al., 2011) and algal
growth (Linderfelt and Turner, 2001). Increasing N content in surface and ground waters has
been commonly explained evoking anthropogenic sources such as agricultural and industrial
activities and sewage wastes (Rodvang and Simpkins, 2001).
2
INTRODUCTION
Many studies have focused on NO3- groundwater contamination in agricultural areas (Bohlke
et al., 2002; Mastrocicco et al., 2011a; Puckett et al., 2002) or in presence of anthropogenic
sources (Aravena and Robertson, 1998); other studies have reported ammonium (NH 4+)
contamination in groundwater due to rich wastewater spillage or landfill leachate plumes
(Brun et al., 2002; Christensen et al., 2001). Less attention has been given to natural sources
of N in ground waters. Thus, identification of potential sources of N and assessment of its
speciation in aquifer is still needed. Elevated NH4+ contents of natural origin have been
reported in aquifers in different parts of the world: U.S. (Glessner and Roy, 2009; Hinkle et al.,
2007), Mexico (Ortega-Guerrero, 2003), Asia (Jang and Liu, 2005; Jiao et al., 2010), Australia
(Linderfelt and Turner, 2001) and Europe (Lorite-Herrera et al., 2009). Potential natural
sources of NH4+ are mostly related to contribution of water/sediment interaction, such as
mineralization of sedimentary organic matter rich in organic N (Lorite-Herrera et al., 2009),
and release of NH4+ from sediments, governed by cation exchange capacity under anoxic
conditions (Appelo and Postma, 2005; Lewandowski and Nützmann, 2010), especially in
association with saline groundwater (Seitzinger et al., 1991).
Another topic where the vertical discretization is unavoidable is the study of trace element
(TE) distribution of in groundwater with variable salinity and their associated sediments. TE
can derive from anthropogenic sources like agricultural, industrial and mining activities
(Buggy and Tobin, 2008; Du Laing et al., 2009), but they can also be related to weathering of
host formations, or by cation or anion exchange reactions promoted by intrusion of saline
waters (Teuchies et al., 2013). Geogenic metal contamination is of upcoming concern,
especially in coastal areas where the ionic strength of groundwater can be very high, thus
promoting substantial mobilization of TE (Acosta et al., 2011; Wang et al., 2012). In fact, many
authors pointed out a potential significant threat related to the elevated trace and rare metals
concentrations found in floodplain saline soils (Cidu et al., 2013; DeLaune et al., 2008; Devai et
al., 2005; Du Laing et al., 2008). A peculiar case are the low lying areas, which represent a
possibly extended problem throughout the world in the midterm future, with the forecasted
sea level rise induced by climate changes (Werner et al., 2013). In these environments, the
land reclamation usually induces upconing of the basal saline and hyperhaline groundwater
(de Louw et al., 2013a; Giambastiani et al., 2013; Mollema et al., 2013). Thus, the ubiquitous
presence of metals obliges to accurately prepare risk assessment studies especially in
floodplains used for agriculture (Hinsby et al., 2008; Vandecasteele et al., 2006).
With the recent refinement of climate change projections, namely sea level rise (SLR) and
changes in recharge and evapotranspiration patterns, the number of studies investigating the
effects of raised pressures on coastal groundwater systems has increased (Ranjan et al., 2006;
Werner et al., 2013). The IPCC 2014 reports (2014a; 2014b) predicted that climate change
will have a range of impacts on water resources: annual runoff will decrease in southern
Europe by 0 to 23% in 2020s and by 6 to 36% in 2070s; summer low flow may decrease down
to 80% in some rivers in southern Europe; storage changes in soil moisture and recharge of
groundwater will be largely reduced in valleys and lowlands; increasing drought risk in the
Mediterranean might be amplified by water withdrawal leading to a strong competition for
available water resources and for reduction of available drinking water (García-Ruiz et al.,
3
INTRODUCTION
2011). These problems may especially affect coastal areas where large socio-economic and
environmental stresses are already observed. Several studies attempted to quantify the effect
of SLR on salt water intrusion into a coastal aquifer. Werner and Simmons (2009) identified
the major hydrogeologic controls on the impact of SLR on salt water intrusion in unconfined
coastal aquifers by using a steady-state analysis with an analytical solution. They
differentiated between flux-controlled and head-controlled systems and showed that SLR is
more problematic for head-controlled systems because inland water levels do not rise with
rising sea level. Chang et al. (2011) also found that for flux-controlled confined aquifers SLR
may not have an impact on fresh water volumes. Watson et al. (2010) investigated the
migration aspect and response time of salt water movement, concluding that in certain
condition several centuries may be required for the salt water interface to reach equilibrium
after SLR. Other studies focused on the effect of SLR on specific coastal settings. Just
considering Europe, vulnerability of low-lying coastal areas to SLR has been addressed by
Lebbe et al. (2008) and by Vandenbohede et al. (2008) for the Belgian coastal plain, by Oude
Essink et al. (2010) for the Netherlands, by Feseker (2007) for North-western Germany, by
Fujinawa et al. (2009) for the eastern Mediterranean coastal region of Turkey, by
Giambastiani et al. (2007) and by Colombani et al. (2015a) for the unconfined coastal aquifers
of Ravenna and Ferrara, Italy. Most of these studies point out that coastal aquifer salinization
often directly endanger surface water resources. Recent studies have looked at quantitative
interactions between surface and ground waters near the coast at large scales (Oude Essink et
al., 2010; Vandenbohede et al., 2008), but less has been done to clarify the small scale
interactions between groundwater and irrigation canals and drainage ditches. Only recently a
few detailed case studies have been published, in which the actual flow and salt transport
processes occurring in drained agricultural lowlands are quantified and modelled (de Louw et
al., 2010; 2011; 2013).
Once a reliable conceptual model of the stresses acting on a territory has been built, a valuable
approach to understand salinization mechanisms (both from seawater intrusion and saline
seepage), their effect on the mobilization of geogenic contaminants, and the impacts that
different climate change scenarios will have on these phenomena, is the implementation of
predictive density-dependent groundwater flow and transport numerical models, like
SEAWAT (Guo and Langevin, 2002; Langevin et al., 2007) and MODFLOW-2000 (Hill et al.,
2000); such models are based on the commonly used groundwater code, MODFLOW,
developed by the U.S. Geological Survey (Harbaugh et al., 2005; McDonald and Harbaugh,
1988). By means of such a comprehensive tool, based on field hydrodynamic and
hydrogeochemical evidences, water managers will be enabled to better manage the
surface/groundwater continuum, and to put in place prevention and/or mitigation measures
to avoid water resources deterioration.
4
INTRODUCTION
1.2 Significance of the selected test site
The areas mostly prone to groundwater deterioration are the densely populated coastal
plains, which more than other areas are exposed to anthropic perturbation and aftermath of
climatic change. Among the issues that these areas have to face, the salinization of both
surface and groundwater resources seem to be the more worrisome, especially in lowlands
intensively farmed.
For this reason, the unconfined coastal aquifer of the Ferrara Province (Northern Italy),
located in the Po River Delta lowland was selected to examine the extent of groundwater
salinization, the mechanisms governing the salinization process, the response of the
salinization rate to the expected impacts of climate change and the role of salinization on the
mobilization of geogenic contaminants.
This territory has been widely studied in the last decades, and a large number of contributes
from different disciplines allows achieving a robust knowledge of the geological and
geomorphological evolution of the coastal aquifer of the Ferrara Province.
However, being the groundwater beneath the Po River Delta poorly exploited due to the
abundant surface water resources, studies on the local aquifers are limited; thus, groundwater
resources still need to be properly characterized and known in order to better manage the
surface/groundwater continuum. To improve the understanding of the hydrological and
hydrochemical aspects of the Ferrara coastal aquifer, since 2009 the Geological Survey of the
Emilia Romagna Region developed a specific network of 30 monitoring wells as a mean of
characterizing the saltwater intrusion along the Northern Adriatic coastal aquifer; 11 of these
monitoring wells, falling within the coastal area of the Ferrara Province, together with 38 preexisting monitoring wells, have been used to carry out the field activities required to improve
the conceptual model of the study area.
Apart from the good level of detail with which this area has been investigated, other factors
make it ideal for achieving the objective of the thesis: (i) the Ferrara coastal area is intensively
farmed (Fig. 1.1) since its reclamation (late 1900), (ii) nowadays the drainage system is
necessary to lower the water table because of the agricultural requirements, (iii) the stiffening
of the hydrographic network due to reclamation activities has augmented land subsidence,
(iv) the increasing difference in the elevation of the ground level and the sea level enhances
seawater encroachment along the rivers, (v) the occurrence of present seawater intrusion and
saline relict groundwater in the deepest part of the coastal aquifer has induced groundwater
and soil salinization harming the agricultural productivity, (vi) the decrease in the availability
of freshwater resources, especially in the summer period, poses a risk in the development of
the tourist activities. All these factors make the area of the Ferrara Province ideal for
conducting the study for the thesis.
The analysis of the results will help local authorities to properly manage coastal fresh water
resources under increasing pressures, and eventually to identify reasonable countermeasures
to decrease the salt load for the future. This thesis, however, only deals with the physical
aspects of the territory and does not take into account the economic aspects, which are as
5
INTRODUCTION
important as the physical ones in the identification and implementation of sustainable
adaptation initiatives. Nevertheless, the characterization protocol employed in this study
could be successfully applied in many other similar coastal low lying areas worldwide,
suffering from the same issues that are stressing the Ferrara coastal aquifer.
Figure 1.1 – Wheat fields in the Ferrara lowland (www.estense.com)
1.3 Objectives of the thesis
Because of the potentially irreversible impact of groundwater quality deterioration in the
Ferrara coastal aquifer, answers concerning the assessment of the extent of the salinization
problem, the understanding of the mechanisms governing salinization processes, and the
sustainability of the current water resources management are urgent.
In this light, the present thesis aims to achieve the following objectives:
6
INTRODUCTION
Characterization of the lowland coastal aquifer of Ferrara: hydrology, hydrochemistry and
evolution of the system
The actual hydrogeochemistry and the geological evolution of the unconfined coastal
aquifer of Ferrara have been addressed in order to understand the drivers of groundwater
salinization. Physical aquifer parameterization was performed to infer the presence of a
vertical hydraulic gradient due to the drainage system. Seepage of relict seawater, trapped
within low permeability sediments during the Holocene transgression, was found to be
responsible for groundwater salinization. The hydrogeochemical investigation allows also
distinguishing between the role of seepage and water–sediment interactions (such as NaCl
solution, ion-exchange, calcite and dolomite dissolution/precipitation, oxidation of organic
matter, and sulphate bacterial reduction) in the evolution of groundwater chemistry.
The importance of data acquisition techniques in saltwater intrusion monitoring
A detailed vertical characterization of the Ferrara coastal aquifer is unavoidable to
properly monitor salinity distribution. Physical-chemical depth-dependent logs were
obtained by different techniques: Open-Borehole Logging (OBL), Integrated-Depth
Sampling (IDS), and Multi-Level Sampling (MLS) via straddle packers. The OBL led to a
satisfactory reconstruction of the extent of the fresh-saltwater interface, but provided a
misleading characterization of the distribution of redox environments within the aquifer.
The IDS conceptual model recalls the phenomenon of actual seawater intrusion, with Cl concentration never exceeding that of seawater and the absence of hypersaline
groundwater all over the study area. On the contrary, good fits between sedimentological,
stratigraphycal and physical-chemical data were obtained using the MLS. In the MLSderived conceptual model the lower portion of the unconfined aquifer shows hypersaline
groundwater, leading to necessarily consider salinization processes other than seawater
intrusion, like the water/sediment interaction or the geochemical evolution of
groundwater due to evapo-concentration. Results demonstrate the need to use MLS in
reconstructing a reliable conceptual model, especially in areas where the density contrast
between fresh and saline groundwater is large.
Predicting salinization trends in the lowland coastal aquifer
A conceptual model based on detailed topography and bathymetry, stratigraphyc
information from analysis of well logs, well driller’s reports and pumping tests was
developed to quantify the possible impacts of climate change on the whole coastal
groundwater system in the Ferrara Province. On the basis of such a conceptual model, a
three-dimensional variable-density groundwater flow model coupled with solute transport
was developed; it was calibrated using highly parameterized inversion techniques with
MODFLOW-2000 to ensure that the model is a reasonable representation of the physical
7
INTRODUCTION
system. The calibrated model was employed to constrain a two-dimensional numerical
simulation, using SEAWAT, capable of investigating the interactions between the coastal
unconfined aquifer and the canals network, and to quantify the possible future effects of
climate change on groundwater salinization with a multiple scenario approach: (i) increase
in evapotranspiration induced by temperature increase; (ii) increase in the frequency of
extreme high rainfall events; (iii) extreme drought conditions; (iv) and canal dewatering
due to salinization of the water courses. The results show that the Ferrara coastal area will
experience a significant salinization and that the major cause is autonomous salinization
via seepage of saline groundwater.
Ammonium occurrence in a salinized lowland coastal aquifer
Impacts induced by climate change may consist in the modification of water/sediment
interaction. A detailed characterization concerning the spatial distribution and the possible
sources of nitrate and ammonium concentrations in the Ferrara coastal aquifer was
completed to understand if salinization processes may play a role in the mobilization of
nitrogen species. Ammonium is the prevalent nitrogen inorganic species in groundwater,
and its concentration increases with depth and salinity. Very high ammonium
concentrations are found in coincidence with peaty sediments in the salinized anoxic
aquifer and in the low-lying aquitard. The results from this study show that the elevated
ammonium concentration derives from the mineralization of organic matter present in fine
sediments deposited in paleo-marsh environments.
Trace elements mobility in a saline coastal aquifer
The effect of elevated salinity on the mobility of trace elements (TE) was studied. The use
of intensive depth profiles gave insights into groundwater and sediment matrix
interactions. The total concentration of TE within the aquifer matrix was higher in the fine
sediments, since usually TE concentrate in the clay fraction, while groundwater
concentrations generally increased with the ionic strength of the solution. Still, some TE
did not follow this trend, being probably released by fertilizers leaching. Thus, the results
from this study highlight the need to distinguish between the release of TE due to
alteration of water/soil interaction equilibrium induced by groundwater salinization and
the release of TE in groundwater due to anthropogenic activities.
8
SITE DESCRIPTION
2.
SITE DESCRIPTION
2.1 Study area
The study area is situated within the Province of Ferrara (Northern Italy) and covers 700 km2.
This coastal floodplain is bounded to the North by one of the Po River branches, the Po di
Goro, and to the South by the brackish water marshes of the Comacchio Lagoons (Fig. 2.1).
Most of the territory is a recently reclaimed land with flat topography below sea level, and
with an altitude ranging from 12 to -11 m above sea level (a.s.l.); the only topographic heights
consist of dune systems, paleodunes and river banks.
The climate is temperate, characterized by cold winters and warm summers with modest
diurnal (10-12 °C) and annual temperature ranges (20-25 °C). Rainfall is on average 670
mm/year, evenly distributed throughout the year. Aquifer recharge from precipitation
surplus is minimal due to a reference evapotranspiration rate of about 650 mm/year: it
occurs mainly during winter months (Mollema et al., 2012).
The area is intensively farmed since its reclamation (late 1900), and the primary land use is
horticultural crops (generally carrots, asparagus, strawberries, melons, watermelons, etc.).
Tourist activities are also developed along the coast, especially during summer periods (JuneSeptember).
Figure 2.1 - Ferrara Province topography map and location of the coastal lowland study area
(dashed rectangle).
9
SITE DESCRIPTION
Several natural and anthropogenic features threaten this area: saltwater intrusion in the
phreatic aquifer and seawater encroachment along the rivers (Giambastiani et al., 2007);
natural (Carminati et al., 2005) and anthropogenic land subsidence (Teatini et al., 2006); soil
salinization; high demand of water during the peak tourist season; insufficient aquifer
recharge and sea level rise (Antonellini et al., 2008).
2.2
Morphological evolution during the last glacial period
The Quaternary evolution of the Po Delta plain is controlled by eustatic and climatic
fluctuations (Stefani and Vincenzi, 2005). The stratigraphyc architecture shows consistent
patterns with changing sea level position recording the depositional evolution from lowstand
continental accumulation to marine transgression and highstand progradation.
The last glacial lowstand, when the sea level dropped 120 m below the present level and the
paleoshoreline shifted 250 km South-East of its current position, resulted in extensive and
repeated basinward shifts of facies, which can be observed across closely spaced
unconformity surfaces associated to alluvial plain sedimentation in the modern coastal area of
Ferrara (Amorosi et al., 2003).
During the Flandrian transgressive phase (18000-5500 yr BP), these deposits were followed
by a retrogradational stacking pattern of coastal plain and littoral facies, reflecting the
landward migration of a barrier-lagoon-estuary system. Back-stepping fluvial and brackish
marsh deposits with widespread organic clay intercalations and peat horizons are followed by
delta-estuarine sand bodies.
At the maximum transgression (6000-5500 yr BP) the paleoshoreline was located 20-30 km
west of its present position (Amorosi et al., 2003). During the early highstand period, large
sand spits and barrier islands grew, progressively turning the previous bays into confined
lagoons (Amorosi et al., 2004). Afterwards, the depositional dynamics were significantly
modified by both climatic change (warmer conditions starting from 2800-2700 yr BP) and
anthropic events (infrastructures, reclamation works and fluvial diversions) causing coastal
progradation and a consequent regressive upward succession of the deposits (Fig. 2.2).
Coastline migration and different phases in the Po Delta progradation resulted in a series of
coastal sand ridges (paleodunes) and paleo marsh lagoons, corresponding to the alternation
of current topographic highs and lows (Simeoni and Corbau, 2009).
10
SITE DESCRIPTION
11
SITE DESCRIPTION
2.3
Recent anthropogenic activity
Since the Middle Ages, the need to acquire new areas to be allocated to agricultural activities
led to a succession of remediation works (Fig. 2.3), intensified from 1850s to 1960s.
The resulting surface water system is very complex, since it is constituted by numerous
natural and non-natural water bodies (Fig. 2.2): the course of the Po River, its tributaries and
an extended hydrological network consisting of WE oriented channels used for irrigation.
These last derive water by gravity directly from the Po River and drainage ditches connected
to pumping stations that maintain this lowland dry, lastly discharging water to the sea
(Antonellini et al., 2008). This induced the stiffening of the hydrographic geometry and
consequently increased the natural subsidence rate (Teatini et al., 2005), which has dropped
most of the territory below sea level, modifying river and groundwater flow regimes.
Nowadays, the drainage system is necessary to maintain the surface water level at about -0.5
m a.s.l. because of the agricultural requirements and to guarantee water discharge toward the
sea to keep this lowland dry (Fig. 2.4).
Figure 2.3 – Waterworks in the Ferrara Province at the beginning of the XX century
(www.bonificaferrara.it).
12
SITE DESCRIPTION
Figure 2.4 – “Valle Lepri” dewatering station in the Ferrara Province lowland
(www.bonificaferrara.it).
2.4
Hydrogeological setting
The complex depositional evolution of the Po Delta system has led to complex aquifer
geometry. The sedimentary sequence is characterized by transgressive bay and barrier
deposits at the base, and prograding sandy deltaic lobes and strand plains at the top. From
East to West, the majority of the sedimentary units consists of a wedge of permeable sand
sediments deposited in shallow sea with intercalation of peat and silty layers, littoral sands
formed in the foreshore and in the adjacent beach, and sand dune systems (Simeoni and
Corbau, 2009).
The coastal aquifer in the study area is generally unconfined; in the westernmost area thin
layers of fine continental alluvial deposits (mostly silt and clay) overlay littoral sands
(Amorosi et al., 2002) making the aquifer semi-confined. The unconfined coastal aquifer is
mainly located within the littoral sands and shallow marine wedge deposits. Fine-grained
deposits of silt and clay at the bottom of sand deposits constitute a 10 m thick confining unit.
The aquifer thickness decreases inland and ranges between 2 and 24 m, while the thickness is
15-20 m.
The aquifer hydraulic conductivity is variable, depending on the sorting of aquifer materials
and ranges from 1x10-6 m/s in depressed areas to 1x10 -3 m/s in the coastal dunes
(Mastrocicco et al., 2012c).
13
SITE DESCRIPTION
The horizontal hydraulic gradient of the shallow unconfined aquifer (0.1-0.5 ‰) is manly
controlled by canals and drains system; the horizontal flow velocity is generally very low,
about 7-10 m/y (Mastrocicco et al., 2013b). On the contrary, quite high vertical hydraulic
gradients have been measured in this aquifer due to the hydraulic head difference between
groundwater and surface waters (about 1 m) that causes a permanent upward groundwater
seepage flux even if in the study area the unconfined aquifer is not stressed by pumping wells
(Giambastiani et al., 2013).
Figure 2.5 - Excerpt of the coastal territory of the Duchy of Ferrara in the XVI century
(www.ilmuseodimirabello.com)
14
SITE DESCRIPTION
The unconfined aquifer in the study area is affected by relict saline and hypersaline
groundwater due to the presence of vast coastal lagoons originally connected to the sea (Fig.
2.5); these lagoons, keeping a high degree of salinity in pore-water, were reclaimed to acquire
new land for agricultural practice. Even nowadays, the upconing of such highly saline
groundwater (Giambastiani et al., 2013) could threaten the agricultural practices
(Mastrocicco et al., 2012a). In fact, in this area, fresh groundwater lenses are created by
infiltration of rainwater or canals leakage in permeable sandy sediments. The freshwater
thickness allows plant growth, thus strongly influencing both agriculture (Mastrocicco et al.,
2013a) and natural vegetation (Antonellini and Mollema, 2010). Given this active exchange
between groundwater and the surface water network (Colombani et al., 2015a), the need to
drain the agricultural land to keep it dry may concurrently induce salinization of surface
water due to saline groundwater upconing. Fortunately, the upconing is often hindered by the
vertical superimposition of low permeability lenses on permeable layers (Fig. 2.2).
15
MATERIALS AND METHODS
3.
MATERIALS AND METHODS
3.1
Field activities
This section describes the methodology used to characterize the unconfined aquifer of the
Ferrara province, in particular the methodology developed via straddle low pressure packers
to simultaneously characterize vertical hydraulic gradients, hydraulic conductivity, and
groundwater quality with the aim of identifying the aquifer dynamics.
3.1.1
Monitoring wells installation
Since 2009, the Geological Survey of the Emilia-Romagna Region (Northern Italy) has
developed a regional monitoring network to monitor saltwater intrusion in the coastal aquifer
facing the Adriatic Sea. A distributed network of 59 monitoring wells (Fig. 2.2) was installed
to monitor water table fluctuations and groundwater quality of the shallow unconfined
aquifer. 46 standard PVC monitoring wells (2.5 cm inner diameter) were drilled manually
with a Ejielkamp Agrisearch auger equipment up to depths between 1.1 and 6 m below
ground level (b.g.l.), depending on the water table depth. Monitoring wells’ location was
selected on the base of the soil characteristic in order to sample the different soil types of the
Ferrara Province. These monitoring wells are located in agricultural fields, in which crop type,
preceding crop, irrigation amount, type and amount of applied fertilizers are well known
(Mastrocicco et al., 2011c). In addition to the shallow monitoring network, 13 deeper
boreholes (5 cm inner diameter) were drilled to get insight of groundwater salinization and
inorganic N concentration in the unconfined aquifer. These wells were screened from 1 m
below ground level (b.g.l.) to a maximum of 22 m b.g.l. to fully penetrate the aquifer. To
prevent cross contamination between sampling points and short-circuits during pumping
tests, the screens were not surrounded by a siliceous gravel pack; a geotextile sock was
instead used to minimize clogging. The monitoring wells were sealed at the top with a mixture
of cement and bentonite to prevent surface-water infiltration. In particular, the monitoring
wells P3, P8, P33, and P34 are located along a flow line and are aligned along fore-stepping
direction of Holocene Po Delta system (Fig. 2.2). With respect to the geomorphologic features
(map in Figure 2.2), P8 and P34 are located in an area of marsh lagoonal deposits
characterized by silts and fine sands, P3 is found on a palaeo-dune characterized by medium
to fine sand deposits, while P33 is placed on silty clays and silt-very fine sands, typical of bay
deposits.
Core samples were also collected in P33 and P34 (Fig. 2.2) every meter or when a change in
stratigraphy was recognized during drilling operations. Samples were stored in a cool box at 4
°C and immediately transported to the laboratory for sediment analysis.
16
MATERIALS AND METHODS
3.1.2 Continuous monitoring of piezometric heads and surface waters levels
Groundwater levels were measured twice in the 46 shallow monitoring wells (Fig. 2.2), once
in summer and once in winter, via a PASI phreatimeter. In the 13 deep monitoring wells,
heads were measured every 4 months from 2009 to nowadays.
In order to define groundwater temporal variations due to surface water/groundwater
interaction and due to tidal fluctuations, two data-loggers LTC M10 Levelogger Solinst® were
inserted at -4.5 m and -8 m a.s.l. inside the monitoring well P8, using Teflon packer rings to
insulate each aquifer portion, from July to December 2012. The data-loggers were set up for
continuous barometric compensated level measurement and electrical conductivity (one
record per hour).
To define the physical boundary conditions of the Ferrara coastal aquifer, surface waters were
monitored over the whole period of the study.
The online data from Porto Garibaldi tide gauge (www.provincia.fe.it), positioned in proximity
of the monitoring well P8, were used to compare groundwater fluctuations with tidal
variations. The Valli di Comacchio and Valle Bertuzzi water level gauges (www.provincia.fe.it)
provide stage data for the two major marsh lagoons of these territories. The average stages of
the Lago delle Nazioni, Po River and irrigation canals were supplied by the Consorzio di
Bonifica Pianura di Ferrara.
To understand in detail the interaction between the irrigation canals and the shallow aquifer,
a datalogger LTC M10, programmed to record every 30 minutes, was inserted in the “Canale
della Gronda” (Fig. 4.24) to monitor hydrometric level variation and EC from July to December
2012. The water level and EC variations in the drain ditch and in the canal were also
monitored every two weeks with a hand meter and a Hanna ® HIcell-21 portable conductivity
meter.
3.1.3
Depth profiles acquisition
Three different methodologies were employed to reconstruct the distribution of the dissolved
species in groundwater over the whole profile of the unconfined aquifer: the Open Borehole
Logging (OBL), the Integrated Depth Sampling (IDS) and the Multi-Level Sampling (MLS).
In order to obtain the OBL profiles, a Hydrolab MS-5 (OTT, Germany) water quality probe was
lowered manually into the monitoring wells to simultaneously record temperature, electrical
conductivity (EC), dissolved oxygen (DO), redox potential (Eh), pH and salinity. The Hydrolab
MS-5 water quality probe consists of six sensors and a data transmitter mounted inside a 4.4
cm diameter housing, with a 30 m long underwater cable and an YSI interface (YSI, California)
at the ground surface. Eh was corrected to the standard H+ electrode potential. After a 5 min
wait, the six parameters were simultaneously acquired every 1 m depth. The hydrochemical
logs were made under natural conditions, without purging the monitoring wells before the
tests.
17
MATERIALS AND METHODS
For the IDS measurements a Proactive® electrical submersible pump with flow controller was
placed in the middle of the screened interval for each monitoring wells and a minimum of four
volumes were purged maintaining a flow rate of 0.1-0.5 l/min to minimize the piezometric
lowering.
For the MLS measurements a Solinst® 800L straddle packers system was used to isolate a
window of 0.2 m within the fully penetrating wells. After the selected port and the tubing
were purged for 3 volumes (approximately 15 l), a groundwater sample was collected at each
depth interval via a low-flow technique using an inertial pump.
The groundwater samples, both for the IDS and the MLS measurements, were collected after
the stabilization of physical-chemical parameters monitored via a Hydrolab flow-through cell
connected to the Hydrolab® MS-5 OTT Corr-Tech® probe able to acquire in situ T, pH, EC, Eh
and DO. Groundwater samples for major ions were then filtered through 0.22 μm Dionex
polypropylene filters, stored in vials with no head space, kept refrigerated in a cool box at 4 °C
and immediately transported to the laboratory to be analyzed. Groundwater samples
intended for TE analysis were filtered through 0.45 µm polypropylene filters and collected in
acid-washed and deionized-water-rinsed HDPE or amber borosilicate glass bottles preserved
with ultrapure acid. Teflon sealed glass vials with no headspace were used to sample for
dissolved H2S.
To compare data obtained by OBL and MLS techniques, the absolute residuals (the absolute
value observed using MLS minus the absolute value observed using OBL at the same depth)
and the linear regression coefficient (R2) with respect to the line YOBL = XMSL was calculated for
each parameter analyzed.
Figure 3.1 - Multilevel Sampler set up and installation
18
MATERIALS AND METHODS
To infer the chemical boundary conditions of the Ferrara coastal aquifer, surface waters were
also sampled: the Po di Gnocca near the monitoring well P4, Po di Volano near the monitoring
well P1 and Canale Circondariale near the monitoring well P10, were sampled in four surveys
from 2010 to 2014; the seawater samples were collected at the shoreline in seven surveys
from 2010 to 2014 near the monitoring wells P1, P8 and P9.
Informations concerning the main surface bodies of this area (Valli di Comacchio, Valle
Betuzzi and Lago delle Nazioni) were also available from the monitoring programs of ARPA
and Ferrara Province.
Finally, the bulk (wet and dry) atmospheric deposition was sampled in four campaigns from
2010 to 2014 using a HDPE bottle and a 20 cm diameter funnel equipped with a 100 μm
Nitex® net to avoid external contamination. Samplers were installed near the monitoring
wells P1 and P8 at a height of 1.5 m above the ground.
3.1.4
Constant rate pumping tests
Constant rate pumping tests were carried out every meter in the 13 deep monitoring wells in
order to define physical parameters of the aquifer and to quantify flux dynamic under
drainage situation. Tests started at the bottom of the well and, after each test, the packer was
lifted 1 m and the test was repeated until the entire well was tested.
To perform constant rate pumping tests, two inflatable Solinst® 800L straddle packers were
used to isolate a window of 0.2 m within the fully penetrating monitoring wells. The original
arrangement was modified by replacing the straddle pipe, within the sampling window, with a
LTC M10 Levelogger Solinst® to monitor head, temperature (T) and electrical conductivity
(EC) every 1 second. The accuracy of the Levelogger employed in this occasion is ± 0.01 m for
a maximum pressure range of 10 m; below this depth the measurements are less reliable.
The system is set at the chosen depth within the monitoring well and the packers are inflated
by a portable air compressor until they stick to the well screen. Before starting the pumping
test, the inflated packers system is left to stand for at least 20 minutes (or longer in finer
sediments) until the piezometric equilibrium is reached and the point head measurement is
recorded. Afterwards a constant pumping rate test is performed using an in line submergible
centrifuge pump able to deliver 20 l/min, and a flow controller to regulate the flow and to
perform the constant rate pumping tests.
Hydraulic conductivity (k) values (m/s) were derived from the equation (Bureau of
Reclamation, 2001):
k
Q
Cs rh
(1)
where Q is the steady flow in the well (m 3/s); Cs (-) is the conductivity coefficient for semispherical flow in saturated materials through partially penetrating cylindrical test wells and
for these condition is equal to 28 based on graphs defined in Bureau of Reclamation study
19
MATERIALS AND METHODS
(2001), r (m) is the radius of the test well, Δh (-) is the head hydraulic gradient between static
head and steady state head under pumping condition.
Since the Levelogger LTC simultaneously acquires the piezometric drawdown induced by
pumping and any variations in T and EC within the aquifer window isolated by the packers,
the equivalent freshwater head (hf) and vertical flow values (qz) can be respectively derived
using the equations (2) and (3) that take into account variable density and salinity values
(Post et al., 2007):
hf 
i   f
i
hi 
zi
f
f
(2)
where hi (m) is the hydraulic head measured at point i, z i (m) is the elevation head and
indicates the depth of the well screen, ρ i and ρf (kg/m3) are respectively the density of the
water in the sampling window and the density of freshwater; and
 h f   a   f
q z   K av 

 

z
f






(3)
where Kav (m/s) is the average kv calculated between two sampling windows (i = 1 and 2), ρ a
(kg/m3) is the average density of the groundwater between the two sampling points, and Δh f
= hf,2 – hf,1 (m) and Δz = z2 – z1 (m) are respectively the difference in freshwater head and
elevation head of two consecutive measures.
Finally, the pore water velocity along each monitoring well was calculated following Darcy’s
law in the form of freshwater heads:
v

k f i
ne
(4)

where v is the average pore water velocity (m/y), i is the freshwater hydraulic head gradient
(-), ne is the effective porosity (-) and kf is the fresh water hydraulic conductivity (m/y). By
assuming that salinity variations have a negligible effect on the groundwater kinematic
viscosity, viscosity kf values can be considered similar to field-measured values of k (Post et
al., 2007). In order to compute velocity values, k values were obtained by equation (1) while
effective porosity values (ne and nev) were calculated for each core sample by equation (3)
using both the measured total porosity (n) and the calculated total porosity (nv) by Vukovic

and Soro (1992). For the shallow coastal aquifer, a regional i influenced by the dewatering
stations equal to 0.4‰ was applied.
20
MATERIALS AND METHODS
3.2
Analytical methods
This section describes the laboratory procedures used to analyze water and sediments
samples with the aim of identifying the spatial distribution of saline groundwater, major ions
and selected contaminants in the Ferrara coastal aquifer.
3.2.1
Sedimentological characterization
Particle size curves were obtained using a sedimentation balance for the coarse fraction and
an X-ray diffraction sedigraph 5100 Micromeritics for the finer fraction; the two regions of the
particle size curve were connected using the computer code SEDIMCOL (Brambati et al.,
1973).
The median of the average grain radius (expressed in φ), the 10th and 60th percentile of the
cumulative curve (d10 and d60 expressed in φ) and the sorting were calculated to describe
grain size distribution analysis. Grain size distribution was then used to calculate hydraulic
conductivity by pedotransfer formulae (Kped), such as Breyer and Shepherd formula (Vukovic
and Soro, 1992), which are appropriate for materials with heterogeneous distributions and
poorly sorted grains. Calculated K ped were compared with the results of the constant rate
pumping tests (see paragraph 3.1.4) carried out in the same monitoring wells to validate the
methodology.
Organic matter content (SOM) of the sediments was measured by dry combustion (Tiessen
and Moir, 1993), then converted to total organic carbon (TOC) using the conversion factor of
0.58 (Danielson and Sutherland, 1986). The water content was measured gravimetrically in
saturated condition after heating for 24 hours at 105°C (Danielson and Sutherland, 1986).
Laboratory porosity measurements (n) on core samples are often affected by errors, as many
sampling methods alter the structure, packing and compaction of the sediment sample
(Vienken and Dietrich, 2011). Hence porosity values were also computed using the following
empirical equation (Vukovic and Soro, 1992):
nv  0.2251  0.83 
(5)
where η stand for uniformity coefficient, defined as d60[φ]/d10[φ].
The effective porosity (ne) was estimated with the formula (Worthington, 1998):
ne  n  CBW
(6)
where CBW is the clay-bounded water, which comprises electrochemically bound water from
clay surfaces and interlayers. The CWB varies in volume according to the clay-type and the
salinity of the formation water, and can be estimated from (Hill et al., 1979):
21
MATERIALS AND METHODS


CBW  n  0.6425  S  0.22  CEC
(7)
where S is the salinity (g/l) and CEC is the Cation Exchange Capacity (meq/ml of pore space).
3.2.2
Chemical analysis of water samples
Major ions in groundwater were analyzed using an isocratic dual pump ion chromatography
ICS-1000 Dionex, equipped with an AS9-HC 4 x 250 mm high capacity column and an ASRSULTRA 4mm self-suppressor for anions. An AS-40 Dionex auto-sampler was employed to run
the analyses; Quality Control (QC) samples were run every 10 samples and the standard
deviation for all QC samples run was better than 4%. The sum of the resulting anions and
cations was referred to as Total Dissolved Solids (TDS).
The EC data from the field depth profiles were also converted to total dissolved solids (TDS)
using a linear relationship obtained from Mastrocicco et al. (2012c), where 0.873 is found to
be the local slope between groundwater EC (dS/m) and groundwater TDS (g/L). This slope
differs from the typical value of 0.64 since the local groundwater ionic composition is quite
complex (Giambastiani et al., 2013).
Trace elements (TE) in groundwater (B, Al, P, Fe, Sr, Sc, Ti, Mn, V, Cr, Co, Ni, Cu, Zn, Ga, As, Rb,
Ba, Hg, Pb, U) were analysed using an X Series Thermo-Scientific spectrometer (ICP-MS).
Specific amounts of Rh, In and Re were added to the analysed solutions as internal standards,
in order to correct for instrumental drift. Accuracy and precision, based on replicated
analyses of samples and standards, are better than 10% for all elements, well above the
detection limit.
The H2S was analysed with LCK 653 cuvette tests and a CADAS 100 UV/Vis
spectrophotometer (Hach-Lange, UK).
Alkalinity was measured via titration using the kit Alkalitats-Test (Aquamerck®).
Saturation index (SI) of gypsum, calcite and dolomite were also calculated using PHREEQC
software (Parkhurst, 1995).
Plots of selected species concentrations of all water samples were built and the relationships
between the groundwater samples, the seawater (Adriatic Sea) and freshwater (irrigation
canals surrounding the monitoring wells) end-members were analyzed.
3.2.3 Geochemical analysis of soil samples
NH4+ was determined in sediments both with ion chromatography and a double beam Jasco V550 UV/VIS spectrophotometer (Bower and Holm Hansen, 1980). The two different analytical
techniques employed returned similar NH4+ concentrations with a correlation coefficient (R2)
of 0.971. KCl 1 M was used to extract NH4+ from the sediment samples, using the sediment to
water weight ratio of 1:10. Sediment and water were mixed and sealed in baker, then shaken
for one hour, and centrifuged at 3500 rpm for one hour at 25 °C to separate the sediment from
22
MATERIALS AND METHODS
the solution. Total leachable nitrogen (TLN) was extracted from the sediment samples by
using Milly-Q (Millipore, USA) water and the sediment to water weight ratio of 1:10; leachates
were analysed with LCK 238 LatoN cuvette tests and a CADAS 100 UV/Vis spectrophotometer
(Hach-Lange, UK).
Major oxides (SiO2, TiO2, Al2O3, Fe2O3, MnO, MgO, CaO, Na2O, K2O and P2O5 expressed in
weight percent) and TE (Sc, V, Cr, Co, Ni, Cu, Zn, As, Rb, Sr, Zr, Ba, La, Ce, Pb, Th and S
expressed in ppm), were analysed by X-ray fluorescence (XRF) on powder pellets, using a
wavelength-dispersive automated Panalytical Axios 4000 spectrometer. Accuracy and
precision based on systematic re-analysis of standards were better than 3% for major oxides;
while accuracy and precision were below 10% for TE (above 10 ppm). Detection limits for TE
results are 3 ppm. As reference standards, the E.P.A. Reference Standard SS-1 (a Type B
naturally contaminated soil) and the E.P.A. Reference Standard SS-2 (a Type C naturally
contaminated soil) were also analysed to cross-check and validate the results.
Calcite and dolomite content in soil was determined with a Chittick gasometrical apparatus.
Pyrite content and pyrite-bound Fe were then calculated from the total S content:
and
(8)
where FeS2 is pyrite content, Fepy is pyrite-bound Fe, Mi is the molar mass of compound i, and
S is total sulphur. These formula work well as long as total reactive Fe is higher than total S,
taking into account stoichiometry, which holds for nearly all samples except for the peaty
sediments, where some kind of overestimation could be expected.
Total reactive Fe was then calculated using the following empirical formula, based on the XRF
analyses:
(9)
where FeTR is total reactive Fe, and Fe2O3 and Al2O3 are total Fe content and total Al content
respectively. This formula assumes that silicate-bound Fe2O3 amounts to about 22.5% of total
Al2O3 content and that total reactive Fe could be viewed as enrichment on top of the silicatebound Fe (Dellwig et al., 2001; 2002).
By subtracting pyrite-bound Fe, the non-pyrite bound reactive Fe (Fereact) can be calculated:
(10)
The potential reduction capacity (PRC) takes into account sediment organic matter (SOM) and
pyrite as major reductants (Hartog et al., 2002):
(11)
23
MATERIALS AND METHODS
where the factor 10,000 is a dimensionless constant for the conversion from % to meq/kg dry
mass. The SOM and the grain size distribution analyses in P33 and P34 were taken from a
previous work (Mastrocicco et al., 2012c).
Carbonate buffering upon pyrite oxidation (CBPO) takes into account the acid buffering
capacity of carbonate following pyrite oxidation by O 2:
(12)
Again, the factor 10,000 is a dimensionless constant for the conversion from % to mmol/kg
dry mass and CaCO3 + MgCa(CO3)2 is the carbonate content.
3.3 Numerical modelling activities
On the basis of the conceptual model drawn from field and laboratory activities, densitydependent flow and transport numerical simulations were implemented and calibrated via
commonly used codes, such as MODFLOW-2005 (Harbaugh, 2005), SEAWAT 4.0 (Langevin et
al., 2007), Hydrus-1D (Šimunek et al., 2008) and PEST (Doherty, 2002).
3.3.1 Site wide steady-state simulation
A non-reactive transport model was used to investigate the freshening and salinization
processes of the unconfined aquifer. SEAWAT 4.0 (Langevin et al., 2007) was used to simulate
the density driven flow and transport of saline and fresh groundwater. For the saltwater
transport, no chemical reactions were considered since this study was planned to investigate
physical controls on conservative chemical transport.
The model domain of the study area extends over 753 km 2, 25 km in the East-West direction
and 45 km in the North-South direction. The length of the simulation period was set equal to
100 years, divided into 1000 time steps.
The grid was defined by 125 columns, 225 rows and vertically discretized into 10 layers of
variable thickness. The top of the grid is represented by the local ground surface, ranging
from 12 m to -10 m a.s.l., obtained by entering the DTM of the Emilia-Romagna Region; the
bottom of the grid, representing the top of the underlying aquitard (the silty-clay unit), ranges
from -12 m to -20 m a.s.l., as deduced from the stratigraphyc logs available in the database of
the Emilia-Romagna Geological Survey.
A mean recharge rate of 155 mm/y was assigned to the entire domain, as a function of mean
precipitation and mean temperature recorded within the site and calculated with the
Thorntwaite method (1948).
24
MATERIALS AND METHODS
The maximum evapotranspiration rate was set to 95 mm/y with an extinction depth of 0.5 m
from ground surface, assuming that evapotranspiration is at maximum rate when the
groundwater table is close to the surface and becomes null below 0.5 m.
Figure 3.2 - Grid discretization and boundary conditions applied to the model domain: no flow
cells (white), constant heads cells (light blue) drain cells (yellow) and river cells (cyan).
The study area was delineated by null flow cells while all other cells are active cells (Fig. 3.2).
The General-Head Boundary (GHB) package was used to simulate the Adriatic Sea, Valli di
Comacchio and Valle Bertuzzi, by assigning a conductance value of 0.1 m2/s and heads of 0 m
for the Adriatic Sea and -0.5 m for the Valli di Comacchio and Valle Bertuzzi.
25
MATERIALS AND METHODS
The River package was employed to simulate the Po River; the weak connection between
groundwater and the river was simulated assigning a small hydraulic conductivity to the
riverbed, with a conductance of 1e-7 m2/s.
The drainage canals network was simulated with the Drain package, assuming an average
depth of the canals of 0.8 m from the topographic level and a conductance of 0.03 m 2/s. This
value ensured an excellent connection between the surface water and groundwater bodies: it
is the maximum value that can be set, given the size of the cell. The dense network of canals
was imported into the model grid from a georeferenced AutoCad DXF file.
The initial pre-calibration estimate of the hydraulic conductivity distribution has been
determined from geologic logs and pumping tests (Fig. 3.3). Vertical anisotropy was set to 0.1
for all layers, while effective porosity was taken as 0.25 for the entire model based on grain
size analysis. Specific storage and specific yield coefficients for sand and silty-clay sediments
(Tab. 3.1) were taken from literature values (Fetter, 2001).
Figure 3.3 - Hydraulic conductivity (m/s) discretization of the model domain based on surface
geology and the decreasing hydraulic conductivity pattern found during the multi-level pumping
tests.
26
MATERIALS AND METHODS
Table 3.1 - Flow and transport parameters used for the simulations
Parameter name
N° of parameters Value (units)
Horizontal hydraulic conductivity 9
2x10-4a – 1x10-8a m/s
Vertical anisotropy
1
0.1b (-)
Effective porosity
1
0.25b (-)
Specific yield
2
0.25b – 0.1b (-)
Specific storage
2
1x10-4b – 1x10-5b (m-1)
Vertical dispersivity
1
8.16x10-3a (m)
Longitudinal dispersivity
1
1.02a (m)
Molecular diffusion coefficient
1
1x10-9b (m2/s)
a Parameter
values optimized by PEST
b Parameter values taken from literature values
Figure 3.4 - Variogram fitted with a power law function (left plot), and distribution of Specified
Concentration in layer 10 derived from the interpolated measured TDS and pilot points (right
plot).
27
MATERIALS AND METHODS
The Initial Concentration for transport simulation was set to 35 g/l for the Adriatic Sea, to 25
g/l for the Valli di Comacchio and Valle Bertuzzi, and 35 g/l for the Lago delle Nazioni, while 1
g/l was used in the rest of the model domain. The initial concentration set at layer 10 has been
derived from measured TDS at the bottom of the aquifer, while the insertion of pilot points
has been deducted from geology: all data are interpolated with the Kriging method using a
power law variogram. The pilot points were entered in geologically similar zones to each
measured point of the study area (Fig. 3.4).
The power law function had the following parameters obtained via least square optimization:
slope 0.77, correlation length 1, power factor 0.635 and anisotropy ratio 2. The pilot points
were entered into the model in those areas geologically similar to each measured point of the
study area. As can be seen from Figure 3.4, within the bodies of actual and paleo-dunes the
TDS concentration is very low due to higher recharge rates. In palaeo back-barrier
environments, where the water stagnated and evaporated, the salt concentration is
significantly higher. In the northern portion of the model domain, a very low TDS
concentration was used for the presence of Po River and its palaeo riverbed environments.
The numerical solver used for the groundwater flow is the Geometric Multigrid Solver (GMG).
The TDV-ultimate scheme was used to solve the advective term of the solute transport
problem and a constant longitudinal dispersivity of 2 m was employed. The ratios of the
horizontal transverse dispersivity to the longitudinal dispersivity, and of the vertical
transverse dispersivity to the longitudinal dispersivity were respectively set to 0.1 and to
0.0001. An effective molecular diffusion coefficient of 1e-9 m2/s was used throughout the
model domain to account for solute diffusion in water (Tab. 3.1).
The observed field data used for model calibration, via trial and error technique, were: the
outflows from drainage basins (data from the Consorzio di Bonifica Pianura di Ferrara), the
hydraulic heads and the TDS concentrations.
3.3.2 Two dimensional transient simulation
Along the transect A-A’, perpendicular to the Canale della Gronda (Fig. 4.24), a non-reactive
transport model was used to investigate the freshening and salinization processes of the
unconfined aquifer.
Groundwater flow simulation was carried out using the MODFLOW-2005 code (Harbaugh,
2005) while SEAWAT 4.0 (Langevin et al., 2007) was used to simulate the interaction between
the saltwater upward flow and the freshwater canal, taking into account the effect of densitydependent transport processes. For the saltwater transport, no chemical reactions were
considered since this study was planned to investigate physical controls on conservative
chemical transport. Groundwater TDS was simulated to track the seasonal vertical variation
of the freshwater-saltwater interface.
28
MATERIALS AND METHODS
The domain of the cross-sectional model consisted of a uniform model grid of 100 columns
and 30 layers with dimension 0.3 m in the horizontal direction and 0.5 m in the vertical
direction, giving a longitudinal extent of 30 m and a thickness of 15 m (Fig. 3.5).
Figure 3.5 - Numerical grid, with the optimized kh discretization, in yellow the Drain cell, and in
light blue the Time-Variant Specified-Head cells pertaining to the canal and to the pro-delta silty
clays connected with the Adriatic sea.
Only the first layer was 1 m thick in order to accommodate groundwater table fluctuations:
this grid configuration was chosen to avoid cell rewetting, which frequently leads to
numerical instabilities and numerical solver’s non-convergence. A steady state simulation was
initially run to achieve representative heads and TDS distribution within the aquifer in
summer 2012. Then the transient simulation time was subdivided into 34 stress periods to
simulate the variation of the canal stage and recharge during the simulated period.
The irrigation canal “Canale della Gronda”, on the right side of the model, was modelled by
time-variant specified head, and the input were extracted from the LTC M10 datalogger,
which was installed in the canal at the start of the monitoring period. The Drain Package was
used to simulate effects of the agricultural drain (left hand side of the model profile), which
removes groundwater from aquifer at a rate proportional to the head difference between the
aquifer and the drain. When the hydraulic head in the aquifer is greater than the drain
elevation, ground water flows into the drain and is removed from the groundwater model.
Recharge from the drain is always zero, regardless of the hydraulic head in the aquifer. The
drainage ditch elevation was set at -1.95 m a.s.l. with 1x10-4 m2/s of drain hydraulic
29
MATERIALS AND METHODS
conductance; a high conductance was used since the drain ditch was reshaped at the
beginning of the 2012.
The Recharge Package was used to simulate the effective precipitation. The daily recharge
was calculated using the finite element model Hydrus-1D (Šimunek et al., 2008).
The top soil of the site was discretized in 101 nodes of 2.5 mm each, to form a regular vertical
grid of 0.25 m. The soil hydraulic parameters were assigned using default values included in
the soil catalogue of Hydrus-1D for loamy sand. Initial water content conditions were
measured gravimetrically for the top soil and assumed constant throughout the model profile.
Groundwater heads and TDS at two depths (-4 and -8 m a.s.l.) in P8 and within the drain ditch
were used for model calibration. TDS profiles were used for model validation.
At the soil surface, variable flux and head conditions were selected to represent the
atmospheric boundary. Runoff was assumed negligible due to the flat topography and based
on other studies in this area (Mastrocicco et al., 2010). Daily reference evapotranspiration
(ET0) was calculated with the FAO-56 recommended Penman-Monteith equation (Allen et al.,
1998). It was assumed that the grass cover at the experimental site resembles the reference
surface as defined by Allen et al. (1998). Potential transpiration and evaporation were split
using an estimated surface cover fraction of 0.8 (Ritchie, 1972). A meteorological station
recording daily rainfall, wind speed, temperature and humidity was located 3.8 km from the
site, while solar radiation data were recorded in another station located 8.6 km from the site
(www.arpa.emr.it/sim/?osservazioni_e_dati/dexter). The actual root water uptake was
simulated using the turf grass dimensionless water stress response function available in the
Hydrus-1D database (Feddes et al., 1978), the constant height of 0.12 m and the maximum
rooting depth of 0.2 m. The dimensionless constant for the radiation extinction by the canopy
was set to 0.463, the interception constant was set to 0.25 mm and the LAI was derived from
grass height.
As shown in figure 3.6, Hydrus-1D calculated the daily actual transpiration and evaporation
fluxes. Free drainage was selected as lower boundary condition and the cumulative bottom
flux expressed as negative flux (recharge towards the aquifer). The daily recharge was
subsequently discretized in the 34 stress period used in the SEAWAT model.
The flux across the impermeable clay basement that constitutes the aquiclude between the
unconfined aquifer and the confined aquifer, with a conductivity value of approximately 1x10 10 m/s, was assumed to be zero. A constant head with value 0 m a.s.l. was needed to properly
simulate the observed vertical heads gradients present within the aquifer (Giambastiani et al.,
2013), since the bottom part of the unconfined aquifer (the prodelta silty-clays) is
hydraulically connected with the Adriatic Sea. Tidal fluctuations were not considered in the
simulation since in this area of the Adriatic Sea tide magnitude is known to influence only the
near shore monitoring wells (Balugani and Antonellini, 2011).
The temporal TDS variation in the “Canale della Gronda” was simulated by means of TimeVariant concentration cells (Fig. 3.5), while constant concentration cells with a TDS of 65 g/L
was set at the bottom of the aquifer. For a conservative solute as chloride, the molecular
diffusion coefficient for porous media was taken as equal to 1x10 -9 m2/s.
30
MATERIALS AND METHODS
Figure 3.6 - In the upper left hand side plot, the actual cumulative transpiration (blue line) and
evaporation rates (black line) computed by Hydrus-1D are shown. In the upper right hand side
plot, the recharge rate computed by Hydrus-1D (blue line) and its discretization as input in
SEAWAT (red bars) are shown. In the lower left hand side plot, the modelled (blue line) and
observed (black line) canal water levels are shown. In the lower right hand side plot, the
modelled (blue line) and observed (black open circles) canal TDS values are shown.
Model parameters were estimated by inverse modelling using the numerical code PEST
(Doherty, 2002). This process has been performed on the following parameters: kh,
longitudinal dispersivity and vertical dispersivity. The specific storage and yield for transient
simulations were judged insensitive in a preliminary sensitivity analysis and thus not
included in the model calibration. Prior information on the ratio between horizontal and kv
31
MATERIALS AND METHODS
values and measured kh values along the profile were used in order to constrain parameter
estimation.
Upper and lower limits of ±200% with respect to the measured kh values were set as
optimization constraints. Upper and lower limits of one order of magnitude with respect to
the initial estimate of longitudinal and vertical dispersivity values were set as optimization
constraints. The objective function consisted of piezometric heads, drain water table
elevations and TDS in both aquifer and drain. Each screen interval was subdivided with the
layer proportion to ensure that calculated heads were related to the observed heads. The kh
parameters were log-transformed, while the other parameters were left linear for the inverse
modelling.
The calibration of transient state model was performed with the objective of matching
modelled with measured hydraulic head and TDS continuously recorded at -4 m and -8 m a.s.l.
in the monitoring well P8 and every two weeks in the drain ditch (from 01/07/2012 to
04/12/2012). The model validation was performed using the calculated TDS profiles against
the TDS profiles carried out in December 2012 and February 2013 in P8. This was done
including other 10 stress periods in the simulation time to reproduce the recharge events
from 05/12/2012 to 28/02/2013 (Fig. 3.6). Thus, the validated model consisted of 44 stress
periods, 10 more than the calibrated one.
3.3.3 Scenario modelling
Four scenarios were simulated and compared to each other. For the first scenario, a 3%
increase in the precipitation amount and increase in evapotranspiration induced by
temperature increase of 2°C over the next 40 years (Gualdi et al., 2013; Giorgi and Lionello,
2008) were simulated. In the second scenario the frequency of severe high rainfall events was
increased. In the third scenario an extreme drought condition was simulated. The fourth
scenario simulated the interruption of canal usage for irrigation purposes.
As regards the first scenario, the average annual precipitations recorded from 2001 to 2012
(676 mm/year) and climatic data used to compute the Penman-Monteith equation were
downloaded from the Emilia-Romagna climatic database. The daily climatic data were then
mediated over the period 2001-2012 to get mean values for each day of the year. Then, the
mean daily values were used as input for a 365 days simulation with Hydrus-1D (keeping the
model parameters described in section 2.3) to estimate the actual evapotranspiration and
recharge rates. The average annual precipitations were increased by 3% and the induced
future actual evapotranspiration (ET a) was calculated relying on the following equation
(Hanson et al., 2012):
(13)
where TDFfuture and TDFcurrent stand for future and current temperature-dependent function,
respectively, on the basis of mean monthly air temperature.
32
MATERIALS AND METHODS
Current mean monthly air temperature (Ta) was set equal to 14.8 °C, based on the climate
data recorded from 2001 to 2012 and downloaded from the on-line meteorological and
climatic database of the Emilia-Romagna region, while future mean monthly air temperature
(T) was increased of 2 °C and set equal to 16.8 °C.
In order to calculate TDF values, the following form was applied (Hanson et al., 2012):
(14)
The mean annual recharge rate calculated by Hydrus-1D and used for this scenario was 116
mm/year, while the bottom constant head was raised from 0 m a.s.l. to 0.13 m a.s.l. (and from
0 m a.s.l. to 0.62 m a.s.l. in a second sub-scenario) to account for the predicted sea level rise.
Prediction of sea level rise were found in Lambek et al. (2011), where the subsidence rate of
1.5 mm/year summed to a minimum projected Adriatic sea level rise of 1.8 mm/year will
produce a net sea level rise for the next 40 years of 0.13 m in the proximity of Po river delta.
The maximum projected Adriatic Sea level rise of 14 mm/year plus the subsidence rate of 1.5
mm/year will instead produce a net sea level rise for the next 40 years of 0.62 m. The
simulation time of this scenario was subdivided in 8 stress periods of 6 months each. Every
stress period was subdivided into 100 time steps, for a total simulation time of 4 years, to
achieve pseudo steady-state conditions. In all the stress periods the recharge rate was kept
constant while the variable constant head and concentration resembling the canal were
turned on and off every 2 stress periods.
In the second scenario the frequency of extreme high rainfall events was increased
(triplicate). The average annual recharge amount was assumed to occur during winter and
spring periods (six months), instead of being distributed over the entire year. The same subscenarios for the predicted sea level rise were run as explained above. The third scenario
simulated an extreme drought condition by setting null recharge for the last year of
simulation, while maintaining the other boundaries as in the first scenario. The same subscenarios for the predicted sea level rise were run also for this third scenario. To account for
the large uncertainties connected with the climate forecasts, the first three scenarios were run
forcing the input values of plus and minus one standard deviation of the predicted variables.
The fourth scenario simulated the interruption of canal usage for irrigation purposes. Here
freshwater recharge from the canal was set equal to zero (canal dewatering) for a total of 6
months in order to simulate an interruption of canal due to degradation of water quality or
salinization of water courses. This scenario was run forcing the canal dewatering for a total of
9 and 3 months, to account for different circumstances of salinization of surface water
courses. This last simulation was important to understand the effect of no freshwater
recharge from canals. The same sub-scenarios for the predicted sea level rise were run also
for this fourth scenario.
33
RESULTS AND DISCUSSION
4.
RESULTS AND DISCUSSION
This chapter reports the main results obtained from the three years of field, laboratory and
modeling activities spent in the implementation of a reliable conceptual model of the
salinization processes and their effects on the hydrogeochemical behaviour of the coastal
unconfined aquifer of the Ferrara Province. The majority of the outcomes, targeting the
different objectives of this study (see paragraph 1.3), have been published as articles in peer
review journals or presented at international conferences (see paragraph 5.4). Excerpts from
these articles will be presented in the following paragraphs and, to help the reader, each
excerpt will be preceded by a brief explanation that highlights its role within the present
study as a whole.
4.1
Characterization of the lowland coastal aquifer of Ferrara: hydrology,
hydrochemistry and evolution of the system
The first objective of this study was to delineate the hydrogeochemistry and the geological
evolution of the unconfined coastal aquifer of Ferrara to identify the possible drivers of
groundwater salinization. Physical aquifer parameterization highlighted the presence of a
vertical hydraulic gradient, due to the drainage system, that may induce groundwater
salinization via the seepage of relict seawater trapped within low permeability sediments
during the Holocene. The hydrogeochemical investigation also allowed distinguishing
between the role of seepage and water-sediment interactions in the evolution of groundwater
chemistry.
4.1.1 Hydrostratigraphic characterization
Grain size distribution analysis performed on the samples of monitoring wells P33 and P34
(Tab. 4.1) reveals the presence of an heterogeneous aquifer with very poorly-sorted silty clay
and peat lenses overlaying a highly permeable well-sorted sandy layer typical of high energy
depositional environments, such as coastal environment (dunes and beaches), intercalated
with poorly-sorted sandy silt lenses. The sandy and sandy silt layers constituting the aquifer
are underlined by an aquitard, in agreement with previous studies (Amorosi et al., 2003).
Porosity values range between 19% and 76%, with an average value of 30%; the highest
values are associated with the presence of peat layers, which exhibit high organic matter
content, with maximum values of 63%.
The hydraulic conductivity (k), resulting from constant rate pumping tests in the monitoring
wells P3, P8, P33 and P34, decreases from 10 -3 to 10-6 m/s from the top to the bottom of the
aquifer (Figure 4.1a), due to the presence of silt and fine sand lenses.
34
RESULTS AND DISCUSSION
Table 4.1 - Core sample analyses of monitoring wells P33 and P34.
i.d.
P33
P34
Elevation O. M. Sand
m a.s.l
%
%
Silt
%
Clay Median
%
φ
d10
φ
d60
Φ
Sorting
φ
C01
-4.93
2.12
22.28
56.36
21.36
5.90
9.25
5.00
2.28
C02
-5.70
3.06
2.48
67.47
30.05
6.74
10.13
5.63
2.02
C03
-6.08
6.15
0.39
46.70
52.91
8.55
11.88
7.63
2.26
C04
-6.35
17.17
0.31
16.68
83.01
9.49
10.75
9.50
1.56
C05
-6.70
62.50
6.90
63.10
30.00
7.80
10.10
7.8
2.00
C06
-7.53
6.59
1.26
70.59
28.15
7.13
11.00
6.00
2.31
C07
-8.60
6.06
1.46
57.77
40.77
7.95
11.75
6.88
2.48
C08
-9.40
5.80
80.48
10.74
8.78
3.36
7.50
2.50
1.96
C09
-10.53
1.59
70.67
23.80
5.53
3.59
7.00
2.50
1.88
C10
-12.45
0.80
96.52
2.43
1.05
2.25
3.00
2.00
0.47
C11
-13.65
3.21
26.98
59.16
13.86
5.39
8.75
5.00
2.29
C12
-14.23
1.60
92.72
6.30
0.98
3.12
3.88
3.00
0.56
C13
-14.80
2.18
62.65
24.10
13.25
4.50
8.88
3.25
2.22
C14
-16.50
2.30
62.78
28.64
8.58
4.39
7.75
3.50
1.79
C15
-18.80
3.16
63.18
30.62
6.20
4.05
6.50
3.63
1.37
C16
-19.40
3.22
76.39
18.09
5.52
3.62
6.25
3.25
1.38
C17
-20.10
4.07
29.73
44.45
25.82
6.07
9.75
5.50
2.59
C18
-20.90
37.27
4.25
51.18
44.57
7.91
10.63
7.25
2.10
C19
-21.60
6.48
50.36
34.67
14.97
5.02
9.25
3.75
2.23
C20
-23.15
6.22
12.26
59.50
28.24
6.82
10.63
6.00
2.45
C21
-23.90
4.94
55.83
38.92
5.25
4.26
6.38
3.75
1.27
C22
-25.20
5.71
5.59
65.83
28.58
6.95
7.13
5.75
2.42
C23
-25.90
5.45
0.45
29.68
69.87
9.28
12.13
8.50
2.09
C01
-3.80
7.81
1.44
42.85
55.71
8.56
11.75
7.75
2.33
C02
-4.60
7.68
7.38
50.52
42.10
7.66
11.13
6.63
2.62
C03
-6.50
1.65
92.69
5.85
1.46
2.80
3.63
2.75
0.58
C04
-7.50
1.09
96.64
2.73
0.63
2.18
2.88
2.00
0.46
C05
-8.50
1.34
94.56
4.00
1.44
2.09
3.00
2
0.69
C06
-12.50
5.04
64.68
29.63
5.69
4.03
6.63
3.50
1.40
C07
-14.10
10.61
55.58
34.08
10.34
4.61
8.13
3.63
1.99
C08
-15.50
4.87
71.51
23.55
4.94
3.99
6.38
3.50
1.26
C09
-16.50
6.67
13.76
48.86
37.38
6.75
10.25
5.25
2.47
35
RESULTS AND DISCUSSION
These lenses pertain to prodelta depositional environment, which is characterized by a lower
formation energy than the sand dune system: the lower energy leads to finer grain size
distribution and consequently to less permeable sediments in the former environment
compared to the latter. The comparison between k values derived by constant rate pumping
tests, calculated by equation (1) (paragraph 3.1.4), and those determined by pedotransfer
formulae shows a poor agreement along the monitoring well profiles (Figure 4.1b).
Figure 4.1 - (a) Profiles of hydraulic conductivity measured by constant rate pumping tests in the
four monitoring wells and (b) Scatter plot of K values calculated by pedotransfer formula (K ped)
and K values resulting from permeability tests in P33 and P34.
The best linear fit between pedotransfer formulae and measured k values gives R 2 of 0.67 (for
Breyer in P34) with a correlation coefficient of 0.81. This low correlation is mostly due to peat
sediments, where discrepancies of more than 1 order of magnitude occur, since grain size
characterization is particularly challenging due to the organic nature of these sediments. In
addition, it has to be considered that pedotransfer formulae are empirical relations and
introduce high uncertainty; in fact, Vukovic and Soro (1992) noted that the applications of
different empirical formula to the same porous medium can yield k values differing by a factor
of 10 or even 20. Secondly, there are scale issues: hydraulic parameters acquired by constant
rate pumping tests are considered to be average values of large volumes of aquifer (Leven and
Dietrich, 2006), while the hydraulic parameters gained from the application of the
pedotransfer formulae to sediment samples are point values: this scale issue could also be
responsible for the observed discrepancies between k values.
36
RESULTS AND DISCUSSION
Tables 4.2 summarize the parameters measured (k) and those derived from empirical
formulas (ne) used to calculate the mean groundwater flow velocities in monitoring wells P33
and P34. Peclet number (Bear, 1972) is always below 1 and therefore the flow is dominated
by diffusion-driven processes, or at least the role of diffusion is comparable to that of
mechanical dispersion (Appelo and Postma, 2005).
The estimated flow velocities show that seawater intrusion is proceeding inland at a very
slow rate due to the effects of the dewatering stations draining the recently reclaimed area of
the Ferrara Province. Mastrocicco et al. (2012c) have showed that for this shallow coastal
aquifer the regional freshwater hydraulic head gradient is equal to 0.4‰ and horizontal flow
velocities are very slow, meaning that lateral seawater intrusion is proceeding inland at a very
slow rate, especially at distance of 2-10 km from the coastline.
Table 4.2 - Values of measured total porosity (n), calculated total porosity by Vukovic and Soro
(1992) (nv), respective effective porosity (ne, nev) and pore-water velocity (v and vv), and
measured hydraulic conductivity (k) in monitoring wells P33 and P34.
i.d.
Elevation
m a.s.l
n
-
nv
-
ne
-
nev
-
k
m/s
v
m/y
vv
m/y
C05
-6.70
0.76
0.37 0.11
0.05 1.1e-4
12.8
26.5
C08
-9.40
0.33
0.29 0.27
0.24 2.1e-4
9.5
10.9
C09
-10.53
0.18
0.29 0.16
0.26 4.8e-5
3.8
2.3
C10
-12.45
0.27
0.35 0.26
0.34 1.2e-4
5.6
4.3
P33 C11
-13.65
0.24
0.33 0.19
0.27 2.8e-5
1.9
1.3
C12
-14.23
0.25
0.37 0.24
0.35 2.6e-5
1.4
0.9
C13
-14.80
0.19
0.29 0.15
0.24 7.5e-6
0.6
0.4
C14
-16.50
0.20
0.31 0.17
0.26 4.3e-6
0.3
0.2
C15
-18.80
0.21
0.33 0.18
0.28 2.3e-6
0.2
0.1
C03
-6.50
0.26
0.37 0.24
0.34 6.7e-5
3.6
2.5
C04
-7.50
0.25
0.35 0.24
0.34 6.2e-5
3.3
2.3
P34 C05
-8.50
0.24
0.35 0.23
0.33 1.3e-5
0.7
0.5
C06
-12.50
0.25
0.32 0.21
0.27 1.9e-6
0.1
0.1
C08
-15.50
0.23
0.33 0.19
0.27 3.0e-6
0.2
0.1
37
RESULTS AND DISCUSSION
This confirms that groundwater salinization in the deep portion of the aquifer should mostly
be due to local salt sources and to water/sediment interaction, rather than present seawater
intrusion. In fact, considering the geomorphologic evolution of the area during the last
transgressive period that was characterized by landward migration of a barrier–lagoon–
estuary system, it is probable that the aquifer could contain significant soluble salt amounts or
trapped fluids.
Figure 4.2 shows freshwater heads (green colour) and the vertical flow (red colour) in
monitoring wells P3, P8, P33 and P34, along with their stratigraphy. An upward vertical flow
is induced by the drainage system, being in some cases quite significant (0.17 m/d in P33). In
some cases this upward flow is contrasted by small lenses of silt and clay that confine the
aquifer at the top and prevent the complete salinization of the shallow aquifer. This feature is
very important, since the reclaimed land is mainly used for agricultural purposes and soil
salinization could cause severe problems to crops sensitive to salinity stress (Gattacceca et al.,
2009; Zalidis et al., 2002). Recharge flows (negative values in Figure 4.2) are present in P3
and in the upper part of P8, both located in the paleodunes (Figure 2.2) that allow surface
water infiltration.
Figure 4.2 - Transect A-A’ across the coastal aquifer (see Figure 2.2 for location) showing
freshwater head (hf in green colour) and seepage (qz in red colour). Note: positive value means
flow from the bottom toward the surface, while negative values indicate downward flow
(recharge).
38
RESULTS AND DISCUSSION
4.1.2 Hydrochemical characterization
The base-exchange index for dolomitic system (BEX D) was used to characterize the status of
salinization or freshening of the coastal aquifer (Stuyfzand, 2008).
The most common water types found in the aquifer are CaHCO 3, NaHCO3, and NaCl. CaHCO 3
and NaHCO3 consist of fresh to fresh-brackish waters with low salinity. The BEXD is positive
in both cases, indicating freshening processes. In fact, these two water types are found in the
shallowest portion of the aquifer, in proximity of canals (P8) that contribute to dilute
groundwater, and topographic heights, such as paleo-dunes (P3). NaHCO3 is also found in the
transition zone between fresh and salt water and contains higher Na+ (31%) than freshwater
of the irrigation canals. NaCl is the most abundant subtype found in the aquifer (88%) and
consists mainly of brackish, salt and hypersaline waters according to the classification of
Vandenbohede and Lebbe (2012).
The anions of these three major water types (HCO 3-, Cl- and SO42-) are plotted against TDS in
Figure 4.3. Salinity clearly correlates with Cl - in a linear regression with an R2 of 0.99. HCO3remains in all samples between 6 and 36 meq/l and does not increase with salinity. HCO 3- is
the dominant anion in freshwater, while Cl - becomes the most frequent in brackish, saline and
hypersaline waters. SO42- is generally high and increases with TDS; low concentrations of SO42are recorded only in low and very high TDS samples. Possible explanations of low Cl - and high
SO42- content are pyrite oxidation or gypsum dissolution (Mollema et al., 2013).
Figure 4.3 - Salinity (TDS in g/l) and concentration of the anions of major salts. The purple line
indicates the TDS concentration of the current Adriatic Sea water.
39
RESULTS AND DISCUSSION
Figure 4.4 - Saturation index for gypsum, calcite and dolomite along the aquifer depth calculated
with PHREEQC.
In order to verify if gypsum dissolution could occur, the related saturation index (SI) has been
calculated (Fig. 4.4). Negative saturation index (SI) values of gypsum for all samples (Fig. 4.4)
suggest that the occurrence of gypsum as the major source of SO42- in groundwater is not
likely. The most probable source of high SO42- detected in brackish groundwater is the
temporary oxidation of pyrite Fe2S or amorphous FeS due to water fluctuation and temporary
exposure to O2-rich waters that have characterized the geomorphologic evolution of this area.
Furthermore, Vandenbohede and Lebbe (2012) and Stuyfzand and Stuurman (2008) have
shown that brackish and salt relict Holocene transgression waters, which must have
infiltrated in low energy areas protected from direct incursions by the sea, allowing abundant
contact with clay and peat, are characterized by high alkalinity combined with extremes in
SO42- reduction due to the oxidation of organic matter. Many samples in the dataset have high
HCO3- concentration and SO42- concentration significantly lower than the freshwatersaltwater mixing line, indicating the presence of brackish and salt relict transgression waters
in the aquifer. This hypothesis is in agreement with the presence of brackish water marshes
between 3300 and 2000 BP (Stefani and Vincenzi, 2005), the high organic matter content of
the sediment, and the occurrence of peat layers in large part of the aquifer.
The comparison between the major ions (Br -, Na+, SO42- and Mg2+) of groundwater samples,
seawater (Adriatic Sea) and freshwater end-members (irrigation canals), displays a mismatch
and deviations from the mixing line (purple line in Fig. 4.5). Br - vs Cl- plot suggests a
salinization mechanism, which follows an evaporation trend rather than a mixing between
freshwater and marine waters (Fig. 4.5a). Deviations from the mixing line are also observed
for Na+ in the most saline samples (Fig. 4.5b): the observed excess and deficit with respect to
the non-reactive mixing may be due to cation-exchange processes occurring in the aquifer.
40
RESULTS AND DISCUSSION
SO42- vs Cl- concentrations show SO42- depletion (Fig. 4.5c) indicating SO 42- reduction in all
monitoring wells, except for samples of P3 and shallow groundwater samples of P34. SO 42depletion is commonly observed in aquifers affected by seawater intrusion (Andersen et al.,
2005). In many cases it can be attributed to SO 42- reduction processes. The supply of reactive
organic C required for SO42- reduction derives from the peat layers enriched with organic
matter found in the aquifer. This result agrees with the strong and typical smell of H 2S noted
in the field during groundwater sampling and with measured concentrations of up to 50 mg/l
of dissolved H2S in P34 (not shown). Mg2+ excess in Fig. 4.5d is likely explained by Ca2+/Mg2+
exchange due to elevated carbonate and dolomite minerals, which are abundant in this
alluvial aquifer (Amorosi et al., 2002), as confirmed by CaCO 3 and MgCa(CO3)2 SI (Fig. 4.4).
Figure 4.5 - Plots of the concentration of Br-, Na+, SO42-, and Mg2+ with respect to Cl-. The purple
lines indicate the non-reactive mixing line between freshwater and modern seawater; the star
symbol indicates the concentration of the modern Adriatic Sea water.
41
RESULTS AND DISCUSSION
Figure 4.6 - Variation of sample concentration of Na+, K+, Mg2+ and Ca2+ with respect to the
theoretical composition expected for a non-reactive mixing of Adriatic seawater and freshwater.
Note: positive values indicate enrichment, while negative values indicate depletion.
The comparison between the major cation composition of groundwater samples and the
theoretical composition expected for a non-reactive mixing between the two end-members
(Adriatic Sea water and freshwater of irrigation canals) (Andersen et al., 2005; Appelo and
Postma, 2005; Capaccioni et al., 2005; Ghiglieri et al., 2009), allows highlighting exchange
processes within the aquifer. Most of the samples show Na+ deficit (Na+) compared to the
non-reactive model (Fig. 4.6a). The depletion is high and comes up to -260 meq/l for the most
saline samples (P34); only occasional Na + enrichment can be observed. K+ and Mg2+ depletion
(K+ and Mg2+) is observed together with Na+ depletion (up to -11 meq/l for K+ and -62
42
RESULTS AND DISCUSSION
meq/l for Mg2+). On the other hand, Ca2+ values show a general enrichment (Ca2+), except for
P34 that shows Ca2+ depletion. The cation exchange process, which is a common process
during saltwater intrusion in coastal aquifers in presence of exchangers (Andersen et al.,
2005; Appelo and Postma, 2005), most likely explains the relative Ca2+ excess, the Na+ and K+
depletions. The sandy-clayey and sandy-silty matrix could be the cation exchanger, but the
organic matter layers also have a higher cation exchange capacity (Appelo and Postma, 2005)
and appear as the most likely exchanger in the transect.
The Ca2+ contribution from the cation exchange (Fig. 4.6d), the variation of Mg 2+ concentration
(Fig. 4.6d, and 4.6c) as well as the HCO 3- increase (Fig. 4.3) from SO42- reduction should affect
the carbonate equilibrium, causing oversaturation with regard to calcite and dolomite, as
confirmed by SI in Fig. 4.4 (de Montety et al., 2008).
These results highlight that each single cation concentration is determined by the whole
equilibrium of many concurrent water-rock reactions. The ionic exchange intervenes as a
buffer by tending to adjust for variations of chemical composition induced by the increase in
salt content: such occurrence leads to feedback loops, producing strong changes of
hydrochemical facies with respect to those indicated by non-reactive mixing. Precipitation
phenomena, in this context, occur more particularly where salinity is high.
4.1.3 The origin of the salinization processes
The Ferrara coastal aquifer shows TDS values variable from a minimum of 0.47 g/l to a
maximum of 74 g/l. The TDS versus depth scatter plot (Fig. 4.7) does not show a preferential
trend and the data are scattered all over the graph.
The freshwater samples grouped on the left side of the graph belong to the monitoring well
P4, whose TDS is between 0.47 and 0.55 g/l: these values testify the freshening of the local
groundwater due to inflow from the Po River, located just 200 m away from the sampling
point (see Fig. 2.2 for location). Low TDS are registered also in the monitoring well P9, located
within a recent dune system, and in the monitoring well P7 due to the great amount of fresh
water historically discharged into this area for rice cultivation. Shallow freshwater lenses, 2-4
m thick, are also found close to irrigation canals in P8 and in correspondence of topographic
highs such as modern coastal dunes along the shoreline and palaeo-dunes, like in P3 and P5
(see Fig. 2.2 for location). The deepest portions of P3, P5 and P7 display brackish
groundwater, which is also present in the top portions of the monitoring wells P2 and P34.
Saline waters were found over the whole profile in P1 that is just 300 m away from the
shoreline, and in the bottom portions of P2 and P3. Hypersaline groundwater was found at the
bottom of P10, P8 and P33, with TDS ranging between 38 and 62 g/l. The highest TDS was
found in P34 with values up to 74 g/l. In general TDS values increase with depth, because of
the interaction between groundwater and the fine sediments of the deepest part of the coastal
unconfined aquifer that were deposited in a back-barrier environment.
43
RESULTS AND DISCUSSION
Figure 4.7 - TDS versus Depth, for 11 monitoring wells sampled via MLS technique. Water quality
classes are drawn according to Davis and De Wiest (1996).
As an example the distributions of TDS with depth along the A-A’ transect (Fig. 2.2) is shown
in figure 4.8. There are no significant differences between the two monitoring campaigns
carried out in June and November 2011; only the thickness of freshwater lenses varies of 0.51 m, while the high TDS values continue characterizing the entire coastal aquifer.
Figure 4.8 - TDS distribution along the A-A’ transect of the coastal aquifer.
44
RESULTS AND DISCUSSION
To investigate the origin of the high salinity of groundwater, Cl- and Br- have been used as
natural tracers. In particular, their molar ratios plotted versus Cl- concentrations could be a
good indicator of groundwater origin as recently proposed by Hoque et al. (2014). Cl- and Brin groundwater are tracers close to ideal conservative behaviour due to their hydrophilic
character and small ionic size. Moreover, they do not take part in significant ion exchange
reactions at low temperatures, nor are adsorbed onto mineral surfaces (Davis et al., 1998).
The major reservoir of water (the ocean) has relatively uniform Cl - and Br- concentrations and
their molar ratio is around 655±4 (Alcalà and Custodio, 2008); however, for the study area the
observed Cl-/Br- Adriatic Sea molar ratio is 696±47. This value is slightly higher than the ones
reported by other authors, but it can be explained by seawater dilution near the delta due to
freshwater input from the Po River branches (Artegiani et al., 1997).
Bromide and chloride data have been compared with those defined by the mixing line of
seawater (Adriatic Sea) and freshwater end-members (irrigation canals). Given the inherent
variability of TDS in both the end-members, a mixing area rather than a simple mixing line
was constructed using the variance of 21 samples collected from 2010 to 2014 (Fig. 4.9, right
panel).
The Cl-/Br- versus Cl- plot (Fig. 4.9, left panel) shows that most of the samples are above the
Adriatic Sea water (local) value: freshwater samples show Cl-/Br- molar ratios highly variable
from 192 to 1440, while brackish, saline and hypersaline samples show almost constant Cl/Br- molar ratios of 870±70, higher than in local seawater. The highest values of Cl-/Br- in the
freshwater samples belong to the monitoring well P9, which is located within the residential
area of Lido di Spina: this suggests that Cl-/Br- values are imputable to the presence of urban
wastewater. The lowest values of Cl-/Br- in the freshwater samples belong to P7: considering
P7 location, such values are ascribable to agricultural pollution due to rice cultivation. P4
values are typical of recharge waters from inland areas and show a perfect correspondence
with the freshwater end-member, thus confirming the inflow from the Po River; the
freshwater samples from the top portion of P3, P5 and P8 have Cl -/Br- molar ratios that seem
to point to an ideal mixing between freshwater and an end-member that could be reasonably
identified in the local bulk deposition (wet and dry) with Cl-/Br- molar ratios of 930±66,
owing to the chemical fractionation during evaporation of sea droplets generated by winds. Cl/Br- molar ratio in bulk deposition controls the Cl -/Br- ratios in groundwater since the water
table in this coastal area is shallow and consequently the recharge water has a short residence
time in the unsaturated zone: physical processes taking place in soil (dilution,
evapotranspiration and mixing) do not significantly modify the Cl -/Br- ratio, but can change
the absolute concentrations of Cl- and Br- (Alcalà and Custodio, 2008; Edmunds, 2001). This
behaviour characterizes both shallow and deep groundwater; on one hand, in fact, P1 samples
show molar ratios up to 888, but with absolute concentrations of Cl- up to 18.700 mg/l and
Br- up to 47 mg/l, and, on the other hand, the samples from the deep portions of P33 and P34
show molar ratios up to 925, but with absolute concentrations of Cl - up to 45.000 mg/l and Brup to 110 mg/l. From the general trend of all the collected MLS samples, it is evident that
salinization is not due to mixing with present seawater, even in monitoring wells located close
to the coastline, like P1.
45
RESULTS AND DISCUSSION
In fact, the line representing the Adriatic Sea water Cl-/Br- molar ratio is located above and
below the trend individuated by groundwater samples (Fig. 4.9, left panel). Moreover, the Brversus Cl- plot (Fig. 4.9, right panel) has a good correlation, indicating a similar origin for the
two ions.
The Cl-/Br- ratio is constant during seawater evaporation. At the onset of halite or other
chloride mineral precipitation, bromides preferentially are left in residual seawater: under
continuous evaporation, while the Cl/Br ratio of the residual brine progressively decreases,
precipitated minerals develop elevated Cl/Br ratios (Stueber and Walter, 1991).
Consequently, leaching of evaporite would produce groundwater with Cl -/Br- ratio higher
than the seawater; on the contrary, mixing with brines will lead to Cl-/Br- ratios in
groundwater lower than seawater. The Cl-/Br- ratio less than 1000 shown by waters of high
salinity cannot be explained by only dissolution and leaching of salts: these values of the ratio
probably derive from both dissolution of salts and mixing with residual brines found in the
deepest part of the aquifer. They were most probably formed in back barrier inter-basins
during the Holocene transgression. The residual brines and evaporite salts have been
probably preserved from flushing being trapped into fine-grained sediments, thus forming
saline inter-layers, which in turn led to the formation of saline and hypersaline groundwater
(Lorenzen et al., 2012).
The inverted head gradient created by the constant pumping of the drainage network has
mobilized and slowly transported upward these saline waters preserved in the deepest
portion of the aquifer within low permeability sediments (Giambastiani et al., 2013).
Figure 4.9 - Plot of Cl-/Br- molar ratio versus Cl- (meq/l) for the groundwater samples (left
panel). Plot of Br- versus Cl-(meq/l) for the groundwater samples (right panel). Red lines indicate
the ideal mixing line between freshwater and present Adriatic seawater, while the red shaded
areas represent the mixing area.
46
RESULTS AND DISCUSSION
To confirm what discussed so far, figure 4.10 shows that evaporation plays an important role
in the evolution of the majority of groundwater samples collected both via MLS and IDS
techniques. In particular, samples collected in P33 and P34 are grouped in the top right area
of the Gibbs diagrams: this testifies a considerable role of evaporation in the geochemical
evolution of saline water that lead to TDS concentration up to 74 g/l. Samples from P1, P2, P8
and P10 also have a clear evaporative trend, especially evident in the deepest samples. On the
other hand, P4 and the top part of P5 and P7 samples fall in the water-sediment interaction
area of the Gibbs diagrams. These samples, affected by freshwater both from the Po River and
irrigation, have a weight ratio of Na++K+/(Na++K++Ca2+) up to 0.6 mg/l and a low weight ratio
of Cl−/(Cl−+HCO3−), which indicate that cation exchange with the clay minerals present in the
aquifer matrix played a role by increasing Na+ and K+ and decreasing Ca2+ under the
background of water-sediment interaction dominance. The low values of Cl−/(Cl−+HCO3−)
weight ratio are imputable to an excess of bicarbonate more due to the organic matter
mineralization processes, which cause an increase in alkalinity, rather than to the dissolution
of carbonates present in the aquifer matrix.
Figure 4.10 - Gibbs diagrams for cations (left panel) and anions (right panel).
4.2
The importance of data acquisition techniques in saltwater intrusion monitoring
The delineation of the results gained from the fulfilment of the first objective of this thesis, i.e
a detailed vertical characterization of the Ferrara coastal aquifer to monitor salinity
distribution. Different approaches exist to perform a depth-dependent characterization of an
aquifer.
47
RESULTS AND DISCUSSION
This paragraph illustrates the application of three different monitoring and sampling
techniques to the coastal aquifer of Ferrara: the open-borehole logging (OBL), the integrateddepth sampling (IDS) and the multi-level sampling techniques (MLS) via straddle packers. The
OBL led to a satisfactory reconstruction of the extent of the fresh-saltwater interface, but
provided a misleading characterization of the distribution of redox environments within the
aquifer. The IDS conceptual model individuates present seawater intrusion as the active
phenomenon, given that measured Cl- concentrations never exceed that of seawater and there
is no indication of hypersaline groundwater all over the study area. On the contrary, good fit
among sedimentological, stratigraphycal and physical-chemical data were obtained using the
MLS. In fact, in the MLS conceptual model the lower portion of the unconfined aquifer
presents hypersaline groundwater: this finding constrains to consider salinization processes
other than seawater intrusion, like the water/sediment interaction or the geochemical
evolution of groundwater due to evapo-concentration.
This study demonstrates that, within coastal shallow aquifers evenly recharged by irrigation
canals, the simple and economical OBL and IDS technique can lead to misleading results when
used to characterize density dependent groundwater stratification. Thus, results entail the
need of using MLS in reconstructing a reliable conceptual model, especially in areas where the
density contrast between fresh and saline groundwater is large.
4.2.1 Open-Borehole Logging and Multi-Level Sampling
Figure 4.11 shows the salinity profiles for the four monitoring wells along transect A-A’ (Fig.
2.2). Considering both OBL and MLS profiles, the aquifer displays fresh water (salinity <0.5
g/l) in the uppermost section of P3 and P8, from the top of the water table down to -9 and -6
m a.s.l. respectively, and a transition zone of brackish water (from 0.5 to 30 g/l) with a
thickness ranging from 1 to 4 m. Finally, saline (from 30 to 50 g/l) to hypersaline (>50 g/l)
water is present in the remaining thickness of the aquifer. Overall, the absolute salinity
residual for all the monitoring wells is 4.31 g/l and the R 2 is 0.92, which is deemed as
acceptable goodness of fit between the two techniques. Since the two different acquisition
techniques, i.e. the OBL and MLS methods, give approximately the same classification and
distribution of the water types in the aquifer, it follows that the OBL method can be
successfully used for a preliminary determination of the position of the freshwater/saltwater
interface in aquifers with normal density stratification (freshwater near the water table and
saline groundwater towards the bottom). Nevertheless, analysing in more detail the OBL
salinity profiles against the MLS profiles, some considerable differences are apparent. In P8
and P34 the OBL technique displays a somewhat constant concentration profile in the lower
part of the aquifer where hypersaline water is present. The MLS technique on the other hand,
is able to distinguish in both monitoring wells a considerable variability in the concentration
profiles. In particular, the salinity distribution in P8 slowly increases from saline (-8 m a.s.l.)
to hypersaline water (-17 m a.s.l.), showing salinities always lower than the ones recorded at
the same depths by OBL.
48
RESULTS AND DISCUSSION
In the hypersaline bottom part of the aquifer the deviation of the MLS profile from the OBL
profile is even more evident; in P34 the MLS profile shows a correlation between the salinity
distribution and the stratigraphy, witnessing water-sediment interactions. Accordingly, in
P33, the MLS profile demonstrates that the change in sediments reflects a change in salinity:
at -20 m a.s.l. salinity increases up to 53 g/l in correspondence with a peaty layer.
Figure 4.11 - Plots of salinity with MLS versus OBL data for each monitoring well along A-A’
transect (see Fig. 2.2 for location); stratigraphy (core logs data) is displayed on each plot to
assist the interpretation.
The misleading distribution of salinity values above seawater concentration recorded via the
OBL technique is thought to be due to both the mixing of water into the monitoring wells
induced by the density dependent flow and to the downward and upward movement of the
probe within the borehole casing during the monitoring procedure. These biases are
obviously avoided using the MLS techniques. As for the salinity values above seawater
concentration, the OBL and MLS techniques present a discordant picture of the physicochemical parameter profiles of the coastal shallow aquifer.
49
RESULTS AND DISCUSSION
Fig. 4.12 describes pH profiles collected via OBL and MLS techniques. The profiles are not in
good agreement, with an absolute residual for all monitoring wells of 0.37 and a R 2 value of
0.72. The pH measured by MLS is regularly higher than the one measured by the OBL
technique, even if in both cases the ambient groundwater is characteristic of the basic
environment. A sharp pH change is recorded with the OBL technique at the fresh-saltwater
interface. On the contrary the MLS profiles display smoother and more regular changes, which
can be directly related to recharge from the canals (having a typical pH around 8.4) and to the
stratigraphy (i.e. in P8 where the sudden increase in pH is imputable to the presence of a peat
layer). From Fig. 4.12 it is clear that the pH observed with the OBL technique reflects the pH
of the stratified water column accumulated in the monitoring well, but it is not strictly
representative of the aquifer pH.
Figure 4.12 - Plots of pH with MLS versus OBL data for each monitoring well along the A-A’
transect (see Fig. 2.2 for location); stratigraphy (core logs data) is displayed on each plot to
assist the interpretation.
50
RESULTS AND DISCUSSION
Fig. 4.13 shows the Eh profiles for the selected monitoring wells. Here the discrepancy
between the OBL and MLS technique is even more pronounced, with an absolute residual for
all the monitoring wells of 0.76 mV and a R2 value of 0.22.
The profiles are not only different with respect to the registered absolute values, but the
trends recorded by the two different techniques do not coincide as well. In P33 the most
negative Eh values coincide with sampling depths at which a strong smell of H 2S was detected
during the sampling procedure: thus it means that sulphate reduction probably takes place at
levels where labile organic matter is present. As for the previously discussed Eh values, the
agreement between OBL and MLS techniques in O 2 measurement is very low, with an absolute
residual for all the monitoring wells of 0.25 mg/l and a R2 value of 0.47.
Figure 4.13 - Plots of Eh with MLS versus OBL data for each monitoring well along the A-A’
transect (see Fig. 2.2 for location); stratigraphy (core logs data) is displayed on each plot to
assist the interpretation.
51
RESULTS AND DISCUSSION
In agreement with the measured negative Eh, the O 2 concentration is always very low in all
monitoring wells (Fig. 4.14). Generally, the O2 concentration recorded via MLS is always
higher at the top of the aquifer compared to concentration measured by the OBL technique.
Oxic conditions in the upper part of the aquifer are due to direct recharge from the canals
network, usually showing O2 concentration around 5 mg/l. O2 concentration recorded via MLS
decreases more rapidly with depth with respect to OBL, reaching anoxic condition in a range
fromn 1 to 3 m below the water table. In the same time anoxic values are recorded between 4
to 7 m below the water table with the OBL technique.
Figure 4.14 - Plots of O2 with MLS versus OBL data, for each monitoring well along the transect
A-A’ (see Fig. 2.2 for location); stratigraphy (core logs data) is displayed on each plot to assist
the interpretation.
The profiles of above reactive sensitive physico-chemical parameters (Figs. 4.12, 4.13 and
4.14) recorded via OBL and MLS techniques give a conflicting picture of the coastal shallow
aquifer, as for the salinity values above seawater concentration (Fig. 4.11). This means that
along the A-A’ transect, even if the two different acquisition techniques give approximately
52
RESULTS AND DISCUSSION
the same classification and distribution of the water types in the aquifer, the OBL method
cannot be employed to characterize in detail the effects due to interactions between surface
and ground waters. The OBL method demonstrates inappropriate in defining the trend in
redox processes occurring within the aquifer, especially in correspondence with sediments
bearing a high content of labile organic matter.
4.2.2 Integrated-Depth Sampling and Multi-Level Sampling
With respect to the comparison of IDS and MLS techniques, it is worth to note that the TDS
concentration in IDS samples never exceeds 40 g/l within the saline groundwater zone, thus
failing in the highlight of the presence of hypersaline water types at the bottom of the
unconfined aquifer (Fig. 4.15).
Accordingly, the IDS technique is not capable to address the high variability of Eh with depth,
thus preventing from recognizing the different terminal electron acceptor processes (TEAP)
taking place within the aquifer (Fig. 4.16). IDS technique displays Eh values usually higher
than MLS ones; an exception is given by the monitoring well P34, where silt and peat layers
constitute the first meters of the unconfined aquifer. MLS plots of Eh versus depth show a
general reducing condition into the aquifer with increasing Eh at increasing depth (Colombani
et al., 2015c).
Figure 4.15 - TDS versus Depth, blue circles indicate the MLS samples and red crosses indicate
the IDS samples. Water quality classes are drawn according to Davis and De Wiest (1996).
53
RESULTS AND DISCUSSION
The more superficial levels tend to be more oxygenated, becoming anoxic moving downward;
in particular this is evident in the monitoring well P5 that passes from the initial oxidizing
conditions to reducing conditions. This behaviour is caused by the presence in the sediments
of the coastal Ferrara plain of large quantities of organic matter due to the geomorphological
history of this deltaic environment (Stefani and Vincenzi, 2005).
In fact, it is generally observed in natural environments that the oxygenated water that enter
an aquifer rich in organic matter will first be depleted in oxygen content and nitrate,
subsequently iron is reduced and finally sulphate reducing condition and methane may
appear in the water (Appelo and Postma, 2005).
Figure 4.16 - Eh versus depth for selected monitoring wells. Blue circles indicate the MLS values
and red lines indicate the IDS values.
The Eh versus pH diagram (Fig. 4.17) can be a helpful tool to distinguish between different
TEAP. The main TEAP derived using MLS data are iron, sulphate reducing and methanogenic
conditions, which are commonly observed in aquifers affected by seawater intrusion
(Andersen et al., 2005). On the other hand, the IDS samples depict nitrate, manganese and
iron reduction as the main TEAP, disregarding sulphate reduction and methane production.
54
RESULTS AND DISCUSSION
Figure 4.17 - Eh versus pH plot. Grey shaded areas identify the main TEAP environments.
4.2.3 Discrepancy between an integral-depth and a depth-dependent monitoring
From the foregoing reasoning it is clear that IDS and MLS techniques often give conflicting
information. In order to quantify the differences, Table 4.3 and Table 4.4 report the
discrepancies between MLS and IDS values concerning the major ions concentrations in
groundwater and the aquifer physical-chemical parameters.
In Table 4.3, nitrate discrepancy has an average value of 141%, with a maximum of 187%; the
phosphate discrepancy has an average of 95%, with a maximum of 188%. For these reactive
species, the artificial mixing induced by the IDS technique within the long screen wells is
therefore detrimental. On the other hand, calcium discrepancy is lower than previous ones,
with an average value of 44% and a maximum of 92%. This behaviour is due to pH buffering,
which involves precipitation/dissolution processes of calcite and dolomite, abundant in the
Ferrara aquifer matrix. Chloride discrepancy shows a high value in P8 (150%) since the TDS
distribution registered by the MLS technique is extremely variable and increases
progressively with depth, from 0.9 mg/l at the aquifer to 61.8 mg/l at the aquifer bottom,
while the IDS technique just gives a TDS value of 37.2 mg/l throughout the whole depth of the
monitoring well.
55
RESULTS AND DISCUSSION
Table 4.3 - Discrepancy between MLS and IDS values concerning the major ions, the symbol – is
used when concentrations are below detection limit.
Chloride Sulphate
(%)
(%)
Nitrate
(%)
Phosphate
(%)
Sodium
(%)
Ammonium Calcium
(%)
(%)
P1
81±70
46±56
187±36
-
72±72
94±65
33±45
P2
63±72
167±110
154±30
-
65±78
125±124
63±37
P3
134±132
71±92
-
152±175
132±141 55±76
48±44
P4
14±8
46±61
-
36±38
9±10
25±22
9±6
P5
87±99
39±43
175±76
92±107
81±91
171±19
41±33
P7
50±64
42±48
109±120 89±107
57±70
86±101
27±31
P8
150±142
161±145
-
155±92
146±134 76±73
92±86
P9
19±17
80±25
94±76
188±176
19±23
46±59
32±5
P10
39±75
31±43
96±58
51±59
47±69
48±82
39±24
P33
19±24
56±75
-
16±24
15±22
24±32
15±22
P34
75±94
54±71
171±79
74±94
88±100
89±97
68±82
Table 4.4 - Discrepancy between MLS and IDS values concerning the physical-chemical
parameters
Eh
(%)
pH
(%)
Alk.
(%)
EC
(%)
DO
(%)
P1
56±14
2±4
23±28
72±66
12±5
P2
25±17
1±1
31±22
56±64
7±4
P3
59±36
1±1
28±46
142±60 14±6
P4
37±28
3±2
7±8
5±5
9±15
P5
101±110 2±2
36±34
65±72
14±9
P7
54±69
8±5
16±17
38±49
13±6
P8
139±2
2±4
14±18
85±104 21±27
P9
62±55
0±0
15±33
17±15
13±8
P10
44±12
1±1
46±69
37±69
4±9
P33
32±34
2±2
13±18
14±21
6±4
P34
18±22
3±4
12±6
69±83
2±7
56
RESULTS AND DISCUSSION
In Table 4.4, the most affected parameter is Eh with an average discrepancy of 57% and a
maximum of 139%. pH, dissolved oxygen and alkalinity show discrepancies from low to
moderate, mostly because they have a limited range of variability or are often below the
detection limit. The EC follows the trend of Cl - (Table 4.3) and its discrepancy is in general
high or moderate, except for P4 where freshwater is present throughout the aquifer vertical
profile.
Figure 4.18 - Cl- distribution maps, according to Stuyfzand classification (1989), derived using
IDS data (left panel) and MLS data from the upper aquifer portion (mid panel) and from the
lower aquifer portion (right panel).
Figure 4.18 shows the conceptual models of the Cl- distribution derived using the IDS data
(left panel) and the MLS data (mid and right panels). The IDS technique returns a conceptual
model that recognizes present seawater as the main salt source. This indication comes from
the following findigns: maximum Cl- concentration never exceeding that of sea water;
presence of saline groundwater (10.000-20.000 mg/l) up to 5 km from the shoreline, in
correspondence with the more recent sandy beach and dune deposits; brackish/saline
groundwater (1.000-10.000 mg/l) in the area characterized by fine interbay deposits;
brackish (300-1.000 mg/l) and fresh/brackish (<300 mg/l) groundwater moving towards the
Po River in the northern part of the study area. A lens of brackish water is also hosted by the
dunes system extending from P8 to P9 in proximity of the Comacchio Lagoon.
57
RESULTS AND DISCUSSION
Hypersaline groundwater (>20.000 mg/l) is not present in the IDS conceptual model and
therefore there is no need to evoke processes other than seawater intrusion. The remarkable
extension of the saline groundwater might be explained by the disturbance of the drainage
network that keeps the water table always below the Adriatic Sea level.
On the contrary, the MLS technique gives conceptual models of the top and bottom parts of
the aquifer both different from the IDS conceptual model and quite different each other. In
particular, the fresh/brackish groundwater area is much broader in the upper portion of the
aquifer than in IDS model, while the same area it is only confined in proximity to the Po River,
mainly within the sandy crevasse splay deposits in the lower portion of the aquifer. A wide
lens of fresh/brackish groundwater is also present in the central part of the study area in
correspondence to the maximum lateral extent of the dune system, testifying the effect of
recharge on the groundwater salinity. In the lower portions of the aquifer, the whole southern
part of the study area shows hypersaline groundwater that cannot be explained by present
seawater intrusion. This leads to consider other sources of salinization like water/sediment
interaction with the fine sediments that were deposited in a back-barrier environment during
the Holocene transgression, or geochemical evolution of groundwater due to a considerable
role of evaporation. Moreover, the Cl/Br molar ratio (Fig. 4.9, left panel), the Gibbs diagrams
(Fig. 4.10) and the Eh versus pH (Fig. 4.17) analysed so far, also support the hypothesis that
other sources of salinization play a role in the unconfined coastal aquifer of Ferrara.
On the light of above findings, it is clear that the IDS technique defines a simple but improper
distribution of the salinity, considering that the actual vertical stratification of salinity is
masked by the artificial mixing within the long screened wells. The conceptual model drawn
on the base of results of the MLS technique allows inferring a more detailed and reliable
distribution of salinity and identifying additional processes involved in the evolution of the
local groundwater.
4.3 Numerical modelling to predict salinization trends in a lowland coastal aquifer
To quantify the impact of groundwater salinization on the whole coastal groundwater system
in the Ferrara Province, a conceptual model was developed based on detailed topography and
bathymetry, stratigraphyc information from analysis of well logs, hydrodynamic information
from heads monitoring and pumping tests, and hydrogeochemical information from MultiLevel Sampling. On the basis of such a conceptual model, a three-dimensional variable-density
groundwater flow model coupled with solute transport was developed and calibrated with
MODFLOW-2000 using highly parameterized inversion techniques, to help ensure that the
model was a reasonable representation of the system. The calibrated model was employed to
investigate the extent of autonomous salinization via seepage of saline groundwater in the
Ferrara coastal aquifer.
To better study the interactions between the coastal unconfined aquifer and the canals
network, and to quantify the possible future effects of climate change on groundwater
salinization, a calibrated two-dimensional model was implemented using SEAWAT with a
58
RESULTS AND DISCUSSION
multiple scenario approach: (i) increase in evapotranspiration induced by temperature
increase; (ii) increase in the frequency of severe high rainfall events; (iii) extreme drought
conditions; (iv) and canal dewatering due to salinization of the water courses. The first three
scenarios had only a minor influence on the aquifer salinization rate, while the fourth one
predicted serious upward flux of the high salinity groundwater actually residing at the bottom
of the unconfined aquifer. The scenarios quantified the possible future effects on groundwater
salinization: they could be useful to find adaptation strategies to manage the water resources
of this and similar areas.
4.3.1 Three-dimensional model calibration and sensitivity analysis
The first step in the calibration process was to compare the outflow from the drainage
network calculated in the numerical simulation with the average discharge data from 2010 to
2013 supplied by the Consorzio di Bonifica Pianura di Ferrara (Tab. 4.5). In order to have a
direct comparison with the discharge data, different sub-drainage basins have been isolated,
according to the Consorzio zonation map (Fig. 4.19).
Table 4.5 - Comparison between the observed and calculated volumes of water (m3/s)
discharged to the sea by the 5 main dewatering stations.
BONELLO ROMANINA GIRALDA ACQUE ALTE ACQUE BASSE Total
5.88E-01
5.16E-01
1.19E+00
3.54E+00
8.03E+00
1.39E+01
Calculated 4.41E-01
4.77E-01
1.29E+00
5.71E+00
4.64E+00
1.26E+01
Observed
To achieve a good fit between simulated fluxes and observed ones, at first the drain
parameters were adjusted changing the conductance (m 2/s) and the elevation of the drains’
bottom (m) that went dry. In addition, to get a closer fit with the observed fluxes, it was
necessary to increase the recharge (from 80 mm/y to 95 mm/y), because the calculated
groundwater outflowing toward the drainage network continued to be insufficient. To assure
a good match between measured and observed hydraulic heads and to reach an acceptable
variance (Fig. 4.20), the location of drains was carefully checked, since they are the main
drivers of fluxes within the model domain.
Regarding salt concentrations, calibration and adjustment of transport parameters like
dispersivities were not necessary. In fact, after the rigorous calibration of flow parameters,
the scatter diagram between observed and calculated concentrations resulted satisfactory
(Fig. 4.20b). The model efficiency (Nash-Sutcliffe, 1970) for concentrations was 0.87, while for
hydraulic heads was 0.72. Given the regional scale of the model, the results were considered
acceptable (Colombani et al., 2015b).
59
RESULTS AND DISCUSSION
Composite scaled sensitivities (CSS) were used to perform a sensitivity analysis on the model
parameters. CSS are calculated for each parameter using the dimensionless scaled
sensitivities for all observations (Hill, 1998). Because CSS are dimensionless, they can be used
to compare the amount of information provided by different types of parameters. Model
simulation results will be more sensitive to parameters with large CSS.
Figure 4.19 - Subdivision of the three-dimensional model domain in drainage basins.
The CSS values indicate which parameter has large influence on the model results. If the CSS
value of a given parameter Pi is high, this parameter Pi is more sensitive to percentage
changes in its value, and the resulting modelled hydraulic heads depending on Pi are more
variable. It can be noticed that the hydraulic conductivity set for the top layers 2 has the
highest CSS; therefore the k value assigned to the model is the most uncertain, since a single
value was used for the entire aquifer (Tab. 4.6). Moreover, this parameter is highly correlated
with the drains conductance, since the drains are all located in the top layers. Secondly, model
results are also sensitive to recharge. Thus, to improve the model efficiency, an accurate
reproduction of the recharge should be introduced, for instance including seasonal variations
60
RESULTS AND DISCUSSION
instead of the single steady state value actually employed. In any case, all the CSS values for
the three-dimensional model have values somewhat low: therefore, any change in the
parameter set will have little effects on the result. The sensitivity analysis also shows that the
drain conductance, the evapotranspiration and the hydraulic conductivities of the lower
layers are relatively insensitive to the change of parameters.
Figure 4.20 - Scatter diagrams of calculated and measured hydraulic head (left plot) and of salt
concentrations (right plot).
Table 4.6 - Hydraulic heads composite scale sensitivity values calculated by PEST.
PARAMETERS
COMPOSITE SCALE
SENSITIVITY
Hydraulic conductivity_1 (layer 2-3)
0.1117
Hydraulic conductivity_2 (layer 4-5)
0.0758
Recharge
0.0671
Hydraulic conductivity_3 (layer 6)
0.0467
Drain
0.0343
Evapotranspiration
0.0199
Hydraulic conductivity_4 (layer 7-8)
0.0045
Hydraulic conductivity_5 (layer 9-10)
0.0003
61
RESULTS AND DISCUSSION
4.3.2 Three-dimensional model results
The piezometric map shown in figure 4.21 highlights that most of the aquifer displays head
values below mean sea level, except for the palaeo-dune system and the present dune system.
The most depressed hydraulic heads are found in the recently reclaimed lands located in the
South-western corner of the model domain (Mezzano).
A Three-dimensional isoconcentration map has been drawn to represent the computed salt
concentrations (Fig. 4.22). In the A-A’ transect the groundwater upward flux with high salts
concentration (up to 70 g/l) is fairly appreciable. This upward flux is responsible for the
salinization of the upper portion of the unconfined aquifer and for the discharge of saline
waters in the drainage network as well.
Figure 4.21 - Piezometric map of the calculated heads (m a.s.l.).
62
RESULTS AND DISCUSSION
Figure 4.22 – Three-dimensional isoconcentration map of the calculated salt concentrations,
along the transect A-A’.
A significant model output consists in the calculation of the salt masses entering and leaving
the model domain. Currently, the salt mass discharging from groundwater into the drainage
network equals 1714 t/y. The Po River recharges the unconfined aquifer with 0.3 t/y of salt,
effective rainfall (recharge) contributes with 48 t/y of salt, while evapotranspiration exports
from the model domain 26 t/y of salt. Converting the salt discharge in the drainage network
into a surface load, this mass flux is equal to 22.7 kg/ha/y, which is consistent with the salt
load found in the Netherland by a similar large scale modeling study (Oude Essink et al.,
2010).
4.3.3
Tidal forcing effects on groundwater fluctuation
To confirm that the main driver of groundwater salinization is the upward flux induced by
dewatering activities necessary to keep the coastal area dry, the effect of tidal fluctuation on
groundwater dynamics and consequently on seawater intrusion were investigated in P8,
which is the monitoring well nearest to the coastline (Fig. 2.2).
Fig. 4.23 shows the relationship between tidal fluctuation and groundwater level in the
monitoring well P8 from August 2009 to May 2010. It is clear from this plot that the tidal
amplitude is approximately 0.4 m, the tidal range is 0.8 m and the average tidal level is 0.18 m
a.s.l., with the highest sea levels occurring during winter time due to storm events (Martinelli
et al., 2010).
63
RESULTS AND DISCUSSION
The piezometric heads recorded in the monitoring well P8 meanwhile are quite stable in
summer and early autumn, with a sudden 0.5 m decline in November 2009 and a steep rise in
April 2010 following the irrigation/drainage canals regime. Overall, the maximum
piezometric head variation during the monitoring period was 0.83 m with a mean value of 1.80 m a.s.l..
The low piezometric head variation during the hydrologic year shows that the unconfined
aquifer is affected by the channel network, and the fact that values are always below mean sea
level confirms that the groundwater flow is always directed inland, favouring seawater
intrusion. The right panel in Fig. 4.23 indicates the results of the analysis during a period of
only one week, when the hydrometric level of the nearby canal was stable and precipitation
absent. In this period of time the correlation with the tidal fluctuation is more evident, with a
lag phase between peaks of 18-19 hours, although the average groundwater fluctuation due to
tidal forcing is only 0.02 m in P8.
Figure 4.23 - Plots of continuous piezometric head monitoring of monitoring well P8 (black line)
and sea level (red line). The left plot shows the entire monitoring period, while the right plot
shows a zoomed timeframe of one week when the tidal oscillation is more evident. Note that the
y scales have different ranges.
It can therefore be inferred that P8 and the nearby surface water level are not affected by tidal
fluctuations. For this reason, a two-dimensional flow and transport model was implemented
across P8 and Canale della Gronda (Fig. 4.24) to study the interactions between the coastal
unconfined aquifer and the canals network, and to quantify the possible future effects of
climate change on groundwater salinization.
64
RESULTS AND DISCUSSION
Figure 4.24 - Location of the study area and conceptual model of the A-A’ transect across
monitoring well P8 and Canale della Gronda, selected for the two-dimensional numerical model.
4.3.4 Two-dimensional model calibration and validation
Figure 4.25 shows the calibration results of measured versus computed TDS and hydraulic
heads in monitoring well P8 (at -4 and -8 m a.s.l.) and in the nearby drain ditch (Fig. 4.24).
Good calibration has been achieved for both heads and TDS throughout the simulated period,
showing the presence of high TDS value in the deep portion of the aquifer and the presence of
a shallow freshwater lens with evident freshening induced by the leaking canal. The inverse
parameters estimation procedure performed with PEST resulted in the absolute mean errors
between simulated and observed heads and TDS of 2.3 cm and 1.64 g/L, respectively (Table
4.7). These errors correspond to 4.0% of the observed piezometric range and to 3.7% of the
observed TDS range during the monitored period. The daily variability of recorded heads and
TDS is very high: as a consequence, a detailed temporal discretization would be required to
accurately reproduce all the variations induced by small changes in canal water level. In any
case, the model conveniently reproduced the piezometric head trend induced by recharge
events and canal water level changes.
65
RESULTS AND DISCUSSION
Figure 4.25 - In the left hand side plots: comparison of modelled hydraulic heads (blue lines)
versus observed hydraulic heads (grey lines) and their running average with daily period (black
lines), In the right hand side plots: comparison of modelled TDS (blue lines) versus observed TDS
(grey lines) and their running average with daily period (black lines). The observed water levels
and TDS in the drain ditch are plotted with black open circles.
66
RESULTS AND DISCUSSION
Table 4.7 - Absolute mean error between observed and calculated values of the calibrated and
validated models.
Simulation
Hydraulic heads
Concentration
absolute mean error (cm) absolute mean error (g/L)
Calibrated model 2.3
1.64
Validated model
2.01
2.3
In particular, figure 4.25 evidences the piezometric heads drop induced by the canal’s
dewatering after the growing season, which also induces the drain ditch dewatering.
The simulated TDS in P8 at -4 m a.s.l. followed the recorded decreasing trend, but it did not
reproduce the various observed sub-peaks: these were probably due to lateral TDS
heterogeneities or geochemical reactions that were not included in the simulation. Also the
TDS in P8 at -4 m a.s.l. followed the observed trend, with an initial decreasing trend due to
downward movement of the more saline groundwater displaced by freshwater coming from
the leaking canal; then, when the canal was dewatered saltwater raised again upward
diminishing the freshwater lens thickness. The hand collected data of both heads and TDS in
the drain ditch matched quite closely the simulated values.
Figure 4.26 shows the TDS spatial maps and TDS profiles along the monitoring well P8 used to
validate the numerical model. The model well reproduced the TDS profiles in both the
sampling rounds, with an absolute mean error between simulated and observed TDS values of
1.52 g/L for the sampling round of the 12th of December 2012, and 2.51 g/l for the sampling
round of the 26th of February 2013 in P8 (Table 4.7). These errors correspond to 2.4% and to
4.0% of the observed TDS range in the first and second sampling round, respectively. Looking
at the freshwater lens shapes of figure 4.26, it could be noted that the mixing zone between
saline and freshwater is quite sharp, approximately 2 m thick, and it moved upward of 1 m
when the canal was dewatered. The sharp gradient between saline and freshwater could be
justified by the relatively low value of vertical dispersivity found via inverse modelling (Table
3.1).
During the irrigation season (spring and summer) the canal stage is kept constant thanks to
the complex agricultural drainage system that controls the water level and guarantees
freshwater for crop irrigation.
After summer the brackish-freshwater interface is deeper and the freshwater lens maximum
thickness is about 4 m due to canal recharge. During winter and autumn the canal is not used
for irrigation purposes and left completely dry. During this period the freshwater lens tends
to shrink, becoming only 3 m thick, and it is preserved only by recharge events (Fig. 4.26). The
results of the validated model indicate the strong link between surface water bodies and
coastal aquifer; the Gronda canal artificially recharges the aquifer during the cropping season,
while during winter time the saltwater interface can migrate upward since the canal is kept
dry for hydraulic protection purposes.
67
RESULTS AND DISCUSSION
Figure 4.26 - In the upper panels the simulated TDS contours for 12/12/2012 (left panel) and for
26/02/2013 (right panel) are shown together with the location of P8. Profiles in P8 of modelled
(black lines) versus observed (black open circles) TDS used to validate the numerical model are
shown in the lower plots.
4.3.5 Future scenarios
The maximum thickness of the shallow freshwater lens in the current situation (TDS<2.5 g/L)
was then compared with the results of the four scenarios in order to investigate the change in
the brackish-freshwater interface depth and geometry. The comparison between different
scenarios allowed quantifing the possible future effects of climate change on groundwater
salinization in the coastal aquifer. In addition, the sub-scenarios exemplified the impact of the
projected low sea level rise (LSLR) and high sea level rise (HSLR) on saltwater upward flow
(Figure 4.27 and Table 4.8).
Figure 4.27 shows the actual average TDS contours at the end of the summer season and at
the end of the winter seasons (upper panels). The future scenarios were all plotted for the end
of the summer conditions, with projected LSLR in the left panels and projected HSLR in the
right panels. To help the reader, the 2.5 g/L TDS isoline was also plotted. In the first scenario,
the increase of precipitation and temperature based on the climate change projections for the
68
RESULTS AND DISCUSSION
Mediterranean region (CIRCE project) (Giorgi and Lionello, 2008; Gualdi et al., 2013) caused
only minor effects on the brackish-freshwater interface. The interface fluctuated about
0.4±0.15 m (Table 4.8) upward in winter with a small decrease in freshwater lens thickness,
especially in the portion of the aquifer close to Gronda canal. No significant change is shown
during the summer period (Fig. 4.27): this means that the increase in evapotranspiration is
basically compensated by the small increase in precipitation. Some remarkable differences
are visible between the LSLR and HSLR sub-scenarios (Fig. 4.27): there is an evident shrinking
of the freshwater lens in the HSLR case respect to the LSLR case. The lens shrinking is
accompanied by the consequent upward movement of high TDS groundwater.
Table 4.8 - Scenario modelling of the predicted variations of freshwater maximum thickness
respect to the actual situation, with sub-scenarios of projected low sea level rise (LSLR) and high
sea level rise (HSLR). Positive values indicate thinning of freshwater lens, while negative values
indicate thickening.
Simulation
Change of freshwater
maximum thickness (m)
LSLR
Change of freshwater
maximum thickness (m)
HSLR
Scenario 1
-0.4±0.15
-0.62±0.33
Scenario 2
0.12±0.05
0.26±0.06
Scenario 3
0.13±0.06
0.26±0.06
Scenario 4
1.15±0.6
1.15±0.6
The second scenario (Scenario 2 in Figure 4.27 and Table 4.8) did not significantly affect the
freshwater lens and aquifer salinization as well. The conditions remained basically those of
the calibrated model, meaning that severe high rainfall events will not improve the natural
recharge of the aquifer and will not prevent an upward movement of the saline groundwater
interface from the bottom. Once again, an evident shrinking of the freshwater lens in the HSLR
case respect to the LSLR is appreciable in figure 4.27. This is a quite important point to stress,
since the impact of the projected sea level rise seems much more important than the
projected climate changes for this zone. The same considerations apply also for the third
scenario, where extreme drought conditions were simulated. This supports the results found
in the first two scenarios, since even without recharge for a prolonged period the freshwater
lens thickness was only slightly changed compared to the base case (Scenario 3 in Figure 4.27
and Table 4.8).
69
RESULTS AND DISCUSSION
Figure 4.27 - In the upper panels the actual average TDS contours for the end of the summer (left
panel) and for the end of the winter (right panel) are plotted, while the future scenarios are all
plotted for the end of the summer conditions. The projected LSLR are shown in the left panels
and HSLR are shown in the right panels. The black line represents the 2.5 g/L TDS isoline.
70
RESULTS AND DISCUSSION
The major variations in freshwater lens maximum thickness are shown in the fourth scenario.
The interruption of water flow within the canal during the irrigation season, due to
hypothetical salinization of the nearby water courses, could create an upward movement of
the high salinity groundwater, which actually resides in the aquifer bottom. In figure 4.27 the
lower panels (representing the scenario with canal dewatering for 6 months) show very thin
freshwater lenses, with an almost complete salinization of the aquifer. In contrast with the
other scenarios, the sub-scenarios of LSLR and HSLR show very similar results, confirming
that the canal is the main driver of aquifer freshening. In the worst case of Scenario 4 (Table
4.8), where the canal interruption lasted for 9 months, the freshwater maximum thickness
decreases from 4.6 to 0.6 m in the summer, and from 4 to 0.9 m in the winter.
In general, the simulations indicate that an increase in temperature or an increase in the
severe high rainfall events have only a minor influence on the aquifer salinization rate, while
an interruption of the canal usage could create serious upward movement of the high salinity
groundwater currently residing in the aquifer bottom. Besides, the impact of the projected sea
level rise seems much more important than the projected climate changes; thus, future
studies will need a better delineation of forthcoming sea level rise at local scale.
4.4 Ammonium occurrence in a salinized lowland coastal aquifer
Recent studies suggested that climate changes can alter the present chemical-physical and
hydrogeological conditions in coastal aquifers, increasing the risk of transfer of contaminants
to the water cycle. For instance, the increase of autonomous salinization due to sea level rise
may even induce the remobilization of geogenic elements. To understand if the increased
ionic strength within the aquifer caused by salinization of the Ferrara coastal aquifer could
shift the chemical equilibrium in the water-soil-air continuum favouring
desorption/dissolution reaction, the contribution of water-sediment interaction and the
anthropogenic influence on groundwater quality have to be firstly distinguished. The
attention was focused on (i) nitrogen species (paragraph 4.4), since ammonium and nitrate
contamination is a well-known issue in the intensively cropped Po plain and on (ii) trace
elements (paragraph 4.5) because in general alpine sediments, that constitute the Po plain,
are known to be rich in some of these compounds.
To understand the spatial distribution and the possible sources of high ammonium
concentration in the unconfined coastal aquifer of the Po River Delta lowland (Italy), a
detailed characterization of inorganic nitrogen species was completed on 59 monitoring wells
(Fig. 2.2). The ammonium resulted the prevalent nitrogen inorganic species in groundwater,
and its concentration increases with depth and salinity. Very high ammonium concentrations
are found in coincidence with peaty sediments in the salinized anoxic aquifer and in the lowlying aquitard. Results from this study show that the elevated ammonium concentration
derives from the mineralization of organic matter present in fine sediments deposited in
palaeo-marsh environments.
71
RESULTS AND DISCUSSION
4.4.1 Water table elevation and ammonium mapping
The analysis of the water table elevation highlights that the hydraulic head is below the sea
level at all locations, with values up to -7 m a.s.l. The central part of the study area is
characterized by a stagnant zone drawing groundwater from both the western and eastern
sides, with head gradients ranging from 0.03% to 0.06%. This area corresponds to a NorthSouth drainage axis induced by dewatering pumping stations, which are used to reclaim
swamps that previously occupied the area.
Figure 4.28 - Location of the study area with the piezometric contour map (black lines); circles
represent the groundwater NH 4+ concentration at 1 m below the water table in the 59
piezometer investigated; in black colour the four selected piezometers along a flow line (dashed
line corresponding to A-A’ transect in Fig. 2.2).
72
RESULTS AND DISCUSSION
NH4+ classed concentration map of Fig. 4.28 shows values ranging from 0 to 2.5 mmol/l with a
mean value of 0.46 mmol/l. The higher concentrations are found in the reclaimed peaty land
in coincidence with lower piezometric heads, with a maximum NH 4+ concentration higher
than 2 mmol/l and a mean concentration of 1.4 mmol/l. The spatial distribution of NH 4+ and
groundwater elevation indicate a reducing environment in the lowland reclaimed area
consisting of peaty soils (Mastrocicco et al., 2011b): this is due to elevated TOC and stagnant
groundwater that often lead to hydrogen sulphide formation (not shown) even at shallow
depths, as the ones investigated by this monitoring network. In the area surrounding the
stagnant zone, N is still present, mainly in the form of NO 3- at much lower concentration than
NH4+ (Fig. 4.29). This behaviour can be explained taking into account the stability of the N
species as a function of the redox potential values distribution (Landon et al., 2011). The
eastern and western parts of the aquifer within the study area are characterized by an
oxidizing aerobic environment, due to the presence of coarser sediments with lower content
of TOC and higher O2 mass flux compared to remaining parts; as a result, away from the
stagnant zone, the dominant N species is the oxidized NO3- form. On the contrary, in the
central part of the study area, where an anaerobic reducing environment is present, the
prevailing form of N is NH4+.
Figure 4.29 - Distribution of NO3- and NH4+ concentrations for the 59 monitoring wells with
respect to redox values (Eh).
4.4.2 Core samples depth profiles
Core sample profiles of monitoring wells P33 and P34 show that sediment TOC and TLN are
very high in recent peaty layers (up to 30 mol/kg and 35 mmol/kg respectively) (Fig. 4.30).
73
RESULTS AND DISCUSSION
The elevated sedimentary organic matter content, rich in organic N, is likely to give rise to the
abundance of NH4+ in the aquifer-aquitard system via organic matter mineralization, as
recently proposed by Jiao et al. (2010).
Figure 4.30 - Depth profiles of sediment TOC, TLN and NH4+; stratigraphy (core logs data) is
plotted on each graph to facilitate the interpretation.
Figure 4.31 - Transect along the selected flow line (refer to Fig. 4.28 for the location) with
concentration profiles of NH4+, NO3- and NO2-; unconfined aquifer thickness (shaded grey area),
water table (blue line) and core logs stratigraphy are also shown. Note that curves are plotted at
different horizontal scale to facilitate the graph interpretation.
74
RESULTS AND DISCUSSION
NH4+ in the sediments increases with depth, with spikes corresponding to the peaty layers,
while the concentration is very low within the coarse sand layers because of the low cation
exchange capacity (CEC) of sandy sediments. On the contrary, higher CEC values of clay and
organic layers favour the adsorption of NH4+, which is not easily displaced from the exchange
sites. This is even confirmed by the groundwater concentration profiles reported in Fig. 4.31
showing high NH4+ concentration within the sandy units and low values where the peaty
layers trap NH4+.
4.4.3 Groundwater quality depth profiles
Considering the water samples collected along the flow line (corresponding to the sequence of
P3, P8, P33 and P34 in Fig. 4.28), the main result of inorganic N speciation is that in all cases
NH4+ is the prevalent species in all the four monitoring wells and NH 4+ concentration
increases with depth (Fig. 4.31). The highest NH4+ concentrations are found in the aquitard,
located at the base of the unconfined aquifer. NO 3- is below detection limits except in the
upper part of the aquifer where NH4+ is very low. Here NO3- is present at low concentration,
especially in the zone where fresh recharge water infiltrates the aquifer causing nitrification
processes. According to Equation (15), the classical nitrification stoichiometry states that all
available NH4+ is transformed into NO3- (Appelo and Postma, 2005) in presence of oxic
conditions:
NH4+ + O2  NO3- + 2H+ + H2O
(15)
Mass balance for Equation (15) shows that NH4+ consumption seems to be sufficient to explain
NO3- increase at the top of the aquifer (Fig. 4.31), at least for monitoring wells P8, P34 and
P33; P3 shows instead a not clear trend, with other possible N sources, such as dry deposition
or fertilizer leaching. In addition, this reaction produces acidity that is largely buffered by the
elevated carbonate presence in these sediments with more than 20% of carbonates (Amorosi
et al., 2002). Accordingly, the measured average pH in the four monitoring wells is 8.4
(ranging between 7.4 and 9.4), thus witnessing basic condition within the aquifer-aquitard
continuum. The elevated pH could also lead to NH3 generation and volatilization, although this
species was not measured. Since the study area is all subjected to intensive agriculture, the
recorded NO3- and NH4+ concentrations in the shallow part of the aquifer could be also due to
fertilizer leaching. Although, recent studies have shown that NH4+ is rapidly nitrified in the top
soil and NO3- leaching is usually limited by denitrification, the latter is stimulated by the
addition of fresh and reactive organic matter (chicken manure) in these recharge areas to
improve the soil fertility (Mastrocicco et al., 2012a).
Fig. 4.32 shows TDS and NH4+ profiles for the four monitoring wells along the selected
transect (Fig. 4.28). The aquifer displays fresh water type (TDS <0.014 mol/l) only in the
upper section of P3 and P8 (approximately 5 to 7 m thick), and a transition zone of brackish
water type (from 0.014 to 0.86 mol/l) with a thickness ranging from 4 to 6 m in P3, P8 and
75
RESULTS AND DISCUSSION
P34, and for the whole depth in P33. Saline (from 0.86 to 1.43 mol/l) to hyper saline (>1.43
mol/l) water types are present in the remaining thickness of the aquifer in P8 and P34.
Groundwater in the coastal unconfined aquifer of the Ferrara Province often displays very
high salinity (>0.2 mol/l), which is likely due to the presence of palaeo-seawater rather than
to modern seawater intrusion. The latter is also unlikely to happen both because groundwater
salinity in this area is often more than twice higher than present seawater concentration (Fig.
4.32), and because over the past few decades groundwater has been only minimally exploited
due to its elevated salinity.
Figure 4.32 - Transect along the selected flow line (refer to Fig. 4.28 for location) with
concentrations profiles of NH4+ and TDS; unconfined aquifer thickness (shaded grey area), water
table (blue line) and core logs data are also shown. Note that curves are plotted at different
horizontal scale to facilitate the graph interpretation.
The upper part of monitoring well P8 displays low TDS and NH4+ concentrations since is
located 10 m from a canal, which contributes to dilute groundwater, and it has TDS and NH 4+
concentrations of 0.019 mol/l and 0.013 mmol/l, respectively. Monitoring well P3 is located in
a palaeo-dune topographically higher than the nearby sediments, thus being a recharge zone
where infiltrating rain creates a freshwater lens floating over the saline groundwater. This
explains the low TDS and NH4+ concentrations recorded at the top of the P3 profile.
Fig. 4.32 also shows a close relation between TDS and NH4+ increases in groundwater. In fact,
although the correlation between TDS and NH4+ is low (R2= 0.28), at elevated ionic strength
(like in P34 and P3) a sharp increase of NH4+ is present. Elevated NH4+ concentrations are
often associated with high groundwater salinity, and recently it has been suggested that
saltwater environments are favourable to the release in groundwater of NH4+ adsorbed onto
sediments (Bratton et al., 2009; Jiao et al., 2010).
76
RESULTS AND DISCUSSION
4.4.4 Groundwater mineralization processes
A comparative analysis of salinity and strontium (Sr 2+) has been performed to further
elucidate the processes that lead to such high NH4+ concentration in both sediments and
groundwater. The left graph in Fig. 4.33 shows that almost all deep samples have a higher
salinity and Sr2+ with respect to the composition of present Adriatic seawater: this suggests
the presence of ancient salt fluids rather than modern saltwater intrusion or mixing
processes.
Figure 4.33 - Salinity versus Sr2+ concentration in deep (from -25 to -9 m a.s.l.), medium (from -9
to -7 m a.s.l.), and shallow (from -7 to +1 m a.s.l.) samples (left graph); grey bars represent
salinity and Sr2+ concentration of the Adriatic seawater. NH4+ versus Sr2+ concentration in deep
(from -25 to -9 m a.s.l.), medium (from -9 to -7 m a.s.l.), and shallow (from -7 to +1 m a.s.l.)
samples (right graph); grey bars represent NH 4+ and Sr2+ concentration of the Adriatic seawater.
This indicates an enrichment of major and minor elements in deep groundwater samples and
consequently a longer transit time of these ground waters with respect to the shallower ones.
Only two deep samples fall on the lower left corner of the graph within the area of low salinity
and Sr2+ concentrations. These two samples belong to monitoring well P3 located within a
recharge area, and their salinity and Sr2+ concentration values are probably due to mixing
processes between fresh recharge water and old saline groundwater (Fig. 4.32). Additional
age tracer studies, e.g. 87Sr/86Sr, are needed to further refine the residence time of water.
The left graph in figure 4.33 shows the presence of some samples (belonging to monitoring
wells P33 and P3) with elevated Sr 2+ concentration and salinity slightly below than the actual
seawater value: these are possibly due to water/sediment interactions. Medium samples lie
between the shallow and deep samples, as expected. The shallow samples are all grouped in
the lower left corner, clearly meaning fast travel time and young groundwater.
77
RESULTS AND DISCUSSION
NH4+ is plotted versus Sr2+ concentration in the right graph of Fig. 4.33. Again the deep
samples show elevated Sr2+ concentrations, this time combined with elevated NH 4+
concentrations. The shallow samples are the less mineralized and the medium samples lie in
between. In particular, two distinct groups can be recognized: (i) high NH4+ - high Sr2+ group
mostly referring to deep samples, and (ii) low NH4+ - high Sr2+ group referring to deep
samples as well. Each of these two groups could define a different end-member along two
different mixing lines drawn to (iii) the low NH4+ - low Sr2+ group of shallow samples. The first
group includes P33 and P34 deep samples, which belong to lagoon and back barrier
sediments usually enriched in TOC. The second group includes P8 and P3 deep samples, which
belongs to beach ridges and prodelta sediments usually depleted in TOC respect to lagoon and
back barrier sediments (Amorosi et al., 2003).
These patterns are indicative of elevated NH4+ contents of natural origin: they are not
ascribable to fertilizer loading, because the elevated NH 4+ concentration are found in strict
association with the more mineralized and presumably older groundwater, and because the
known type and amount of fertilizers applied in the study area are not sufficient to explain
NH4+ concentration measured in deep groundwater samples (Mastrocicco et al., 2011b).
4.5 Trace elements mobility in a saline coastal aquifer of the Po river lowland
The effect of elevated salinity on the mobility of trace elements (TE) was studied within the
shallow coastal aquifer. The use of intensive depth profiles gave insights into groundwater
and sediment matrix interactions. The total concentration of TE within the aquifer matrix was
higher in the fine sediments, since usually TE concentrate in the clay fraction, while
groundwater concentrations generally increased with the ionic strength of the solution.
The only exceptions were V and As that did not followed this trend, being most probably
released by fertilizer leaching. The increase of salinity in conjunction with a high content in
carbonates stimulated a release of Pb and Zn, suggesting Pb and Zn surface complexation on
carbonate and dolomite minerals. In addition, the elevated organic carbon content of the
sediments induced reducing conditions throughout the aquifer, which further promoted TE
dissolution.
4.5.1 Sediment composition
The chemical composition of the cores P33 and P34 is reported in Table 4.9, whereas in Table
4.10 are reported the calculated parameters values along with the previously published SOM
and clay content (Mastrocicco et al., 2012c).
The values that have been found are in agreement with previous studies of soils (Amorosi and
Sammartino, 2006) and sediments (Amorosi et al., 2002; 2004; Curzi et al., 2006) in the area:
in particular, these soils and sediments are characterized by high concentrations of Cr and Ni
related to the provenance of the sediments (Po river contribution).
78
RESULTS AND DISCUSSION
Figure 4.34 shows the clear positive trends between the clay content of the sediment samples
and the selected TE, that highlight a typical enrichment in the fine fraction often reported in
literature (Loring, 1991; Dinelli et al., 2007; Dou et al., 2013): this is displayed not only by the
trends of V and Rb, but also by those of Zn and Pb (Fig. 4.34) and other possible hazardous
metals, like Co, Ni and Cu. On the contrary, the enrichment trend in peat and SOM is negligible,
indicating that TE are not adsorbed or complexed onto SOM.
Table 4.9 - Chemical composition of samples from P33 and P34. Depth is expressed as relative to
mean sea level.
79
RESULTS AND DISCUSSION
Table 4.10 – Derived parameter values in the sediment samples collected from P33 and P34 (in
italics). SOM and clay content reported from Mastrocicco et al. (2012c). Depth are expressed as
relative to mean sea level
Sample Depth FeS2
Fepy
FeTr
FeReact PRC
CBPO
SOM* Clay*
i.d.
(m)
(wt%) (wt%) (wt%) (wt%) (meq/kg) (mmol/kg) (wt%) (wt%)
P33-01
P33-02
P33-03
P33-04
P33-05
P33-06
P33-07
P33-08
P33-09
P33-10
P33-11
P33-12
P33-13
P33-14
P33-15
P33-16
P33-17
P33-18
P33-19
P33-20
P33-21
P33-22
P33-23
P34-01
P34-02
P34-03
P34-04
P34-05
P34-06
P34-07
P34-08
P34-09
P34-10
P34-11
-4.9
-5.7
-6.1
-6.4
-6.7
-7.5
-8.6
-9.4
-10.5
-12.5
-13.7
-14.2
-14.8
-16.5
-18.8
-19.4
-20.1
-20.9
-21.6
-23.2
-23.9
-25.2
-25.9
-3.8
-4.6
-6.5
-7.5
-8.5
-9.5
-10.5
-12.5
-14.1
-15.5
-16.5
0.04
0.04
0.04
0.76
8.56
2.46
2.28
2.41
0.70
0.41
0.61
0.10
0.37
0.37
0.65
0.88
0.04
1.63
0.84
0.90
0.49
1.71
3.15
2.44
2.13
1.02
0.66
0.43
0.60
0.40
0.21
1.35
0.29
0.18
0.02
0.02
0.02
0.35
4.00
1.15
1.07
1.12
0.33
0.19
0.28
0.05
0.18
0.17
0.30
0.41
0.02
0.76
0.39
0.42
0.23
0.80
1.47
1.14
0.99
0.47
0.31
0.20
0.28
0.19
0.10
0.63
0.14
0.08
1.31
2.39
1.97
1.90
1.48
1.77
1.84
0.80
0.98
0.04
1.52
1.03
1.07
0.96
0.96
1.34
1.73
2.02
1.02
1.34
0.83
1.77
3.56
2.59
2.02
0.81
0.16
0.16
0.20
1.13
1.06
2.58
1.02
1.67
1.30
2.37
1.95
1.54
0.00
0.62
0.77
0.00
0.65
0.00
1.23
0.98
0.90
0.79
0.66
0.93
1.71
1.26
0.63
0.92
0.60
0.97
2.09
1.45
1.02
0.34
0.00
0.00
1.00
0.95
0.97
1.95
0.88
1.59
80
1462
2085
4155
12397
52403
7479
6902
6882
1940
1041
2904
1193
1921
1996
2920
3250
2757
26888
5377
5273
3910
5957
7584
8269
7784
2375
1557
1435
1705
1448
3617
8767
3617
4671
2707
2410
2283
1187
100
1709
2099
100
1893
1126
2228
2447
2708
2528
2829
3636
3278
2786
3701
3293
4077
2774
100
1052
2191
2059
1462
1592
1635
3530
3537
2843
3270
2602
2.1
3.1
6.2
17.2
62.5
6.6
6.1
5.8
1.6
0.8
3.2
1.6
2.2
2.3
3.2
3.2
4.1
37.3
6.5
6.2
4.9
5.7
5.4
7.8
7.7
1.7
1.1
1.3
1.4
1.4
5.0
10.6
4.9
6.7
21.4
30.1
52.9
83.0
5.0
28.2
40.8
8.8
5.5
1.1
13.9
1.0
13.3
8.6
6.2
5.5
25.8
44.6
15.0
28.2
5.3
28.6
69.9
55.7
42.1
1.5
0.6
1.4
2.4
5.6
5.7
10.3
4.9
37.4
RESULTS AND DISCUSSION
Figure 4.34 - Scatter plots of selected TE in sediments versus clay content and their relative
regression coefficients (R2) for linear fitting.
4.5.2 Hydrochemical conditions of the shallow aquifer
The studied groundwater is characterized in P34 and P35 boreholes by extremely elevated
salinity (Fig. 4.35), which is well above the present Adriatic Sea salinity. Reducing conditions
are observed throughout the aquifer thickness, as testified by the low Eh (Fig. 4.35) typical of
Fe and SO42- reduction in carbonate buffered systems (McMahon and Chapelle, 2008). The pH
values are comprised between 8.4 to 9.4 in P34 and 8.2 to 8.9 in P33. The principal electron
acceptors like O2 and NO3- are not present except for the first 1-2 meters of aquifer
(Mastrocicco et al., 2013b).
4.5.3 TE distribution in the shallow aquifer
Table 4.11 reports the TEs concentrations throughout the aquifer profile in P33 and P34. In
general, the TE linked with saltwater intrusion (B, Sr and Ba) increase coherently with the
salinity (Colombani et al., 2014). The maximum values of these saltwater intrusion markers
are well above the actual mean ocean composition, with B up to 8.8 mg/l, Sr up to 16.5 mg/l
and Ba up to 308 µg/l.
81
RESULTS AND DISCUSSION
Figure 4.35 - Depth profiles of TDS, Eh and selected redox species in groundwater and sediments;
stratigraphy (core logs data) is reported on each plot to facilitate the interpretation.
82
RESULTS AND DISCUSSION
Table 4.11 - TE concentrations in groundwater samples from P33 and P34. Depth is expressed as
relative to mean sea level.
The reported mean composition of B, Sr and Ba for present ocean seawater is 4.5 mg/l, 8.8
mg/l and 21 µg/l, respectively (Turekian, 1968). The quite high Mn and Fe just reflect the
reducing conditions present in the aquifer, as explained in the previous section. Another
important redox sensitive TE is As, which is often present at remarkable concentrations,
exceeding 20 µg/l in most of the samples. In general, some of the analysed TE (as Al, Ti, Zn and
Pb) display high or very high concentrations for groundwater. . Their trends will be analysed
more in detail in the next sections.
4.5.4 Trace elements profiles in sediments and groundwater
Fig. 4.36 shows a few selected profiles of TE in both sediment and groundwater..
83
RESULTS AND DISCUSSION
Figure 4.36 - Depth profiles of TE in groundwater and sediments; stratigraphy (core logs data) is
reported on each plot to facilitate the interpretation.
84
RESULTS AND DISCUSSION
The V concentration in sediments follows the grain-size pattern (see Fig. 4.36) with the lowest
values in the most permeable layers in both the boreholes P33 and P34. In P34 the V in
groundwater (concentrations ranging from 28 to 156 µg/l) shows a marked peak, probably
due to agrochemical leaching, at the top of the aquifer,, and stable concentrations downward
despite the elevated salinity. In P33 V shows various peaks not clearly related to salinity or Eh
variations, but in general with higher concentrations than in P34.
V could be sensitive to changes in redox conditions. It has two prominent redox species: V(V),
which prevails under oxic conditions, and V(IV), which prevails in reduced environments
(Breit and Wanty, 1991).
Both species are surface reactive, but V(IV) will be adsorbed far more strongly in natural
waters: consequently, V is thought to be removed from solution under reducing conditions
(Emerson and Huested, 1991). The elevated salinity and the organic ligands present in this
reducing groundwater keep V in solution.
Zn, as already shown, is concentrated in the fine grained sediments; in groundwater
(concentrations ranging from 33 to 667 µg/l) Zn displays an enrichment in the fine sediments
comparable to V enrichment and tends to increase with depth in both P33 and P34, although
with a pronounced vertical variability.
Cr displays the same pattern than Zn in the sediments, but in groundwater (concentrations
ranging between 1 and 140 µg/l) it shows peaks (both in P33 and P34 wells) both in fine
graded sediment and coarse sand sediments.
Pb is below detection limits in some sandy sediment; in silt and clay it never exceeds 25 ppm,
while in groundwater (concentrations ranging between 1.6 and 37.6 µg/l) Pb increases with
ionic strength of solution, showing peaks in coincidence with peaty layers. All the selected TE
evidence a hundred fold increase in concentration compared to concentrations in mean
seawater (Turekian, 1968): this occurrence witnesses the high degree of interaction between
liquid and solid phases.
4.5.5 Trace elements partitioning between groundwater and sediments
The direct comparison between sediment and groundwater samples is a quite efficient way to
analyse the interaction between liquid and solid phases.
First of all, a correlation matrix can be constructed to evaluate which element or sediment
property best matches with other variables. To be concise, only the high correlation degrees
will be reported and only Pearson coefficients above 0.6 or below -0.6 will be discussed.
The analysis gives some insights into the interaction between dissolved TE, major ions in
groundwater and the aquifer matrix: for example, B present in groundwater is positively
correlated to the dissolved Na (0.84) and salinity (0.79), as usually found in saline aquifers
(Mondal et al., 2011; Giménez Forcada and Morell Evangelista, 2008); in addition, B displays
the highest affinity for dissolved Mg (0.87), however witnessing the role of dolomite
dissolution in TE abundance in groundwater. B also correlates quite well with dissolved Ti
(0.80), Ga (0.78), Sr (0.78), Ba (0.72) and Pb (0.75).
85
RESULTS AND DISCUSSION
As mentioned in the previous section, B concentrations in groundwater largely exceed
average B seawater concentration (Turekian, 1968): this is consistent with the very high TDS
displayed by depth profiles of both the monitoring wells P33 and P43 (Fig. 4.35). Al in
groundwater shows a week positive correlation with salinity (0.62) and with dissolved P
(0.68), suggesting that its mobility and the elevated concentrations found in groundwater
(Table 4.11) are strongly linked with aqueous complexes and formation of ultrafine colloids.
To prove this hypothesis, Al was analysed in a few selected duplicates of groundwater
samples filtered through 0.22 µm polypropylene cartridges: the Al concentration dropped
below 0.1 mg/l. The same behaviour was observed for V, Cr, Co, Ni, Cu and U with
concentration decreases of more than 80%; all the other TE groundwater concentrations
were unaffected by filtration. These findings point to an important control of colloids in the
occurrence of TE in groundwater.
P concentration in groundwater is correlated with dissolved K (0.77) and Na (0.73): the most
probable source of this nutrient, like for the ammonium, is the decomposition of deep buried
SOM (Mastrocicco et al., 2013b). The dissolved Sc has positive correlations with dissolved Mo
(0.82) and Cu (0.72), suggesting a common origin of these TE in this saline aquifer. Ti in
groundwater is well correlated with dissolved Sr (0.86) and dissolved Co (0.75): Ti
concentrations in groundwater are quite elevated and, as mentioned for B, well above the
seawater concentration, suggesting dissolution from sediment in reducing alkaline-saline
conditions. The filtering through 0.22 µm polypropylene cartridges did not produce
appreciable decrease in Ti concentrations. The dissolved Cu is positively correlated with Sr
(0.81) and Mo (0.82): in general dissolved Cu concentrations are not elevated even at high
salinities (Table 4.11).
On the contrary, Zn groundwater concentrations are much higher, with concentration peaks
of more than 400 µg/l (Fig. 4.36 and Table 4.11). Dissolved Zn is weakly correlated with
dissolved Fe (0.65), suggesting a release of this metal in iron reducing conditions. The
dissolved As is weakly correlated with most of the TE, except dissolved V (0.66), suggesting a
possible agricultural origin of Zn from fertilizers (Otero et al., 2005) beside the sedimentary
origin. Surprisingly, As is not correlated with pyrite or SOM as it is usually found (Wang et al.,
2012), although the As groundwater concentrations are generally high throughout both
profiles (Table 4.11). The quite elevated (up to 38 µg/l) dissolved Pb concentrations in
groundwater are positively the Fereact (0.72) meaning that Pb in groundwater is possibly
sourced by iron hydroxides in saline conditions; in addition, Pb correlates with dissolved Mg
(0.72), indicating that dolomite is a possible source for the elevated Pb observed groundwater
concentrations as well (Table 4.11).
86
SUMMARY AND CONCLUSIONS
5. SUMMARY AND CONCLUSIONS
5.1 Synthesis
The primary objective of this thesis was to gain a quantitative understanding of the processes
that caused groundwater salinization in the Ferrara coastal aquifer. In this final chapter, the
main outcomes of the applied multi-disciplinary approach will be evaluated to determine to
what extent they contribute to improve current knowledge. Some implications for practical
hydrogeological studies will be discussed and research topics needing further understanding
will be outlined.
Finally, a list of the related publication is supplied at the end of this chapter (paragraph 5.4) to
provide the reader with more detailed information concerning the different aspects on which
this thesis has focused over the last three years.
5.2 Conclusions
A detailed hydrodynamic and hydrogeochemical characterization of the Ferrara coastal
aquifer, subject to profound natural and anthropogenic changes since the last glaciation to
nowadays, demonstrated that groundwater salinization is a factual issue for current and
future management of water resources (see chapter 4.1).
Saline and hypersaline water are associated with fine sediments of back barrier and lagoonal
environments, relic of Flandrian transgression, while brackish and fresh waters are related to
relict dune-beach systems. Hypersaline groundwater found in the deepest portion of the
aquifer cannot be associated with lateral seawater intrusion or upconing of present seawater:
its features suggest the presence of relict seawater and salts trapped in the back-barrier
dunes and lagoonal environments during the Holocene interglacial phases. Relic seawater and
salt deposits in fine grained sediments could have been preserved from rapid flushing and
gradually released, leading to the formation of hypersaline groundwater.
Field data show that the development of anomalously-high salt contents (i.e., greater than in
seawater) may be attributed to direct evaporation of seawater prior to infiltration. Even the
high Cl- concentrations in groundwater from the coastal area can be attributed to this process.
Depth-dependent physical aquifer parameterization highlights the presence of an upward
vertical gradient due to the dense drainage system that controls the hydrodynamics of the
coastal area. As a consequence, saline groundwater previously preserved in the deepest
portion of the aquifer, and often confined by low-permeability lenses, has been mobilized and
transported via seepage to the shallow part of the aquifer and even discharged in the
superficial network of irrigation canals.
The hydrogeological and hydro-geochemical investigations suggest that aquifer salinization
does not result in a linear process. In fact, mixing between freshwater end-members and relict
87
SUMMARY AND CONCLUSIONS
seawater causes the onset of several concurring and competing water-sediment interactions
as cation exchange, calcite and dolomite dissolution, oxidation of organic matter, and SO 42bacterial reduction, which overlap to salinization. Deviations from the non-reactive mixing
line are observed, indeed, for the main ions (Br-, Na+, SO42-, Ca2+, and Mg2+) especially in the
most saline samples. The more saline groundwater is characterized by the effects of cation
exchange and SO42- reduction where fossil organic matter is present (paleo-marsh deposits,
peat layers). The whole of the processes results in the increase of Ca2+ and HCO3- contents in
the ground waters. The increase in Ca2+, Mg2+, and HCO3- shifts the carbonate equilibrium and
leads to dolomite/calcite precipitation.
Experience in the field revealed that a detailed vertical characterization is the sole approach
that can return a reliable distribution of dissolved species and geochemical patterns within
the unconfined aquifer. For this reason different monitoring techniques were applied to
acquire depth-dependent profiles of the physical-chemical parameters: they were compared
in order to find the most reliable and effective one (see chapter 4.2).
The comparison between the Open-Borehole Logging (OBL) technique and the more robust
(but more expensive and time consuming) Multi-Level Sampling (MLS) technique shows that,
excluding the distribution of saline to hypersaline water types at the bottom of the aquifer, no
significant differences are observed between salinity data recorded by means of the two
techniques. It follows that, when the density stratification is from denser groundwater at the
bottom to less dense groundwater towards the water table, the OBL method can be
successfully used for a preliminary determination of the position of the saltwater wedge and
for a general classification of the water types present in the aquifer. However, hydrochemical
data (pH, Eh and O2) collected with the OBL technique do not agree with data acquired by the
MLS technique. As matter of fact, more detailed studies on the reactive geochemical processes
involving micro and meso-scale structures, like peat layers, can be undertaken only through
MLS technique and. this technique is also preferable to OBL when a high level of accuracy in
depicting salinity distribution is required to infer recharge from surface water bodies.
A direct comparison between Integrated-Depth Sampling (IDS) and MLS performance was
also performed. The conceptual models based respectively on the IDS and the MLS monitoring
techniques return different representations of groundwater salinization and therefore of the
causes that led to it. The disagreement in the results gained by the two sampling techniques is
particularly evident when analysing the TDS versus depth plot, the Cl -/Br- molar ratios and
the Eh versus pH diagram. Discrepancies between MLS and IDS values concerning major ions
concentrations and physical-chemical parameters give a quantitative insight on this
disagreement, especially in the case of the redox-dependent dissolved species, testifying that
in the IDS technique the artificial mixing within the long well screens masks the actual vertical
stratification along the aquifer thickness. The results demonstrate that the IDS technique is
not only inaccurate if compared with the MLS technique, but it might also give misleading
representations of reality.
Therefore, conceptual models gained via IDS technique cannot be considered in a preliminary
study for a coastal aquifer affected by salinization, but they should be decisively avoided,
especially when the density contrast between fresh and saline groundwater is large.
88
SUMMARY AND CONCLUSIONS
The current situation of the Ferrara coastal aquifer was reconstructed through the use of
numerical modelling, to confirm the initial hypothesis on the origin of salinization as well (see
chapter 4.3). The effects of aquifer autonomous salinization via upward flux and those of the
consequent salt loads to surface waters were calculated with a three-dimensional variabledensity numerical flow model coupled with a non-reactive solute transport model.
Results show that since seepage water is brackish to saline, salinization of shallow
groundwater and surface waters is by now considerable and discharge of salt loads in the
superficial network of irrigation canals would intensify in the near future. This process, that is
the result of the drainage of lands over the past years, will affect surface water quality and
reduce shallow fresh groundwater volumes. Thus, in the future, fresh water resources for
agricultural purposes and ecosystems will come under increasing pressure.
A two-dimensional numerical model was implemented along the selected transect A-A’ to
help deepening the understanding of the key factors that control the hydrodynamic behaviour
and the salinity in the local aquifer with respect to a nearby drainage canal.
Results show that the current irrigation scheme governs the groundwater fluxes. In spring
and summer the canal used for agricultural irrigation recharges the aquifer, thus diminishing
the shallow groundwater salinity also in case of temperature increase (and associated
evapotranspiration), as forecasted by future regional climate change. On the contrary, during
the cold season the canal is no longer used for irrigation and consequently the freshwater
recharge stops, leading to a flow reversal that provokes an increase in shallow groundwater
salinity connected to the upward flux of hyper-saline groundwater contained in the fine
prodelta sediments. This upward flux slowly salinizes the surface waters bodies connected
with the aquifer. Simulations indicate that an increase in temperature or an increase in the
severe high rainfall events have only minor influence on the aquifer salinization rate, while an
interruption of canal usage could create serious upward movement of the high salinity
groundwater currently residing in the aquifer bottom. Besides, the impact of the projected sea
level rise seems much more important than temperature and rainfall increase, thus new
studies will need to better delineate the forthcoming sea level rise at local scale.
Since climactic changes may enhance the rate of autonomous salinization in the unconfined
aquifer of Ferrara, the contribution of water-sediment interaction and the anthropogenic
influence on groundwater quality were studied in order to assess whether the change in the
present chemical-physical and hydrogeological conditions may also induce the remobilization
of geogenic elements. The study focused the attention on nitrogen species (see chapter 4.4 )
and trace elements (see chapter 4.5).
The prevalent dissolved inorganic N species was NH4+, especially where reducing condition
dominates, while NO3- prevails where oxidizing condition prevails. Very high NH4+
concentrations were found in the salinized anoxic aquifer in coincidence with peaty
sediments, and even higher concentrations were in the low-lying aquitard. NH4+ extracted
from core samples and SED-TOC determinations suggested that NH4+ is of natural origin and
is likely due to mineralization of organic rich fine sediments, originated in paleo-marsh
environments.
89
SUMMARY AND CONCLUSIONS
Plots of NH4+ versus salinity and Sr2+ confirm the hypothesis of sediment being the source of
NH4+, with the deepest and most mineralized samples displaying the highest NH 4+
concentrations.
Accordingly, the total trace elements (TE) concentration within the aquifer matrix was higher
in the fine sediments, since usually these elements concentrate in the clay fraction, while
groundwater concentrations generally increased with the ionic strength of the solution. The
only exceptions were V and As that did not follow this trend, being most probably released by
fertilizers leaching in the top part of the aquifer and by buried peaty layers in the lower part
of the aquifer. In addition, the elevated SOM induced reducing conditions throughout the
aquifer profile, which further promoted TE dissolution. The presence of carbonates in the
aquifer matrix probably stimulated a high release of Pb and Zn in reducing alkaline-saline
conditions, suggesting Pb and Zn surface complexation on carbonate and dolomite minerals.
Some TE, like aluminium, were found to be affected by filtration, suggesting the presence of
aqueous complexes and formation of ultrafine colloids in coincidence with fine lenses. This
latter aspect could be relevant for TE transport and should be accurately studied in future
researches.
5.3 Recommendations
5.3.1 Practical implications
Discriminating between past and actual salinization is often difficult or even impossible,
especially when only salinity data are collected. It is evident that the current hydrogeological
environment and its present border conditions (natural and human) are completely different
from those, which allowed originating relic seawater and successively preserving it . The
interpretation of present-day observations must take into account that water quality patterns
continue evolving starting from the past, by adapting to contemporary boundary conditions.
The knowledge of the geological history of the area under investigation is then essential to
recognize the potential contribution of past hydrological processes to the present situation.
Moreover, collection of reliable samples resulted of vital importance for the correct
identification of aquifer physical properties and groundwater quality distribution. The
selection of appropriate monitoring techniques, such as depth-dependent sampling, are
therefore as important as that of right analytical and modelling techniques. MSL technique
does not require highly-specialized skills or expensive equipment, but still today studies on
aquifer salinization are rarely taking advantage of this approach that is conversely common in
contaminant fate and transport assessment. The implications of what said so far are not
limited to the local situation, but have a wider impact.
In fact, the rational demonstration, for a real site, of how severe can be the misleading in using
ease but incorrect conceptual models based on unreliable monitoring techniques, can
90
SUMMARY AND CONCLUSIONS
represent a warning to water resources managers who have to make difficult choices for
planning future groundwater protection and exploitation.
Finally, even if field data are indispensable to understand the hydrogeological system under
investigation, numerical models provide a means to test hypothesis based on physical and
chemical theory. With the right approach, the conjunctive use of both measurements and
simulation leads to better understanding and new insights. In this study numerical scenarios
quantified the possible climate change impacts on groundwater salinization and its
consequences on water resources quality: moreover, they could be helpful in outlining
mitigation strategies to decrease future salt load to groundwater and surface water systems.
As matter of fact, successful choice and implementation of mitigative countermeasures
requires a solid knowledge base, through which the relevant conditions and processes are
well characterized: however, planning also needs to take into account the economic aspects
that were not considered in this study. Nevertheless, the characterization protocol described
and applied to this study could be successfully applied in many other similar coastal low lying
areas worldwide, suffering from the same issues that are stressing the Ferrara coastal aquifer.
5.3.2 Remaining challenges
The present study used knowledge and methodologies from different earth-scientific
disciplines to investigate the salinization of the Ferrara coastal aquifers. None of the
methodologies provided complete understanding of the hydrogeological system under
consideration when applied separately. The strength of the adopted multi-methodological
approach provides an important progress of previous studies that mainly addressed the
subject from a single view in this area. Notwithstanding the profuse efforts, some of the topics
that were discussed in this thesis are still not fully understood.
For instance, the understanding of the hydrogeological behaviour of low-permeability
Holocene strata is inadequate. The observed spatial salinity variations in the pore-water of
these layers imply complex flow patterns, which contrast with the often-made assumption of
a uniform vertical flow field. This scale issue is relevant on ecological and pollution-oriented
applications that require detailed knowledge of the small-scale flow field, such as the
quantification of high nutrient loads to surface waters in groundwater discharge areas. It
requires the development of methods to characterize groundwater flow in greater detail than
those provided by sparse observation wells.
91
SUMMARY AND CONCLUSIONS
5.4 Related publications
Contributions concerning the characterization of the lowland coastal aquifer of Ferrara and its
salinization:
 Colombani N.*, Bonzi L., Calabrese L., Ferrari V., Giambastiani B.M.S., Mastrocicco M., Severi
P. (2012). Correlation between hydrochemical facies and depositional environments in a
salinized lowland coastal aquifer (Ferrara, IT). Proceedings of 7th EUREGEO, 12-15 giugno
2012, Bologna (IT). Eds. Scappini S., Zapparoli S. Centro Stampa Regione Emilia-Romagna,
pp. 363-364.
 Giambastiani B.M.S.*, Colombani N., Fidelibus M.D., Severi P., Mastrocicco M. (2012).
Groundwater hypersalinization in a lowland coastal aquifer (Po River Plain, Italy).
Proceedings of 22nd Salt Water Intrusion Meeting. 17-22 June 2012, Búzios (BRA). Eds.
Cardoso da Silva G.Jr., Gico de Lima Montenegro S.M. - Meta marketing e eventos Publisher;
ISBN: 978-85-63243-03-4, pp. 82-85.
 Giambastiani B.M.S., Colombani N., Mastrocicco M.* (2012). Artificial recharge assessment
in salinized coastal aquifers: Comacchio case study (IT). Proceedings of the 39th IAH
Congress. 16-21September 2012, Niagara Falls (CAN).
 Giambastiani B.M.S., Colombani N., Mastrocicco M.*, Fidelibus M.D. (2013). Characterization
of the lowland coastal aquifer of Comacchio (Ferrara, Italy): hydrology, hydrochemistry and
evolution of the system. JOURNAL OF HYDROLOGY, 501, pp. 35-44.
 Giambastiani B.M.S., Colombani N., Mastrocicco M.*, Severi P. (2012). Salinization and
freshening processes in the coastal aquifer of Ferrara (IT). Proceedings of BALWOIS2012: V
International Scientific Conference on Water, Climate and Environment; 28 May - 2 June
2012, Ohrid (FYROM). Eds Morell M., Popovska C., Morell O., ISBN 978-608-4510-10-9.
 Mastrocicco M., Colombani N.*, Antonellini M. (2012). Freshwater-seawater mixing
experiments in sand columns. JOURNAL OF HYDROLOGY, 448–449, pp. 112–118.
 Volta G., Colombani N.* (2013). Continuous monitoring of surface-groundwater
interactions in a lowland coastal aquifer. REND. ONLINE SOC. GEOL. IT., 25, 137-139.
Contributions concerning the importance of data acquisition techniques in saltwater intrusion
monitoring:
 Colombani N., Giambastiani B.M.S.*, Mastrocicco M. (2015). Combined use of electrical
conductivity and temperature to estimate aquifer properties in a forced gradient test.
Submitted to JOURNAL OF HYDROLOGY, HYDROL18711.
92
SUMMARY AND CONCLUSIONS
 Giambastiani B.M.S.*, Colombani N., Mastrocicco M. (2012). Multilevel characterization of
vertical hydraulic gradients, permeability, temperature and salinity in a shallow coastal
aquifer (Ferrara, Italy). Proceedings of 22nd Salt Water Intrusion Meeting. 17-22 June
2012, Búzios (BRA). Eds. Cardoso da Silva G.Jr., Gico de Lima Montenegro S.M. - Meta
marketing e eventos Publisher; ISBN: 978-85-63243-03-4, pp. 140-143.
 Giambastiani B.M.S., Colombani N., Mastrocicco M.*, Severi P. (2012). Cost and timeeffective methodology to evaluate aquifer parameters in coastal aquifers. Proceeding of
FLOWPATH 2012, 20-22 giugno 2012 Bologna (IT). Percorsi di Idrogeologia.
 Giambastiani B.M.S., Colombani N.*, Salemi E., Mastrocicco M. (2012). Metodica di
caratterizzazione multilivello di gradienti idraulici, permeabilità e conducibilità elettrica in
acquiferi costieri salinizzati. IV Congresso Nazionale AIGA - Perugia, 6-7 febbraio 2012.
ENGINEERING, HYDRO, ENVIRONMENTAL GEOLOGY, 14 (B), pp. 132-133.
 Mastrocicco M., Colombani N.*, Volta G. (2014). Misleading reconstruction of seawater
intrusion via integral depth sampling. Submitted to JOURNAL OF HYDROLOGY, HYDROL
18381.
 Mastrocicco M., Giambastiani B.M.S., Severi P., Colombani N.* (2012). The importance of
data acquisition techniques in saltwater intrusion monitoring. WATER RESOURCES
MANAGEMENT, 26 (10), pp. 2851-2866.
Contributions concerning the implementation of numerical modelling to assess climate
change impact on the salinization of the Ferrara coastal aquifer:
 Colombani N., Giambastiani B.M.S., Mastrocicco M.* (2014). Predicting salinization trends in
a lowland aquifer: Comacchio (Italy). WATER RESOURCES MANAGEMENT, 29(2), 603-618.
 Colombani N., Giambastiani B.M.S, Mastrocicco M.* (2013). Forecasting Salinization Trends
In A Lowland Aquifer: Comacchio (Italy). Proceedings of EWRA: 8th International
Conference, 26-29 June 2013, Porto (PT). In Water Resources Management In An
Interdisciplinary And Changing Context. Eds. Maia R., Guerreiro de Brito A., Seca Teixeira
A., Tentúgal Valente J., Pêgo J.P., FEUP Publisher. ISBN: 978-989-95557-8-5

Colombani N., Osti A., Volta G., Mastrocicco M.* (2015). Impact of climate change on
salinization of coastal water resources. Submitted to WATER RESOURCES MANAGEMENT.
 Mastrocicco M., Colombani N.*, Gargini A. (2014). Modelling present and future Po river
interactions with the alluvial aquifers (low Po river plain, Italy). JOURNAL OF WATER AND
CLIMATE CHANGE, 5(3), 457-471.
Contributions concerning the effect that salinization could have on the mobilization of major
and trace elements:
93
SUMMARY AND CONCLUSIONS
 Castaldelli G., Colombani N.*, Vincenzi F., Mastrocicco M. (2013). Linking dissolved organic
carbon, acetate and denitrification in agricultural soils. ENVIRONMENTAL EARTH
SCIENCES, 68 (4), pp. 939-945.
 Castaldelli G.*, Pierobon E., Soana E., Salemi E., Aschonitis V., Colombani N., Mastrocicco M.,
Bartoli M. (2012). Nitrogen mass balance in a coastal lowland declared vulnerable to
nitrate (WFD 2000/60/EC): the relevance of secondary canals in excess nitrogen removal.
Proceedings of the 17th Nitrogen Workshop, 26-29th June 2012, Wexford (IRE).
Innovations for sustainable use of nitrogen resources, Eds. Richards K.G., Fenton O.,
Watson C.J. ISBN-10: 1-84170-588-8, pp. 172-173.
 Colombani N., Mastrocicco M.*, Dinelli E. (2014). Trace elements mobility in a saline coastal
aquifer of the Po river lowland (IT). Submitted to JOURNAL OF GEOCHEMICAL
EXPLORATION, GEXPLO3735.
 Colombani N.*, Mastrocicco M., Vincenzi F., Castaldelli G. (2012). Nitrogen speciation and
denitrification in a low land territory receiving synthetic fertilizers (Province of Ferrara,
Northern Italy). Proceedings of BALWOIS2012: V International Scientific Conference on
Water, Climate and Environment; 28 May - 2 June 2012, Ohrid (FYROM). Eds Morell M.,
Popovska C., Morell O., ISBN 978-608-4510-10-9.
 Giambastiani B.M.S., Colombani N., Mastrocicco M.* (2014). Detecting small scale variability
of trace elements in a shallow aquifer. WATER AIR AND SOIL POLLUTION, doi:
10.1007/s11270-014-2283-4
 Giambastiani B.M.S, Colombani N.*, Mastrocicco M. (2013). High phosphate bearing
sediments in a salinized coastal aquifer. Proceedings of GQ13: 8th IAHS Groundwater
Quality Conference, 21-26 aprile 2013, Gainesville, Florida (USA).
 Mastrocicco M., Colombani N. *, Castaldelli G. (2013). A step-wise approach to assess the
fate of nitrogen species in agricultural lowlands. In "Wastewater reuse and management".
Edito da Sharma S.K. and Sanghi R. (Eds.) Springer-Verlag New York, LLC; ISBN 978-94007-4941-2, 474 p.
 Mastrocicco M., Giambastiani B.M.S., Colombani N.* (2013). Ammonium occurrence in a
salinized lowland coastal aquifer. HYDROLOGICAL PROCESSESS, 27 (24), pp. 3495-3501.
94
REFERENCES
REFERENCES
Acosta J.A., Jansen B., Kalbitz K., Faz A., Martínez-Martínez S., 2011. Salinity increases mobility
of heavy metals in soils. Chemosphere 85(8), 1318-1324.
Alcalà F.J. and Custodio E., 2008. Using the Cl/Br ratio as a tracer to identify the origin of
salinity in aquifers in Spain and Portugal. Journal of Hydrology 359, 189-207.
Allen R.G., Pereira L.S., Raes D., Smith M., 1998. Crop evapotranspiration. Guidelines for
computing crop water requirements. Irrigation and Drainage Paper No. 56. FAO, Rome.
Amorosi A., Centineo M.C., Colalongo M.L., Pasini G., Sarti G., Vaiani S.C., 2003. Facies
Architecture and Latest Pleistocene–Holocene Depositional History of the Po Delta
(Comacchio Area), Italy. The Journal of Geology 111, 39-56.
Amorosi A., Centineo M.C., Dinelli E., Lucchini F., Tateo F., 2002. Geochemical and
mineralogical variations as indicators of provenance changes in Late Quaternary deposits of
SE Po Plain. Sedimentary Geology 151, 273-292.
Amorosi A., Colalongo M. L., Fiorini F., Fusco F., Pasini G., Vaiani S.C., Sarti G., 2004.
Palaeogeographic and palaeoclimatic evolution of the Po Plain from 150-ky core records.
Global and Planetary Change 40(1-2), 55-78.
Amorosi A. and Sammartino I., 2006. Influence of sediment provenance on background values
of potentially toxic metals from near-surface sediments of Po coastal plain (Italy).
International Journal of Earth Sciences 96(2), 389-396.
Andersen M.S., Nyvang V., Jakobsen R., Postma D., 2005. Geochemical processes and solute
transport at the seawater/freshwater interface of a sandy aquifer. Geochimica and
Cosmochimica Acta 69, 3979-3994.
Antonellini M. and Mollema P.N., 2010. Impact of groundwater salinity on vegetation species
richness in the coastal pine forests and wetlands of Ravenna, Italy. Ecological Engineering
36(9), 1201-1211.
Antonellini M., Mollema P.N., Giambastiani B.M.S., Bishop K., Caruso L., Minchio A., Pellegrini
L., Sabia M., Ulazzi E., Gabbianelli G., 2008. Salt water intrusion in the coastal aquifer of the
southern Po Plain, Italy. Hydrogeology Journal 16(8), 1541-1556.
Appelo C.A.J. and Postma D., 2005. Geochemistry, groundwater and pollution, 2nd Edition,
Balkema, Rotterdam, The Netherlands.
95
REFERENCES
Aravena R. and Robertson W.D., 1998. Use of multiple isotope tracers to evaluate
denitrification in ground water: Study of nitrate from a large-flux septic system plume.
Ground Water 36(6), 975-982.
Artegiani A., Paschini E., Russo A., Bregant D., Raicich F., Pinardi N., 1997. The Adriatic Sea
general circulation. Part I: Air−sea interactions and water mass structure. Journal of Physical
Oceanography 27, 1492-1514.
Balugani E. and Antonellini M., 2011. Barometric pressure influence on water table
fluctuations in coastal aquifers of partially enclosed seas: An example from the Adriatic coast,
Italy. Journal of Hydrology 400, 176-186.
Balugani E. and Antonellini M., 2010. Measuring Salinity within Shallow Piezometers:
Comparison of Two Field Methods. Journal Water Resource and Protection 2, 251-258.
Barlow P.M. and Reichard E.G., 2010. Saltwater intrusion in coastal regions of North America.
Hydrogeology Journal 18, 247-260.
Bear J., 1972. Dinamics of fluids in porous media. Elsevier, Amsterdam.
Bear J., Cheng A.H.-D., Sorek S., Quazar D., Herrea I., 1999. Seawater intrusion in coastal
aquifers -concepts, methods and practices. Kluwer Academic Publishers.
Böhlke J.K., Wanty R., Tuttle M., Delin G., Landon M., 2002. Denitrification in the recharge area
and discharge area of a transient agricultural nitrate plume in a glacial outwash sand aquifer,
Minnesota. Water Resources Research 38(7), 101-1026.
Bower C.F. and Holm-Hansen T., 1980. A salicylate-hypoclorite method for determining
ammonia in seawater. Canadian Journal of Aquatic Sciences 37, 794-798.
Brambati A., Candian C., Bisiacchi G., 1973. Fortran IV program for settling tube size analysis
using CDC 6200 computer. Istituto di Geologia e Paleontologia, Università di Trieste.
Bratton J.F., Bohlke J.K., Krantz D.E., Tobias C.R., 2009. Flow and geochemistry of groundwater
beneath a back-barrier lagoon: The subterranean estuary at Chincoteague Bay, Maryland,
USA. Marine Chemistry 113(1-2), 78-92.
Breit G.N. and Wanty R.B., 1991. Vanadium accumulation in carbonaceous rocks: a review of
geochemical controls during deposition and diagenesis. Chemical Geology 91, 83-97.
Britt S.L., 2005. Testing the in-well horizontal laminar flow assumption with a sand-tank well
model. Groundwater Monitoring and Remediation 25(3), 73-81.
96
REFERENCES
Brun A., Engesgaard P., Christensen T.H., Rosbjerg D., 2002. Modelling of transport and
biogeochemical processes in pollution plumes: Vejen landfill, Denmark. Journal of Hydrology
256(3-4), 228-247.
Buggy C.J. and Tobin J.M., 2008. Seasonal and spatial distribution of metals in surface
sediment of an urban estuary.Environmental Pollution 155(2), 308-319.
Bureau of Reclamation, 2001. Water testing for permeability. In: Engineering Geology Field
Manual, US Dept of the Interior, Washington. Vol. 2, Chap. 17.
Candela L., von Igel W., Elorza F.J., Aronica G., 2009. Impact assessment of combined climate
and management scenarios on groundwater resources and associated wetland (Majorca,
Spain). Journal of Hydrology 376(3-4), 510-527.
Capaccioni B., Didero M., Paletta C., Didero L., 2005. Saline intrusion and refreshening in a
multilayer coastal aquifer in the Catania Plain (Sicily, Southern Italy): dynamics of
degradation processes according to the hydrochemical characteristics of groundwater.
Journal of Hydrology 307, 1-16.
Carminati E., Dogliosi C., Scrocca D., 2005. Magnitude and causes of natural subsidence of
Venice. In: Fletcher C, Spencer T (Eds.) Flooding and environmental challenges for Venice and
its lagoon. Cambridge University Press, Cambridge, pp. 21-28.
Cartwright I., Weaver T.R., Fulton S., Nichol C., Reid M., Cheng X., 2004. Hydrogeochemical and
isotopic constrains on the origins of dryland salinity, Murray basin, Victoria, Australia. Applied
Geochemistry 19, 1233-1254.
Chang S.W., Clement T.P., Simpson M.J., Lee K.K., 2011. Does sea-level rise have an impact on
saltwater intrusion? Advances in Water Resources 34(10), 1283-1291.
Christensen T.H., Kjeldsen P., Bjerg P.L., Jensen D.L., Christensen J.B., Baun A., Albrechtsen H.J.,
Heron G., 2001. Biogeochemistry of landfill leachate plumes. Applied Geochemistry 16(7-8),
659-718.
Cidu S., Vittori Antisari L., Biddau R., Buscaroli A., Carbone S., Carbone S., Da Pelo S., Dinelli E.,
Vianello G., Zannoni D., 2013. Dynamics of rare earth elements in water–soil systems: The case
study of the Pineta San Vitale (Ravenna, Italy). Geoderma 193-194, 52-67.
Colombani N., Mastrocicco M., Dinelli E., 2014. Trace elements mobility in a saline coastal
aquifer of the Po river lowland (IT). Submitted to Journal of Geochemical Exploration,
GEXPLO3735.
97
REFERENCES
Colombani N., Mastrocicco M., Gargini A., Davis G.B., Prommer H., 2009. Modelling the fate of
styrene in a mixed petroleum hydrocarbon plume. Journal of Contaminant Hydrology 105(12), 38-55.
Colombani N., Mastrocicco M., Giambastiani B.M.S., 2015a. Predicting salinization trends in a
lowland coastal aquifer: Comacchio (Italy). Water Resources Management doi:
10.1007/s11269-014-0795-8.
Colombani N., Osti A., Volta G., Mastrocicco M., 2015b. Impact of climate change on salinization
of coastal water resources. Submitted to Water Resources Management.
Colombani N., Volta G., Osti A., Mastrocicco M., 2015c. Misleading reconstruction of seawater
intrusion via integral depth sampling. Submitted to Journal of Hydrology, HYDROL18381.
Curzi P.V., Dinelli E., Ricci Lucchi M., Vaiani S.C., 2006. Palaeoenvironmental control on
sediment composition and provenance in the late Quaternary deltaic successions: a case study
from the Po delta area (Northern Italy). Geological Journal 41, 591-612.
Custodio E., 2010. Coastal aquifers in Europe: an overview. Hydrogeology Journal 18, 269280.
Danielson R.E. and Sutherland P.L., 1986. In: Klute A. (ed.) Methods of Soil Analysis, Part I.
Physical and Mineralogical Methods, 2nd Edition. Agronomy Monograph 9, 443-461.
Davis S.N., Whittemone D.O., Fabryka-Martin J., 1998. Uses of chloride/bromide ratios in
studies of potable water. Ground Water 36(2), 338-350.
Davis S.N. and De Wiest R.J.M., 1996. Hydrogeology. John Wiley & Sons, New York, pp. 463.
Delaune R.D., Gambrell R.P., Jugsujinda A., Devai I., Hou A., 2008. Total Hg, methyl Hg and
other toxic heavy metals in a northern Gulf of Mexico estuary: Louisiana Pontchartrain basin.
Journal of Environmental Science and Health - Part A Toxic/Hazardous Substances and
Environmental Engineering 43(9), 1006-1015.
Dellwig O., Böttcher M.E., Lipinski M., Brumsack H.-J., 2002. Trace metals in Holocene coastal
peats and their relation to pyrite formation (NW Germany). Chemical Geology 182(2-4), 423442.
Dellwig O., Watermann F., Brumsack H.-J., Gerdes G., Krumbein W.E., 2001. Sulphur and iron
geochemistry of Holocene coastal peats (NW Germany): A tool for palaeoenvironmental
reconstruction. Palaeogeography, Palaeoclimatology, Palaeoecology 167 (3-4), 359-379.
98
REFERENCES
de Louw P.G.B., Eeman S., Oude Essink G.H.P., Vermue E., Post V.E.A., 2013a. Rainwater lens
dynamics and mixing between infiltrating rainwater and upward saline groundwater seepage
beneath a tile-drained agricultural field. Journal of Hydrology 501, 133-145.
de Louw P.G.B., Eeman S., Siemon B., Voortman B.R., Gunnink J., Van Baaren S.E., Oude Essink
G.H.P., 2011. Shallow rainwater lenses in deltaic areas with saline seepage. Hydrology and
Earth System Sciences 15, 3659-3678.
de Louw P.G.B., Oude Essink G.H.P., Stuyfzand P.J., van der Zee S.E.A.T.M., 2010. Upward
groundwater flow in boils as the dominant mechanism of salinization in deep polders. Journal
of Hydrology 394 (3-4), 494-506.
de Louw P.G.B., Vandenbohede A., Werner A.D., Oude Essink G.H.P., 2013b. Natural saltwater
upconing by preferential groundwater discharge through boils. Journal of Hydrology 490, 7487.
de Montety V., Radakovitch O., Vallet-Coulomb C., Blavoux B., Hermitte D., Valles V., 2008.
Origin of groundwater salinity and hydrogeochemical processes in a confined coastal aquifer:
case of the Rhône delta (Southern France). Applied Geochemistry 23 (8), 2337-2349.
Devai I., Patrick Jr. W.H., Neue H.-U., DeLaune R.D., Kongchum M., Rinklebe J., 2005. Methyl
mercury and heavy metal content in soils of rivers Saale and Elbe (Germany). Analytical
Letters 38(6), 1037-1048.
Dhar R.K., Zheng Y., Stute M., van Geen A., Cheng Z., Shanewaz M., Shamsudduha M., Hoque
M.A., Rahman M.W., Ahmed K.M., 2008. Temporal variability of groundwater chemistry in
shallow and deep aquifers of Araihazar, Bangladesh. Journal of Contaminant Hydrology 99(14), 97-111.
Dillon P., 2005. Future management of aquifer recharge. Hydrogeology Journal 13(1), 313316.
Dinelli E., Tateo F., Summa V., 2007. Geochemical and mineralogical proxies for grain size in
mudstones and siltstones from the Pleistocene and Holocene of the Po River alluvial plain,
Italy. In J. Arribas, S. Critelli, and M.J. Johnsson (Eds) Sedimentary Provenance and
Petrogenesis: Perspectives from Petrography and Geochemistry. Geological Society of
America Special Paper, 420, 25-36.
Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000.
Establishing a framework for Community action in the field of water policy. Official Journal of
the European Communities L327/1.
99
REFERENCES
Doherty J., 2002. PEST: Model-Independent Parameter Estimation: User’s Manual, 5th ed.,
Watermark Numer. Comput., Brisbane, Australia.
Dou Y., Li J., Zhao J., Hu B., Yang S., 2013. Distribution, enrichment and source of heavy metals
in surface sediments of the eastern Beibu Bay, South China Sea. Marine Pollution Bulletin
67(1-2), 137-45.
Du Laing G., De Vos R., Vandecasteele B., Lesage E., Tack F.M.G., Verloo M.G., 2008. Effect of
salinity on heavy metal mobility and availability in intertidal sediments of the Scheldt estuary.
Estuarine, Coastal and Shelf Science 77(4), 589-602.
Du Laing G., Rinklebe J., Vandecasteele B., Meers E., Tack F.M.G., 2009. Trace metal behaviour
in estuarine and riverine floodplain soils and sediments: A review. Science of the Total
Environment 407(13), 3972-3985.
Edmunds W.M., 2001. Palaeowaters in European coastal aquifers – the goals and main
conclusions of the PALAEAUX project. In: Edmunds, W.M., Milne, C. (Eds.), Paleowaters in
Coastal Europe: Evolution of Groundwater since the Late Pleistocene. Special Publications
189. Geological Society, London, pp. 1-16.
Einarson M.D. and Cherry J.A., 2002. A new multilevel ground water monitoring system using
multichannel tubing. Ground Water Monitoring and Remediation 22 (4), 52-65.
Elci A., Gregory P.F., Molz F.J., 2003. Detrimental effects of natural head gradients on chemical
and water level measurements in observation wells: Identification and control. Journal of
Hydrology 281, 70-81.
Emerson S.R. and Huested S.S., 1991. Ocean anoxia and the concentrations of molybdenum
and vanadium in seawater. Marine Chemistry 34, 177-196.
Fakir Y., Mernissi M.E., Kreuser T., Berjami B., 2002. Natural tracer approach to characterize
groundwater in the coastal Sahel of Oualidia (Morocco). Environmental Geology 43, 197-202.
Fass T., Cook P.G., Stieglitz T., Herczeg A.L., 2007. Development of saline groundwater through
transpiration of seawater. Ground Water 45 (6), 703-710.
Feddes R.A., Kowalik P.J., Zaradny H., 1978. Simulation of field water use and crop yield.
Pudoc, Wageningen, the Netherlands.
Feseker T., 2007. Numerical studies on saltwater intrusion in a coastal aquifer in
northwestern Germany. Hydrogeology Journal 15(2), 267-279.
100
REFERENCES
Fetter C.W., 2001. Applied hydrogeology, 4th ed. Prentice Hall, Englewood Cliffs, USA, pp. 598.
Fidelibus M.D., Calò G., Tinelli R., Tulipano L., 2011. Salt ground waters in the Salento karstic
coastal aquifer (Apulia, Southern Italy). In Lambrakis N., Stournaras G & Katsanou K. (Eds.),
Advances in the Research of Aquatic Environment, Environmental Earth Sciences Series,
Springer-Verlag, Berlin. 1, 407-415.
Fujinawa K., Iba T., Fujihara Y., Watanabe T., 2009. Modeling Interaction of Fluid and Salt in an
Aquifer/Lagoon System. Ground Water 47(1), 35-48.
García-Ruiz J.M., López-Moreno J.I., Vicente-Serrano S.M., Lasanta T., Beguería S., 2011.
Mediterranean water resources in a global change scenario. Earth Science Reviews 105, 121139.
Gattacceca J.C., Vallet-Coulomb C., Mayer A., Claude C., Radakovitch O., Conchetto E., Hamelin
B., 2009. Isotopic and geochemical characterization of salinization in the shallow aquifers of a
reclaimed subsiding zone: the southern Venice Lagoon coastland. Journal of Hydrology 378,
46-61.
Gelberg K.H., Church L., Casey G., London M., Roerig D.S., Boyd J., Hill M., 1999. Nitrate levels in
drinking water in rural New York State. Environmental Research 80(1), 34-40.
Ghiglieri G., Oggiano G., Fidelibus M.D., Alemayehu T., Barbieri G., Vernier A., 2009.
Hydrogeology of the Nurra Region, Sardinia (Italy): basement-cover influences on
groundwater occurrence and hydro-geochemistry. Hydrogeology Journal 17(2), 447-466.
Giambastiani B.M.S., Antonellini M., Oude Essink G.H.P., Stuurman R.J., 2007. Saltwater
intrusion in the unconfined coastal aquifer of Ravenna (Italy): a numerical model. Journal of
Hydrology 340, 91-104.
Giambastiani B.M.S., Colombani N., Mastrocicco M., Fidelibus M.D., 2013. Characterization of
the lowland coastal aquifer of Comacchio (Ferrara, Italy): hydrology, hydrochemistry and
evolution of the system. Journal of Hydrology 501, 35-44.
Giménez Forcada E. and Morell Evangelista I., 2008. Contributions of boron isotopes to
understanding the hydrogeochemistry of the coastal detritic aquifer of Castellon Plain, Spain.
Hydrogeology Journal 16, 547-557.
Giorgi F. and Lionello P., 2008. Climate change projections for the Mediterranean region.
Global Planetary Change 63, 90-104.
101
REFERENCES
Glessner J.J.G. and Roy W.R., 2009. Paleosols in central Illinois as potential sources of
ammonium in groundwater. Ground Water Monitoring and Remediation 29(4), 56-64.
Grassi S., Cortecci G., Squarci P., 2007. Groundwater resource degradation in coastal plains:
the example of the Cecina area (Tuscany – Central Italy). Applied Geochemistry 22, 22732289.
Gualdi S., Somot S., Li L., Artale V., Adani M., Bellucci A., Braun A., Calmanti S., Carillo A. et al.,
2013. The CIRCE simulations – Regional climate change projections with realistic
representation of the Mediterranean Sea. Bulletin of the American Meteorological Society 94,
65-81.
Gulis G., Czompolyova M., Cerhan J.R., 2002. An ecologic study of nitrate in municipal drinking
water and cancer incidence in Trnava District, Slovakia. Environmental Research 88(3), 182187.
Guo W. and Langevin C.D., 2002. User's guide to SEAWAT: A computer program for simulation
of three-dimensional variable-density ground-water flow. Techniques of Water-Resources
Investigations 6-A7, pp. 77.
Hanson R.T., Flint L.E., Flint A.L., Dettinger D.M., Faunt C.C., Cayan D., Schmid W., 2012. A
method for physically based model analysis of conjunctive use in response to potential
climate changes. Water Resources Research 48(2), W00L08.
Harbaugh A.W., 2005. MODFLOW-2005, the U.S. Geological Survey Modular Ground-water
Model - The Ground-water Flow Process: U.S. Geological Survey Techniques and Methods 6,
A16.
Hartog N., Griffioen J., Van der Weijden C.H., 2002. Distribution and reactivity of O2-reducing
components in sediments from a layered aquifer. Environmental Science and Technology
36(11), 2338-2344.
Henderson T.H., Mayer K.U., Parker B.L., Al T.A., 2009. Three-dimensional density-dependent
flow and multicomponent reactive transport modeling of chlorinated solvent oxidation by
potassium permanganate. Journal of Contaminant Hydrology 106, 195-211.
Hill M.C., Banta E.R., Harbaugh A.W., Anderman E.R., 2000. MODFLOW-2000, the U.S.
Geological Survey Modular Ground-Water Model - User guide to the observation, sensitivity,
and parameter-estimation processes and three post-processing programs. U.S. Geological
Survey Open-File Report 00-184, pp. 210.
102
REFERENCES
Hill H.J., Shirley O.J., Klein G.E., 1979. Bound water in shaly sands—its relation to Qv and other
formation properties. The Log Analyst, Society of Professional Well Log Analysts XX(3), 3-19.
Hinkle S.R., Böhlke J.K., Duff J.H., Morgan D.S., Weick R.J., 2007. Aquifer-scale controls on the
distribution of nitrate and ammonium in ground water near La Pine, Oregon, USA. Journal of
Hydrology 333(2-4), 486-503.
Hinsby K., Condesso de Melo M.T., Dahl M., 2008. European case studies supporting the
derivation of natural background levels and groundwater threshold values for the protection
of dependent ecosystems and human health. The Science of the Total Environment 401(1-3),
1-20.
Hoque M.A., McArthur J.M., Sikdar P.K., Ball J.D., Molla T.N., 2014. Tracing recharge to aquifers
beneath an Asian megacity with Cl/Br and stable isotopes: the example of Dhaka, Bangladesh.
Hydrogeology Journal 22(7), 1549-1560.
IPCC, 2014a. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and
Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change [Field CB, VR Barros, DJ Dokken, KJ Mach, MD
Mastrandrea, TE Bilir, M Chatterjee, KL Ebi, YO Estrada, RC Genova, B Girma, ES Kissel, AN
Levy, S MacCracken, PR Mastrandrea, LL White (eds.)]. Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA.
IPCC, 2014b. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part B: Regional
Aspects. Contribution of Working Group II to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change [Barros, VR, CB Field, DJ Dokken, MD
Mastrandrea, KJ Mach, TE Bilir, M Chatterjee, KL Ebi, YO Estrada, RC Genova, B Girma, ES.
Kissel, AN Levy, S MacCracken, PR Mastrandrea, LL White (eds.)]. Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA.
Jang C.S. and Liu C.W., 2005. Contamination potential of nitrogen compounds in the
heterogeneous aquifers of the Choushui River alluvial fan, Taiwan. Journal of Contaminant
Hydrology 79(3-4), 135-155.
Jayasingha P., Pitawala A., Dharmagunawardhane H.A., 2011. Vulnerability of coastal aquifers
due to nutrient pollution from agriculture: Kalpitiya, Sri Lanka. Water Air and Soil Pollution
219, 563-577.
Jiao J.J., Wang Y., Cherry J.A., Wang X., Zhi B., Du H., Wen D., 2010. Abnormally high ammonium
of natural origin in a coastal aquifer-aquitard system in the Pearl River Delta, China.
Environmental Science and Technology 44, 7470-7475.
103
REFERENCES
Jorstad L.B., Jankowski J., Acworth R.I., 2004. Analysis of the distribution of inorganic
constituents in a landfill leachate-contaminated aquifer: Astrolabe Park, Sydney, Australia.
Environmental Geology 46(2), 263-272.
Kass A., Gavrieli I., Yechieli Y., Vangosh A., Starinsky A., 2005. The impact of freshwater and
wastewater irrigation on the chemistry of shallow groundwater: a case study from the Israeli
Coastal Aquifer. Journal of Hydrology 300, 314-331.
Kim K.-Y., Chon C.-M., Park K.-H., Park Y.-S., Woo N.C., 2008. Multi-depth monitoring of
electrical conductivity and temperature of groundwater at a multilayered coastal aquifer: Jeju
Island, Korea. Hydrological Processes 22(18), 3724-3733.
Konikow L.F. and Hornberger G.Z., 2006. Modeling effects of multimode wells on solute
transport. Ground Water 44 (5), 648-660.
Kurtzman D., Netzer L., Weisbrod N., Graber E., Ronen D., 2011. Steady-state homogeneous
approximations of vertical velocity from EC profiles. Ground Water 49(2), 275-279.
Lambeck K., Antonioli F., Anzidei M., Ferranti L., Leoni G., Scicchitano G., Silenzi S., 2011. Sea
level change along the Italian coast during the Holocene and projections for the future.
Quaternary International 232, 250-257.
Landon M.K., Green C.T., Belitz K., Singleton M.J., Esser B.K., 2011. Relations of hydrogeologic
factors, groundwater reduction-oxidation conditions, and temporal and spatial distributions
of nitrate, Central-Eastside San Joaquin Valley, California, USA. Hydrogeology Journal 19,
1203-1224.
Langevin C.D., Thorne D.T.Jr., Dausman A.M., Sukop M.C., Guo W., 2007. SEAWAT Version 4: A
Computer Program for Simulation of Multi-Species Solute and Heat Transport: U.S. Geological
Survey Techniques and Methods, 6-A22, pp 1-39.
Langevin C.D. and Zygnerski M., 2013. Effect of Sea-Level Rise on Salt Water Intrusion near a
Coastal Well Field in Southeastern Florida. Ground Water 51(5), 781-803.
Lebbe L., Van Meir N., Viaene P., 2008. Potential Implications of Sea-Level Rise for Belgium.
Journal of Coastal Research 24(2), 358-366.
Lenahan M.J. and Bristow K., 2010. Understanding sub-surface solute distributions and
salinization mechanisms in a tropical coastal floodplain groundwater system. Journal of
Hydrology 390, 131-142.
104
REFERENCES
Leven C. and Dietrich P., 2006. What information can we get from pumping tests?-comparing
pumping test configurations using sensitivity coefficients. Journal of Hydrology 319, 199-215.
Lewandowski J. and Nützmann G., 2010. Nutrient retention and release in a floodplain's
aquifer and in the hyporheic zone of a lowland river. Ecological Engineering 36(9), 11561166.
Li Q., Cai S., Mo C., Chu B., Peng L., Yang F., 2010. Toxic effects of heavy metals and their
accumulation in vegetables grown in a saline soil. Ecotoxicology and Environmental Safety
73(1), 84-88.
Li J., Li F., Liu Q., Zhang Y., 2014. Trace metal in surface water and groundwater and its
transfer in a Yellow River alluvial fan: evidence from isotopes and hydrochemistry. The
Science of the Total Environment 472, 979-988.
Linderfelt W.R. and Turner J.V., 2001. Interaction between shallow groundwater, saline
surface water and nutrient discharge in a seasonal estuary: The Swan - Canning system.
Hydrological Processes 15(13), 2631-2653.
Lorenzen G., Sprenger C., Baudron P., Gupta D., Pekdeger A., 2012. Origin and dynamics of
groundwater salinity in the alluvial plains of western Delhi and adjacent territories of
Haryana State, India. Hydrological Processes 26, 2333-2345.
Loring D.H., 1991. Normalization of heavy-metal data from estuarine and coastal sediments.
ICES Journal of Marine Science 48(1), 101-115.
Lorite-Herrera M., Hiscock K., Jiménez-Espinosa R., 2009. Distribution of dissolved inorganic
and organic nitrogen in river water and groundwater in an agriculturally-dominated
catchment, South-East Spain. Water Air and Soil Pollution 198, 335-346.
Martinelli L., Zanuttigh B., Corbau C., 2010. Assessment of coastal flooding hazard along the
Emilia Romagna littoral, IT. Coastal Engineering 57, 1042-1058.
Mastrocicco M., Colombani N., Castaldelli G., Jovanovic N., 2011a. Monitoring and modelling
nitrate persistence in a shallow aquifer. Water Air and Soil Pollution 217(1-4), 83-93.
Mastrocicco M., Colombani N., Di Giuseppe D., Faccini B., Coltorti M., 2013a. Contribution of
the subsurface drainage system in changing the nitrogen speciation of an agricultural soil
located in a complex marsh environment (Ferrara, Italy). Agricultural Water Management
119, 144-153.
105
REFERENCES
Mastrocicco M., Colombani N., Petitta M., 2011b. Modelling the density contrast effect on a
chlorinated hydrocarbon plume reaching the shore line. Water Air and Soil Pollution 220(14), 387-398.
Mastrocicco M., Colombani N., Salemi E., Castaldelli G., 2011c. Groundwater nitrogen
speciation in intensively cultivated lowland areas. In Advances in the Research of Aquatic
Environment, Springer, XXVIII, pp. 512, ISBN 978-3-642-24075-1. Eds. Lambrakis N.,
Stournaras G., Katsanou K. 2, 291-298.
Mastrocicco M., Colombani N., Salemi E., Castaldelli G., 2010. Numerical assessment of
effective evapotranspiration from maize plots to estimate groundwater recharge in lowlands.
Agricultural Water Management 97(9), 1389-1398.
Mastrocicco M., Colombani N., Salemi E., Vincenzi F., Castaldelli G., 2012a. The role of the
unsaturated zone in determining nitrate leaching to groundwater. In “Groundwater Quality
Sustainability”, IAH Book Series (IAH Selected Papers), Eds. Zuber A., Maloszewski P., Witczak
S., Malina G. - Taylor & Francis Books (UK) ISBN-13:9780415698412.
Mastrocicco M., Colombani N., Sbarbati C., Petitta M., 2012b. Assessing the effect of saltwater
intrusion on petroleum hydrocarbons plumes via numerical modelling. Water Air and Soil
Pollution 223(7), 4417-4427.
Mastrocicco M., Giambastiani B.M.S., Colombani N., 2013b. Ammonium occurrence in a
salinized lowland coastal aquifer. Hydrological Processes 27(24), 3495-3501.
Mastrocicco M., Giambastiani B.M.S., Severi P., Colombani N., 2012c. The importance of data
acquisition techniques in saltwater intrusion monitoring. Water Resources Management 26,
2851-2866.
Mc Donald J.M. and Harbaugh A.W., 1988. MODFLOW, a modular 3D finite difference groundwater flow model. U.S. Geological Survey Open File Report 83-875, pp. 586.
McMahon P.B. and Chapelle F.H., 2008. Redox processes and water quality of selected
principal aquifer systems. Ground Water 46(2), 259-71.
Mollema P.N., Antonellini M., Dinelli E., Gabbianelli G., Greggio N., Stuyfzand P.J., 2013.
Hydrochemical and physical processes influencing salinization and freshening in
Mediterranean low-lying coastal environments. Applied Geochemistry 34, 207-221.
Mollema P.N., Antonellini M., Gabbianelli G., Laghi M., Marconi V., Minchio A., 2012. Climate
and water budget change of a Mediterranean coastal watershed, Ravenna, Italy.
Environmental Earth Sciences 65, 257-276.
106
REFERENCES
Mondal N.C., Singh V.P., Singh S., Singh V.S., 2011. Hydrochemical characteristic of coastal
aquifer from Tuticorin, Tamil Nadu, India. Environmental Monitoring and Assessment 175,
531-550.
Nash J.E. and Sutcliffe J.V., 1970. River flow forecasting through conceptual models. Part I: A
discussion of principles. Journal of Hydrology, 10(3), 282–290.
Nativ R. and Weisbrod N., 1994. Management of a multilayered coastal aquifer – an Israeli
case study. Water Resources Management 8, 297-311.
Netzer L., Weisbrod N., Kurtzman D., Nasser A., Graber E.R., Ronen D., 2011. Observations on
Vertical Variability in Groundwater Quality: Implications for Aquifer Management. Water
Resources Management 25, 1315-1324.
Ortega-Guerrero A., 2003. A. Origin and geochemical evolution of groundwater in a closedbasin clayey aquitard, Northern Mexico. Journal of Hydrology 284(1-4), 26-44.
Otero N., Vitòria L., Soler A., Canals A., 2005. Fertilizer characterisation: Major, trace and rare
earth elements. Applied Geochemistry 20, 1473-1488.
Oude Essink G.H.P., van Baaren E.S., de Louw P.G.B., 2010. Effects of climate change on coastal
groundwater systems: A modeling study in the Netherlands. Water Resources Research
46(10), W00F04.
Park S.C., Yun S.T., Chae G.T., Yoo I.S., Shin K.S., Heo C.H., Lee S.K., 2005. Regional
hydrochemical study on salinization of coastal aquifers, western coastal area of South Korea.
Journal of Hydrology 313(3-4), 182-194.
Parkhurst D.L., 1995. User’s guide to PHREEQC – A computer program for speciation,
reaction-path, advective-transport, and inverse geochemical calculations: U.S. Geological
Survey Water - Resources Investigations Report 95-4227, pp. 143.
Post V. and Abarca E., 2010. Preface: saltwater and freshwater interactions in coastal aquifers.
Hydrogeology Journal 18, 1-4.
Post V., Kooi H., Simmons C., 2007. Using hydraulic head measurements in variable-density
ground water flow analyses. Ground Water 45(6), 664-671.
Prommer H., Tuxen N., Bjerg P.L., 2006. Fringe-controlled natural attenuation of phenoxy
acids in a landfill plume: Integration of field-scale processes by reactive transport modelling.
Environ Science Technology 40(15), 4732-4738.
107
REFERENCES
Puckett L.J., Cowdery T.K., McMahon P.B., Tornes L.H., Stoner J.D., 2002. Using chemical,
hydrologic, and age dating analysis to delineate redox processes and flow paths in the riparian
zone of a glacial outwash aquifer-stream system. Water Resources Research 38(8), 91-920.
Ranjana P., Kazamaa S., Sawamoto M., 2006. Effects of climate change on coastal fresh
groundwater resources. Global Environmental Change 16(4), 388-399.
Ridd P.V. and Stieglitz T., 2002. Dry season salinity changes in arid estuaries fringed by
mangroves and saltflats. Estuarine Coastal and Shelf Science 54(6), 1039-1049.
Ritchie J.T., 1972. Model for predicting evaporation from a row crop with incomplete cover.
Water Resources Research 8(5), 1204-1213.
Robbins G.A. and Martin-Hayden J.M., 1991. Mass balance evaluation of monitoring well
purging: Part I. Theoretical models and implications for representative sampling. Journal of
Contaminant Hydrology 8(3-4), 203-224.
Rodvang S.J. and Simpkins W.W., 2001. Agricultural contaminants in Quaternary aquitards: A
review of occurrence and fate in North America. Hydrogeology Journal 9(1), 44-59.
Sanchez-Martos F., Pulido-Bosch A., Molina-Sanchez L., Vallejos-Izquierdo A., 2002.
Identification of the origin of salinization in groundwater using minor ions (Lower Andarax,
Southeast Spain). Science of the Total Environment 297, 43-58.
Seitzinger S.P., Gardner W.S., Spratt A.K., 1991. The effect of salinity on ammonium sorption in
aquatic sediments. Implications for benthic nutrient recycling. Estuaries 14(2), 167-174.
Shalev E., Lazar A., Wollman S., Kington S., Yechieli Y., Gvirtzman H., 2009. Biased monitoring
of fresh water-salt water mixing zone in coastal aquifers. Ground Water 47(1), 49-56.
Simeoni U. and Corbau C., 2009. A review of the Delta Po evolution (Italy) related to climatic
changes and human impacts. Geomorphology 107, 64-71.
Šimunek J., Šejna M., Saito H., Sakai M., van Genuchten M.Th., 2008. The HYDRUS-1D Software
Package for Simulating the Movement of Water, Heat, and Multiple Solutes in Variably
Saturated Media, Version 4.0, HYDRUS Software Series 3, Department of Environmental
Sciences, University of California Riverside, Riverside, California, USA, pp. 315.
Stefani M. and Vincenzi S., 2005. The interplay of eustasy, climate and human activity in the
late Quaternary depositional evolution and sedimentary architecture of the Po Delta system.
Marine Geology 222-223, 19-48.
108
REFERENCES
Stueber A.M. and Walter L.M., 1991. Origin and chemical evolution of formation waters from
Silurian-Devonian strata in the Illinois basin, USA. Geochimica and Cosmochimica Acta 55,
309-325.
Stuyfzand P.J., 2008. Base exchange indices as indicators of salinization or freshening of
(coastal) aquifers. In: Proc. 20th SWIM, Naples (Fl), USA, 2008, pp. 262-265.
Stuyfzand P.J., 1989. A new hydrochemical classification of watertypes. IAHS Publication 182,
89-98.
Stuyfzand P.J. and Stuurman, R.J., 2008. Origin, distribution and chemical mass balances of
non-anthropogenic, brackish and (hyper)saline groundwaters in the Netherlands. In: Proc. 1 st
SWIM-SWICA Joint Saltwater Intrusion Conference, Cagliari (Italy), 151-164.
Teatini P., Ferronato M., Gambolati G., Bertoni W., Gonella M., 2005. A century of land
subsidence in Ravenna, Italy. Environmental Geology 47(6), 831-846.
Teatini P., Ferronato M., Gambolati G., Gonella M., 2006. Groundwater pumping and land
subsidence in the Emilia-Romagna coastland, Italy: modeling the past occurrence and the
future trend. Water Resources Research 42, 1-19.
Teuchies J., Vandenbruwaene W., Carpentier R., Bervoets L., Temmerman S., Wang C., Maris T.,
Cox T.J.S., Van Braeckel A., Meire P., 2013. Estuaries as Filters : The Role of Tidal Marshes in
Trace Metal Removal. PLoS ONE 8(8), e70381.
Thornthwaite C.W., 1948. An approach toward a rational classification of climate.
Geographical Review 38, 55-94.
Tiessen H. and Moir J.O., 1993. Total and organic carbon. In: Carte M.E. (Eds.) Soil Sampling
and Methods of Analysis, Lewis Publishers, MI (USA), pp. 187-211.
Turekian K.K., 1968. Oceans. Prentice-Hall.
Vandecasteele B., Buysse C.A., Tack F.M.G., 2008. Metal uptake in maize, willows and poplars
on impoldered and freshwater tidal marshes in the Scheldt estuary. Soil Use and Management
22(1), 52-61.
Vandenbohede A. and Lebbe L., 2012. Groundwater chemistry patterns in the phreatic aquifer
of the central Belgian coastal plain. Applied Geochemistry 27, 22-36.
Vandenbohede A., Luyten K., Lebbe L., 2008. Effects of global change on heterogeneous coastal
aquifers: a case study in Belgium. Journal of Coastal Research 24(2B), 160-170.
109
REFERENCES
Vienken T., Dietrich P., 2011. Field evaluation of method for determining hydraulic
conductivity from grain size data. Journal of Hydrology 400, 58-71.
Vukovic M. and Soro A., 1992. Determination of Hydraulic Conductivity of Porous Media From
Grain-Size Composition, Water Resources Publication, Highlands Ranch, Colorado, pp. 83.
Wang Y., Jiao J.J., Cherry J.A., 2012. Occurrence and geochemical behaviour of arsenic in a
coastal aquifer-aquitard system of the Pearl River Delta, China. Science of the Total
Environment 427-428, 286-297.
Watson T.A., Werner A.D., Simmons C.T., 2010. Transience of seawater intrusion in response
to sea level rise. Water Resources Resesrch 46(12), W12533.
Werner A.D., Bakker M., Post V.E.A., Vandenbohede A., Lu C., Ataie-Ashtiani B., Simmons C.T.,
Barry D.A., 2013. Seawater intrusion processes, investigation and management: Recent
avances and future challenges. Advances in Water Resources 51, 3-26.
Werner A.D. and Gallagher M.R., 2006. Characterisation of sea-water intrusion in the Pioneer
Valley, Australia using hydrochemistry and three-dimensional numerical modelling.
Hydrogeology Journal 14, 1452-1469.
Werner A.D. and Simmons C.T., 2009. Impact of sea-level rise on sea water intrusion in coastal
aquifers. Ground Water 47(2), 197-204.
Wood W.W., Rizk Z.S., Alsharhan A.S., 2003. Timing of recharge and the origin, evolution and
distribution of solutes in a hyperarid aquifer system, in Alsharhan, A.S., Wood, W.W., (Eds.),
Water Resources Perspectives: Evaluation, Management and Policy. Elsevier, Amsterdam, The
Netherlands, pp. 295-312.
Worthington P.F., 1998. Conjunctive interpretation of core and log data through association of
effective and total porosity models. In: Harvey, P.K. & Lovell, M.A. (Eds), Core-Log Integration,
Geological Society, London, Special Publications 136, pp. 213-223.
Yamanaka M. and Kumagai Y., 2006. Sulfur isotope constraint on the provenance of salinity in
a confined aquifer system of the southwestern Nobi Plain, central Japan. Journal of Hydrology
325, 35-55.
Zalidis G., Stamatiadis S., Takavakoglou V., Eskridge K., Misopolinos N., 2002. Impacts of
agricultural practices on soil and water quality in the Mediterranean region and proposed
assessment methodology. Agriculture Ecosystems and Environment 88, 137-146.
110
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement