ArticlePDF Available

Combining flux estimation techniques to improve characterization of groundwater-surface-water interaction in the Zenne River, Belgium


Abstract and Figures

The management of urban rivers which drain contaminated groundwater is suffering from high uncertainties regarding reliable quantification of groundwater fluxes. Independent techniques are combined for estimating these fluxes towards the Zenne River, Belgium. Measured hydraulic gradients, temperature gradients in conjunction with a 1D-heat and fluid transport model, direct flux measurement with the finite volume point dilution method (FVPDM), and a numerical groundwater flow model are applied, to estimate vertical and horizontal groundwater fluxes and groundwater–surface-water interaction. Hydraulic gradient analysis, the temperature-based method, and the groundwater flow model yielded average vertical fluxes of –61, –45 and –40 mm/d, respectively. The negative sign indicates upward flow to the river. Changes in exchange fluxes are sensitive to precipitation but the river remained gaining during the examined period. The FVPDM, compared to the groundwater flow model, results in two very high estimates of the horizontal Darcy fluxes (2,600 and 500 mm/d), depending on the depth of application. The obtained results allow an evaluation of the temporal and spatial variability of estimated fluxes, thereby helping to curtail possible consequences of pollution of the Zenne River as final receptor, and contribute to the setup of a suitable remediation plan for the contaminated study site.
Content may be subject to copyright.
Combining ux estimation techniques to improve characterization
of groundwatersurface-water interaction in the Zenne River,
J. Dujardin &C. Anibas &J. Bronders &P. Jamin &
K. Hamonts &W. Dejonghe &S. Brouyère &O. Batelaan
Abstract The management of urban rivers which drain
contaminated groundwater is suffering from high uncer-
tainties regarding reliable quantication of groundwater
uxes. Independent techniques are combined for estimat-
ing these uxes towards the Zenne River, Belgium.
Measured hydraulic gradients, temperature gradients in
conjunction with a 1D-heat and uid transport model,
direct ux measurement with the nite volume point
dilution method (FVPDM), and a numerical groundwater
ow model are applied, to estimate vertical and horizontal
groundwater uxes and groundwatersurface-water inter-
action. Hydraulic gradient analysis, the temperature-based
method, and the groundwater ow model yielded average
vertical uxes of 61, 45 and 40 mm/d, respectively.
The negative sign indicates upward ow to the river.
Changes in exchange uxes are sensitive to precipitation
but the river remained gaining during the examined
period. The FVPDM, compared to the groundwater ow
model, results in two very high estimates of the horizontal
Darcy uxes (2,600 and 500 mm/d), depending on the
depth of application. The obtained results allow an
evaluation of the temporal and spatial variability of
estimated uxes, thereby helping to curtail possible
consequences of pollution of the Zenne River as nal
receptor, and contribute to the setup of a suitable
remediation plan for the contaminated study site.
Keywords Contamination .Groundwater/surface-water
relations .Groundwater management .Risk
management .Multiple methodology
Since contaminated sites pose a signicant risk to water
resources (EC 2006; EEA 2007), national and interna-
tional regulations like the European Water Framework
Directive (EU 2000) mandate the protection of linked
groundwatersurface-water systems and ask for a reliable
assessment of uxes across the groundwatersurface-
water interface (Schmidt et al. 2008). Efcient remedia-
tion of polluted sites needs integrated management
practices, especially for complex and large-scale pollution;
risk management therefore is gaining importance in
science and engineering (Bardos et al. 2002). Van Keer
et al. (2009) describe a methodology for risk management
for polluted sites as developed in the framework of the EU
FP5 Welcome project. Another risk management plan for
browneld sites (or brownelds) has been introduced in
Flanders, Belgium, by Bronders et al. (2007,2008). The
characterization phase of a risk management plan and the
quality and reliability of its risk assessment relies strongly
on the information on the water uxes in the system.
Cirone and Duncan (2000) and Smith (2005) for example
outline the importance of proper estimates of the ground-
water ux and groundwatersurface-water interaction of
contaminated sites. In saturated conditions, the movement
of water is the main vector for pollutant transport; its
determination is needed to analyze the sourcereceptor
pathway and to evaluate movement, behaviour and fate of
pollutants. Hence, the measurement, calculation and
modelling of groundwater uxes are very important; great
Received: 22 May 2013 / Accepted: 6 June 2014
Published online: 9 July 2014
*Springer-Verlag Berlin Heidelberg 2014
J. Dujardin ()):C. Anibas :O. Batelaan
Department of Hydrology and Hydraulic Engineering,
Vrije Universiteit Brussel, Pleinlaan 2, Brussels, 1050, Belgium
Tel.: 003226293035
J. Dujardin
Cartography and GIS Research Unit Department of Geography,
Vrije Universiteit Brussel, Pleinlaan 2, Brussels, 1050, Belgium
J. Bronders :K. Hamonts :W. Dejonghe
VITO (Flemish Institute for Technological Research), Boeretang
200, Mol, 2400, Belgium
P. Jamin :S. Brouyère
Hydrology Unit Department ArGEnCo,
University of Liège, Building B52/3, Sart Tilman, 4000, Belgium
O. Batelaan
National Centre for Groundwater Research and Training School of
the Environment,
Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia
Hydrogeology Journal (2014) 22: 16571668
DOI 10.1007/s10040-014-1159-4
care has to be taken that proper methods are used and
Many groundwatersurface-water related exchange
processes take place in the hyporheic zone, i.e. the
saturated sediments beneath and beside streams and
rivers where groundwater and surface water is actively
mixed (Hayashi and Rosenberry 2002). Characterized by
relatively strong biogeochemical process rates (McClain
et al. 2003;Triskaetal.1993), the hyporheic zone is
seen as a potential sink or source of pollutants (Smith
2005). Due to hydrological connectivity, the hyporheic
zone is connected with other landscape compartments
including the aquifer and the riparian zone (Bracken and
Croke 2007; Lexartza-Artza and Wainwright 2009); the
exchange processes, therefore, are site dependent and
may have a large variability in time and space
(Sophocleous 2002).
Field methods providing spatial and/or temporal
distributed estimates of groundwatersurface-water inter-
action have been comprehensibly described by Kalbus
et al. (2006) and Rosenberry and LaBaugh (2008). They
can be based on the following: Darcys law, e.g. using
piezometer nests and boreholes placed in the riverbed and/
or in the adjacent riparian zone (Baxter et al. 2003; Cey
et al. 1998); differential stream discharge gauging (Becker
et al. 2004); numerical modeling (Cardenas and Zlotnik
2003; Fleckenstein et al. 2004); use of tracers like heat
(Anderson 2005; Anibas et al. 2011), dye, salt, chloride or
stable isotopes (Carey and Quinton 2005; Unland et al.
2013); and remote sensing (Dujardin et al. 2011; Loheide
and Gorelick 2006).
Due to scale problems (Kikuchi et al. 2012), the
heterogeneity of the underground (Schornberg et al.
2010) and limited possibilities for direct measurement,
the quantication of the groundwatersurface-water ux is
a challenging task. Moreover, all methodologies to
quantify groundwatersurface-water interaction have dis-
tinct limitations and can only capture the exchange at a
specic spatial or temporal scale (Kalbus et al. 2006).
Methods based on Darcysux are hampered by difcul-
ties in estimating the hydraulic conductivity (Chen 2000).
This realization results in a focus on using different
methods and combinations of them. Becker et al. (2004)
combine stream and streambed temperature surveys with
stream ow measurements to assess groundwater
discharge to a stream. In another study, Unland et al.
(2013) state that by combining several methods, including
differential stream gauging and chemical mass balances,
with temperature and electrical conductivity surveys, the
applicability of each technique can be evaluated. Anibas
et al. (2011), for example, state that using heat as a natural
tracer as a stand-alone technique is possible, but it is
preferably combined with other eld methods. It can be
concluded that because of the limitations and uncertainties
associated with a single method, any attempt to reliably
characterize groundwatersurface-water interactions will
benet from a multi-scale approach combining different
techniques (Kalbus et al. 2006; Hyun et al. 2011; Kikuchi
et al. 2012).
This case study, therefore, aims to determine the
vertical and horizontal groundwatersurface-water inter-
action from a browneld towards an urban river by
independently applying and combining different eld
methods and a numerical groundwater ow model of the
study area, Vilvoorde-Machelen and the Zenne River,
Belgium. The eld methods are (1) measurement of
hydraulic gradients; and (2) thermal method (Anibas
et al. 2009) to determine vertical groundwater uxes;
and (3) the nite volume point dilution method (FVPDM;
Brouyère et al. 2008) which investigates the horizontal
groundwater uxes.
Study area
The different groundwater ux estimation techniques were
applied on the industrial area of Vilvoorde-Machelen,
located about 10 km north-east of Brussels, Belgium
(Fig. 1). The study area of 10 km
is located in the Zenne
catchment, which covers an area of about 600 km
some minor rivers (including the Trawool River, the Woluwe
River and the Vondelgracht), the Brussels-Scheldt Canal
ows parallel to the Zenne River in a SN direction through
the study area (Boel 2008). Figure 2illustrates the domain of
the groundwater ow model with the positions of the
boreholes SB1, SB2 and PB9, and point T, which are the
locations where eld measurements were performed.
Figure 2also shows the contours of the phreatic groundwater
level indicating groundwater head gradients in the N and N
W direction and towards the Zenne River.
The topography in the study area ranges from 10 to
50 m, with an average value of 16 m above sea level and
with a mean slope of 1.3 %. The average precipitation is
852 mm/year (average values for 19812010; KMI 2013).
The dominant soil type of the area is silty loam, while in
the northeastern part some clay-loam occurs.
Since 1835, the Vilvoorde area has been a major
industrial site, with considerable chemical industrial activity.
The study area contains a number of well-known contam-
inated sources (OVAM 20032006; Boel 2008). Field
investigations indicate the presence of an extensive regional
contaminant plume, containing a mixture of BTEX (ben-
zene, toluene, ethylbenzene and xylene), polycyclic aromatic
hydrocarbons (PAH) and chlorinated aliphatic hydrocarbons
(CAH). While Bronders et al. (2007) estimated the size of
the contaminant plume as 1.2 by 0.6 km, Dujardin et al.
(2011)identied the Zenne River as its nal receptor.
Methods and measurements
Measured hydraulic gradients
Vertical hydraulic gradients across the streambed can be
derived by comparing water level measurements above the
streambed (level of the river) and the piezometric
groundwater levels below the riverbed. This allows
estimation of uxes through the streambed using Darcys
Law (Kalbus et al. 2006). Vertical water uxes (q) through
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
the streambed can be estimated as
where K
is the vertical hydraulic conductivity of the
streambed and iis the vertical hydraulic gradient. Surface-
water levels of the Zenne River were measured at location
T (Fig. 2), near the right riverbank, while groundwater
levels were measured at SB2 (89 m depth). The
measurements were continuously measured, every
30 min, from November 2005 till April 2007.
Porous aquifers often show substantial differences
between the horizontal and vertical hydraulic conductivity
of one or two orders of magnitude (Freeze and Cherry
1979; Chen 2000). A horizontal hydraulic conductivity of
2.5 m/d was used in the groundwater model (Boel 2008);
Fig. 1 Location of the study area within Belgium. The domain of the groundwater model is indicated by the white line; the major surface
water bodies are shown as blue lines
Fig. 2 The model area with phreatic groundwater level contours (black lines; m above sea level) showing a gradient in the northern direction and
towards the Zenne River. Position of the boreholes SB1,SB2 and PB9, as well as the locations where temperatures and hydraulic heads were
measured in the river (point Tindicated by a red dot;highlighted on the left); slug tests and FVPDM tests were performed in SB2 and PB9
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
following Anibas et al. (2011), an anisotropy factor of 9
was used to obtain a vertical hydraulic conductivity K
0.28 m/d. In case no other estimates are available to
characterize riverbed conductivities, the sandy aquifers
occurring in Flanders are assumed to have anisotropies of
about 10 (e.g. Woldeamlak 2007). Here, an anisotropy
value was used which was derived for the Flemish
lowland Aa River (Anibas et al. 2011). This approach is
justied by the fact that the Aa River and the Zenne River
have a comparable hydrology and hydrogeology.
Thermal method
At position T (Fig. 2), temperature measurements in the
riverbed were performed every hour between November 2005
and April 2007 using probes at about 0.2, 0.6 and 1.20 m depth.
For the heat transport model, however, temperatures averaged
over 6-h intervals were used. Between May 2006 and
September 2006, no temperature measurements were available.
Temperature gradients measured in the riverbed can be
used to estimate interaction between groundwater and
surface water (Anderson 2005; Constantz 2008). Given
the fact that groundwater temperatures are relatively stable
throughout the year and stream temperatures vary on a
seasonal and daily basis, the ow of groundwater causes
disturbances of the natural temperature-depth distribu-
tions. By applying combined one-dimensional (1D) heat
and uid transport modeling, these variations can be used
to compute point estimates of vertical groundwater
surface-water exchange uxes. This thermal method has
proved to be reliable (Anibas et al. 2009; Lautz 2010), not
least because gathering of thermal data, the parameter
estimation, the establishment of model boundary condi-
tions and the model calibration are relatively simple.
Different implementations of the thermal method exist
(Anderson 2005); most commonly exchange rates are
quantied by inverse modeling of measured temperature
proles (Schmidt et al. 2007; Anibas et al. 2011) or their
time series (Keery et al. 2007; Anibas et al. 2009)in
riverbeds composed of unconsolidated sediments.
In this research STRIVE (Stream RIVer Ecosystem), a
package of subroutines introduced in the FEMME
ecosystem modeling platform (Soetaert et al. 2002;
Anibas et al. 2009) was applied to determine groundwa-
tersurface-water interaction. Based on Lapham (1989)
STRIVE contains a numerical 1D heat transport module.
ρc∇⋅ TqðÞ¼
Equation (2) describes the combined heat and uid
transport of an incompressible uid through a homoge-
neous porous media, where Tis the temperature at depth z
[m] and time t[s] in the soil in K [˚C], c
the specic heat
capacity of the uid [J/kgK], ρ
the density of the uid
], cand ρare the specic heat capacity and density
of the sediment-uid matrix [J/kgK] and [kg/m
respectively. qis the seepage velocity or specic discharge
vector [m/s]. For presentation purposes, however, the units
mm/d are used to indicate uxes. Fluxes with a negative
sign stand for movement of groundwater in direction to
the surface (i.e. a gaining reach), whereas uxes of water
from the river into the soil (i.e. a losing reach) have a
positive sign. κ
is the effective thermal conductivity of
the soil-water matrix [J/smK].
The main advantage of the thermal method is its simple
parameterization, since the physical parameters (e.g. κ
or c)
have a limited range (Stonestrom and Constantz 2003;
Anderson 2005). The thermal parameters, based on Anibas
et al. (2011), who successfully simulated groundwater
surface-water interaction at the Aa River, Belgium, were
applied in this study. The Aa River is in its thermal and
hydrogeologic characteristics comparable to the Zenne River.
Tab le 1shows thermal parameters used for the STRIVE model
for the Zenne River.
A time series of surface-watertemperatures and a constant
groundwater temperature at 5 m depth constitute the
boundary conditions of the heat transport model. The
groundwater temperature is based on the mean annual
surface temperature of the study area (i.e. average 1981
2010 was 10.5 ˚C; KMI 2013). At a depth of 5 m the
groundwater temperature shows a relative constant behavior
in time, which is about 12˚C above the mean annual surface
temperature (Anderson 2005). In this case, a groundwater
temperature of 12.0 ˚C was used.
STRIVE contains routines to calibrate the vertical ux
until a best t is obtained between the simulated temperature
proles and the measured temperature distributions (Anibas
et al. 2009). Transient thermal modeling is applied in which
the rst temperature prole of a measured time series at three
depths (0.2, 0.6 and 1.2 m) is used to initialize the model,
while the others are used for calibration. The computation of
the temporal resolution of groundwatersurface-water inter-
action is possible with STRIVE by splitting the time series in
equal parts of the length of 1 week (Dujardin et al. 2011).
Groundwater ow model
The groundwater ow in the study area is simulated with a
steady-state groundwater ow model (Touchant et al.
2007; Dujardin et al. 2011) built in MODFLOW-2000
(Harbaugh et al. 2000). Figure 3shows a hydrogeological
Table 1 Applied input parameters for the STRIVE model
Parameter Symbol Value Unit
Density of the saturated
ρ1,965 kg/m
Specic heat capacity of
the saturated sediment
c1.365 J/kgK
Thermal conductivity of
the saturated sediment
1.833 J/smK
Density of water ρ
1,000 kg/m
Specic heat capacity of
4,180 J/kgK
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
fence diagram of the ve model layers (from top to
1. Silty eolian sands (Q1; 212 m thick)
2. Gravel in a silty sand matrix (Q2; 310 m thick)
3. Wedge-shaped Ghent Formation of which the upper
part consists of densely packed sand layer (GePi; 4 m
4. The lower part of the Ghent Formation consists of clay-
containing silty sand (GeMe; 2 m thick)
5. Tielt Formation (TiEg; 2030 m thick) consisting of
silty, glauconite ne sands
The hydrostratigraphy was interpreted by 68 borehole
loggings and 179 cone penetration tests, available from
the Geological Database for the Subsoil of Flanders (DOV
2008). Parameters for hydraulic conductivity of the model
layers are described in Table 2; the storage coefcient S
1/m was obtained from literature (i.e. Morris and
Johnson 1967).
The model boundary conditions are a combination of
the employed MODFLOW packages: River, General
Head, Specied Head and Recharge. The lower boundary
is dened at the top of the Tertiary clay-rich Kortrijk
Formation; considered as almost impervious, it is thus
implemented as a no-ow boundary. Beginning from the
north, the other model boundaries are dened in clockwise
direction as follows (Fig. 1): no ow and river boundary
(at the Trawool River), constant head at the eastern
boundary (based on head measurements inside and outside
the model area), no ow boundary in the south and
constant head due to the Brussels-Scheldt Canal in the
west. The no ow boundaries are justied by the regional
groundwater ow which is parallel to the northern and
southern model boundaries.
The upper boundary is dened by groundwater
recharge. Because the spatial distribution of the ground-
water recharge has an impact on the modeled groundwa-
tersurface-water interaction, distributed recharge
estimations were acquired from Dujardin et al. (2011)
using the WetSpass methodology (Batelaan and De Smedt
2007). The average recharge used in this model is
159 mm/year with a standard deviation of 91 mm/year.
To study groundwatersurface-water interaction in the
Zenne River the River package (RIV) is employed. Field
investigations delivered average values for river water
level h=9.6 m and river depth d=1.2 m. The conductance
of the river sediments C=2.66 m
/d is calculated as
where, Kthe hydraulic conductivity of the bottom sedi-
ments [m/d], Athe surface area of the bottom sediments
] and bis the thickness of the sediment layer [m]. A
more extensive description of the applied groundwater
model and its internal and external boundary conditions
can be found in Touchant et al. (2007), Boel (2008) and
Dujardin et al. (2011).
FVPDM tests
Slug tests, hydraulic gradients and pumping tests in
piezometers have traditionally been used to estimate
hydraulic conductivities and to constrain groundwater
Table 2 Applied hydraulic conductivity (K) values for the ground-
water ow model
Layer Designation K
[m/d] K
1 Q1 2.5 2.5
2 Q2 12.5 6
3 GeMe 0.5 0.5
4 GePi 4 3
5 TiEg 6 4
Fig. 3 NS and EW hydrogeological cross-sections of the subsurface of the study area (Q1 silty eolian sands; Q2 gravel; GePiGeMe
Ghent Formation; TiEg Tielt Formation; Dujardin et al. 2011, with permission from Elsevier)
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
uxes (Butler 1998). Another technique applicable on
piezometers is the nite volume point dilution method
(FVPDM; Brouyère et al. 2008).
This novel method was developed to overcome the
difculties of implementing classical point dilution meth-
ods (PDM), like instantaneous and uniform mixing of
tracer into the well without disturbance of groundwater
uxes (Drost et al. 1968; Haveley et al. 1967). The
FVPDM is a tracer technique that has been introduced in
2003 (Brouyère 2003; Brouyère et al. 2008), generalizing
all the single-well point dilution tests to almost any tracer
injection scenario. This method is based on a mathemat-
ical and numerical model for the tracer injection into a
well and considers the mass balance of the injection of
tracer uid and transiting groundwater ow passing
through the well screen (Brouyère et al. 2005). The
analytical solution of this model applied to a single well
tracer technique enables the accurate measurement of
transit ow rate and, thus, of the Darcy uxes (Eq. 4).
As a eld method, the FVPDM is based on a controlled
continuous injection of a tracer into a well and on the
monitoring of its concentration over time. During the
whole experiment, the water column within the well is
mixed to insure a homogeneous repartition of the tracer
mass. The tracer concentration is proportional to the
groundwater uxes that ush the tracer out of the well.
The FVPDM experiment can be maintained as long as
tracer uid and power supply for injection and mixing are
available, in which case a continuous temporal monitoring
of the variations of groundwater uxes is possible.
QinCin QinCin QoutCw;0
exp Qout
Qout ¼Qin þQin
and C
are tracer concentrations [M/L
] in the
well, in the injection water and in the well at time t
respectively. Q
and Q
/T] are the injection
rate, the transit ow rate corresponding to the groundwater
ow intercepted by the well screen that is directly related
to the apparent Darcysux (vD) and the ow rate leaving
the well through the screen carrying tracer at concentra-
tion C
is the volume of water in the injection well,
assumed to be constant.
The FVPDM tests have been executed in SB2 and
PB9, beginning with piezometer SB2_F2 in September
2008 (Fig. 2). This piezometer with a diameter of 2 inches
(5.1 cm) consists of two well screens, F1 (78 m depth)
and F2 (910 m depth). The tracer uranine was injected
with two ow steps: T
at 0.01 L/min for 181 min and T
at 0.02 L/min for 128 min. Piezometric head, temperature,
turbidity and uranine concentrations were monitored for
350 min from the beginning of the test using a LevelTroll
probe (In-Situ Inc., Ft. Collins, CO, USA) and with a eld
uorimeter (GGUN-FL30#1370; Geomagnetism Group,
University of Neuchâtel, Neuchâtel, Switzerland) respec-
tively. A second FVPDM test was performed on PB9_F1
(screen 4.36.3 m depth) in November 2008. Uranine
tracer was injected at 0.006 L/min for 120 min, with
piezometric head, temperature, turbidity and uranine
concentration monitored for 180 min from the beginning
of the test respectively using the LevelTroll probe and the
eld uorimeter.
Measured hydraulic gradients
Based on hydraulic gradients between the piezometric
groundwater level and the river stage, Fig. 4shows the
ux estimates as a blue line. The uxes are negative (i.e. a
gaining river), varying around 60 mm/d. The lowest
vertical uxes occur in February 2006, with an average of
36 mm/d, the highest values are observed in December
2006 and March 2007, with averages of 84 mm/d.
Thermal gradients
Figure 4also indicates average weekly vertical water
uxes (red line) simulated with STRIVE. It can be
observed that only groundwater discharge to the Zenne
River occurred, on average about 45 mm/d; missing parts
in the line graph indicate periods were no reliable output
could be generated or where no temperature data were
available. This is especially the case for periods where no
strong temperature gradients are present between the
groundwater and the surface water; hence, in spring and
autumn (Anibas et al. 2009).
Groundwater flow model
The groundwater model was calibrated using observed
heads from 27 observation wells with measured heads
between 1999 and 2006. The calibration resulted in a root
mean squared error (RMSE) of 0.32 m and an absolute
error (AE) of 0.02 m. The general orientation of the
groundwater ow is north-west (Fig. 2). In the western
part of the area, the groundwater ow is oriented towards
the east, because of the draining effect of the Zenne River.
A water budget for the Zenne River was calculated; Fig. 5
and Table 3show the global inow and outow of the
Zenne River in the study area.
It is clear that the Zenne River is receiving groundwa-
ter from:
1. The Brussels-Scheldt Canal (B-S Canal), producing a
ux from the west
2. The regional groundwater ow, coming from the south-
The B-S Canal discharges 7,054 m
/d (over a length of
3,830 m), representing 72 % of the total inow into the
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
Zenne River; the remaining 28 % of the total inow is
coming from the regional groundwater ow (2,690 m
The upper geological layer (Q1) is characterized by
horizontal uxes between 0 and 80 mm/d. Figure 6shows
vertical groundwater uxes in the study area as simulated
with the numerical groundwater ow model. The colored
cells show upward uxes, while the white cells have
downward uxes. From Fig. 6it is clear that the Zenne is
gaining, the uxes vary from 12 to 40 mm/d.
FVPDM tests
The results of the rst FVPDM test (SB2_F2) showed that
the tracer injection rate was too high for the very low
groundwater ow prevailing during the test. A better
dimensioning was achieved in the second FVPDM test
PB9_F1. Figure 7shows the comparison of measured and
simulated concentration of tracer using FVPDM method.
The groundwater ow velocity was determined by
simulating the tracer elution curve and comparing this
curve with experimental data. Simulated concentrations
were adjusted by modifying only the apparent Darcys
ux (vD). The other terms of Eq. (3) are based on
experimental conditions.
In Fig. 7, the solid line corresponds to the best
adjustment of Darcysux for the experimental con-
ditions. The dashed and dotted lines indicate the
sensitivity of the method to the magnitude of Darcys
ux. The ascending part of the simulated curves matches
almost perfectly with the experimental measurements. A
small gap can be observed at the beginning of the test on
PB9_F1. This is probably due to a longer homogeniza-
tion time of the tracer over the whole height of the well.
The FVPDM tests allowed calculating a horizontal
Darcysux of 2600 mm/d for SB2_F2 and a Darcys
ux of 500 mm/d for PB9_F1. The differences observed
between the horizontal Darcysux of SB2_F2 and
PB9_F1 can be explained by local variation of aquifer
hydraulic conductivity as is frequently observed in such
alluvial aquifers.
Table 4summarizes the different estimations of vertical
uxes using three methodologies: hydraulic gradients, the
thermal method and the groundwater ow model. To
compare the resulting vertical uxes from the former with
the vertical uxes obtained from the groundwater ow
model, Table 4contains the vertical ux simulated in the
model cell located around SB2.
Table 4indicates that all three methods give fairly
similar vertical ux estimates, observing an upward ux
from the groundwater towards the Zenne River. Method 1
gives an average of 61 mm/d of water discharge; method
Fig. 4 Calculated daily vertical groundwater ux (blue line) based on measured hydraulic gradients and weekly measured thermal
gradients using the STRIVE model (red line) and daily precipitation (black lines). Short-term uctuations, sensitive to precipitation,
dominate. The river is gaining throughout the 16-month period; no clear seasonal pattern is visible
Fig. 5 Schematic view of the different components of the water
budget of the Zenne River (B-S Canal Brussels-Scheldt Canal, GW
groundwater). See Table 3
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
2, the thermal method, gives a bit less, around 45 mm/d;
and method 3, the groundwater ow model, gives around
40 mm/d. Since the results of the three independent
methodologies show similar estimates, it can be expected
that they give a realistic and reliable value of the vertical
groundwatersurface-water interaction at the Zenne River.
Notice that a complete similitude is unlikely; the two
methods, the hydraulic gradient method and the thermal
method, are methodologically distinct and are based on
different assumptions. The disagreement in exchange
uxes between the hydraulic gradient method and the
thermal method can be attributed to the uncertainty
regarding the hydraulic conductivity of the riverbed and
the alluvial sediments.
Regarding the thermal method, the magnitude of the
estimated exchange uxes is comparable with other
studies of Flemish rivers as described by Anibas et al.
(2009,2011), but are fairly low in comparison with some
other works like Keery et al. (2007). The fact that
STRIVE integrates the exchange uxes over the vertical
model domain of 5 m depth may partly explain the big
differences in ux estimates.
It is possible to use the ux estimates of the thermal
method together with the measured hydraulic gradient for
the estimation of the vertical hydraulic conductivity. By
doing so, the obtained ux estimates of the hydraulic
gradient method would be similar to those of the thermal
method; both methods are not applied independently
anymore. For the Zenne River, since the hydraulic
gradient method yields higher results then the others the
hydraulic conductivity is reduced to 0.21 m/d. It can be
stated that the used vertical hydraulic conductivity was
overestimated. However, keeping in mind the large
uncertainties regarding the estimation of (vertical) hydrau-
lic conductivities, the initial value of 0.28 m/d seems to be
well chosen. In fact, Anibas et al. (2011) derived their
estimate of the vertical riverbed hydraulic conductivity
with a similar approach.
The thermal method and the hydraulic gradient method
also resolve the temporal behavior of the groundwater
surface-water interaction. In Fig. 4, it can be seen that the
results of both methods correspond and show comparable
values and trends. It is clear that hydraulic gradients
deliver ux estimates with the highest temporal resolution.
Table 3 Absolute (m
/d) and relative (%) inows and outows for the Zenne River (length 3,830 m). The values in the table are related to
the values in Fig. 5(B-S Canal Brussels-Scheldt Canal, GW groundwater)
Absolute inow
Relative inow
Absolute outow
Relative outow
West B-S Canal discharge (1in Fig. 5) 7,054 72 0 0
South-eastern regional GW ow (2in Fig. 5) 2,690 28 0 0
River drainage (1+2in Fig. 5) 0 0 9,744 100
Fig. 6 Vertical groundwater uxes in the study area as simulated with the numerical groundwater ow model. The colored cells show
upward uxes, while the white cells have downward uxes
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
Figure 4shows a fairly constant ux for the whole
simulation period, and strong seasonal variations are not
indicated. The short-term uctuations in groundwater
surface-water interaction can be explained by the large
variations in stream discharge of the Zenne River. As an
urban river, it is quite sensitive to precipitation; hence, the
stream discharge shows a quick response in function of
rainfall events, which has an impact on the vertical
groundwatersurface-water interaction of the Zenne
River. Because of changing hydraulic gradients, the
vertical discharge will decrease when river water levels
rise; when the stream discharge decreases again, the
vertical discharge into the Zenne River will increase
again; however, the river remained gaining for the whole
investigated period.
The left side of Table 5compares horizontal Darcys
uxes simulated by the groundwater ow model for two
locations, while the right side presents horizontal Darcys
uxes estimated by the FVPDM for the same locations.
The groundwater ow model estimated a Darcysux of
150 mm/d in SB2_F2, and of 100 mm/d in PB9_F2. The
FVPDM estimated a Darcysux of 2,600 mm/d in
SB2_F2 and of 500 mm/d in PB9_F2, which is much
higher, 517 times, than obtained from the groundwater
ow model. This can be explained by the fact that
FVPDM, as a single-well dilution technique, only pro-
vides a groundwater ow representative of the moment
when the test is performed and at close vicinity of the
tested well. FVPDM uxes are instantaneous estimations,
which depend strongly on experimental conditions, like
e.g. river stage variations. These variations generate local
pressure changes in the aquifer and thus changes in
groundwater uxes around the injection wells (Brouyère
et al. 2008). This is especially the case for SB2_F2 where
the injection well is situated at 3.5 m of the Zenne River,
making it very sensitive to river stage variations and the
differences in pressure due to these variations. The
groundwater ow model on the other hand is a steady-
state model, which smooths out the temporal variation of
groundwater ow.
The hydraulic conductivity of the aquifer in the vicinity
of the tested well can be deduced from the groundwater
ux measured by the FVPDM and the local hydraulic
gradient. A gradient of 0.0024 around well PB9 and
0.0033 around SB2 has been measured at the time of the
experiment based on head measurements in neighboring
piezometers, which gives a hydraulic conductivity of 208
and 788 m/d respectively for PB9 and SB2. The hydraulic
properties of the groundwater ow model are average
aquifer values, representing a much larger area than the
ones investigated by a single well dilution test, leading to
the much lower mean hydraulic conductivity of layer (Q2)
of 12.5 m/d. Hence, the differences between the ground-
water uxes measured by the FVPDM in the eld and the
ones used in the groundwater ow model are caused by
spatial heterogeneities of the porous media of the aquifer
and by the temporal dynamics of groundwater ow close
to the Zenne River.
The results show that horizontal uxes might be larger
than the vertical ones, and they might be much more
variable as well. Not considering the extreme value of
2,600 mm/d, it is expected that the contribution of a
0 30 60 90 120 150 180
Uranine concentration (ppb)
vD = 6.0·10 m/s
vD = 5.0·10 m/s
vD = 4.0·10 m/s
Experimental measured data
Tracer injection flow rate
vD = 1.0 10 m/s
vD = 3.5·10 m/s
vD = 3.0·10 m/s
vD = 2.7·10 m/s
vD = 1.0·10 m/s
0 30 60 90 120 150 180 210 240 270 300 330 360
Uranine concentration (ppb)
Tracer injection flow rate (×10 m/h)
-3 3
SB2_F2 PB9_F1
Tracer injection flow rate (×10 m/h)
-3 3
Time (min)
Time (min)
Fig. 7 Monitored and modelled concentration evolution and injection ow rates. The solid black curve provides the Darcysux with the
t. The rst FVPDM test was at piezometer SB2_F2, the second text at PB9_F1
Table 4 Comparison of estimated vertical uxes obtained from various methods [unit: mm/d]. Negative values indicate upward uxes
Method 1: hydraulic gradients Method 2: temperature gradients Method 3: GW ow model
Time-dependent Time-dependent Steady-state
Period November 2005April 2007 November 2005April 2007 Calibration period 2005
Results Vertical ux at the river Vertical ux at the river Vertical ux at the river
Location SB2 8 m upstream from SB2 Around SB2
Flux (mm/d) 61 45 40
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
horizontal ux component to groundwatersurface-water
exchange will still be smaller than the vertical component.
As is described in literature (e.g. Rosenberry and LaBaugh
2008), the results from the groundwater ow model (Boel
2008; Dujardin et al. 2011) show that the ow lines bend
strongly below the river cells, leading to predominantly
vertical groundwatersurface-water exchange at the river
riverbed interface. Hence, with respect to an assessment of
contaminant transport towards the Zenne River, the
vertical groundwatersurface-water interaction is the most
important pathway. Boel (2008) and Dujardin et al. (2011)
show that the pollution sources are at such a distance from
the river that their potential horizontal pollution pathways
are predominantly located in model layers Q2 and TiEg
(Fig. 3).
The differences between the horizontal and vertical
estimates and the differences between the estimates of
the different methods show that there is a possibility to
over or underestimate the uxes if one makes use of
only a single method. While the groundwater model in
principal simulates the total groundwater ux towards
the river, these estimates are based on assumptions
made for the whole model domain, including the more
or less realistic denition of the riverriverbed inter-
face. Especially for small-scale groundwater models,
the simplication of this interface may lead to sub-
optimal estimates of groundwatersurface-water
Field methods have the advantage to better describe the
interaction, since they are based on actual eld measure-
ments at discrete locations. As point measurements, on the
other hand, they lack the possibility to extrapolate these
values on spatial scales; thus, only the combination of
different methods will deliver a reliable perception of the
variation of groundwatersurface-water interaction in
space and time and enables understanding of the uncer-
tainty and heterogeneity of the investigated physical
Knowledge about water uxes is important for the
assessment of the contaminant transport from the sur-
rounding brownelds towards the Zenne River. This
study, therefore, compared four independent methodolo-
gies in order to characterize groundwatersurface-water
interactions between a river, the adjacent hyporheic zone
and aquifer.
As the nal receptor, the ux estimates show that
the Zenne River receives a little less than one third of
the discharging groundwater from the eastern part of
the study area. Using the different methodologies,
vertical and horizontal exchange in the study area
were investigated. The hydraulic gradient and the
thermal method yield temporal variations and the
steady-state groundwater ow model indicates the
spatial variation in the whole study area, including
horizontal uxes. Since the hydraulic gradient and the
thermal method are limited to vertical exchange rates,
a fourth methodology, the FVPDM was chosen to give
an independent estimation of horizontal Darcysuxes.
The vertical ux estimates resulted in comparable
groundwater discharge uxes of 61 to 40 mm/d to the
receiving Zenne River. Being in line with studies of
other Flemish lowland rivers, these values have therefore
a strong reliability. The hydraulic gradient method and
the thermal method also show comparable results and
trends in the temporal distribution of groundwater
surface-water exchange. Regardless of different rainfall
events, the river remained gaining throughout the entire
investigated period of 16 months (Fig. 4). With a weak
seasonal pattern, short-term variations dominate the
vertical exchange, which is explained by the sensitivity
of the groundwater discharge on the water level of the
mostly urban Zenne River in connection with heavy
rainfall events.
The obtained results for horizontal ow showed
differences, explained by the strong sensitivity to eld
conditions of the transient FVPDM method. These differ-
ences in estimations indicate how strong the uncertainties
are when relying solely on estimations of a single eld
method or on a groundwater ow model and also
emphasize the importance of investigating temporal
variations of groundwater uxes and spatial heterogene-
ities of the hydraulic properties of an aquifer. A risk
management plan, hence, should not rely on site character-
izations from a single eld campaign, since these
measurements and results might not be representative for
the uctuations of the groundwater ow in particular and
aquifer dynamics in general.
The combination of eld techniques, therefore,
improves the capacity of a risk management plan for
brownelds and adjacent surface water and groundwater
bodies. By application of different techniques, uncertain-
ties of the estimates are reduced, while the condence as
well as the credibility of the applied methods and the risk
management is improved.
Table 5 Comparison of estimated horizontal Darcysuxes obtained from the groundwater ow model and the FVPDM method [unit:
Method 3: GW ow model Method 4: FVPDM
Steady-state Steady-state
Period Calibration period 2005 Sept 2008 Nov 2008
Results Darcysux (horizontal) Darcysux (horizontal)
Location (depth) SB2_F2 (9.5 m) PB9_F1 (5.3 m) SB2_F2 (9.5 m) PB9_F1 (5.3 m)
Flux (mm/d) 150 100 2,600 500
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
Acknowledgements Financial support for this research came
from the Belgian Science Policy Ofce in the framework of the
Flux-based Risk Assessment of the impact of Contaminants on
Water resources and ECOsystems (FRAC-WECO SD/TE/02)
Anderson MP (2005) Heat as a groundwater tracer. Ground Water
Anibas C, Fleckenstein J, Volze N, Buis K, Verhoeven R, Meire P,
Batelaan O (2009) Transient or steady state? using vertical
temperature proles to quantify groundwatersurface water
exchange. Hydrol Process 23:21652177. doi:10.1002/
Anibas C, Buis K, Verhoeven R, Meire P, Batelaan O (2011) A
simple thermal mapping method for seasonal spatial patterns of
groundwatersurface water interaction. J Hydrol 397(12):93
104. doi:10.1016/j.jhydrol.2010.11.036
Bardos P, Lewis A, Nortcliff S, Matiotti C, Marot F, Sullivan TL
(2002) Review of decision support tools for contaminated land
management, and their use in Europe. Austrian Federal
Environment Agency, Vienna on behalf of CLARINET, Vienna
Batelaan O, De Smedt F (2007) GIS-based recharge estimation by
coupling surface-subsurface water balance. J Hydrol 337:337
Baxter C, Hauer FR, Woessner WW (2003) Measuring groundwa-
terstream water exchange: new techniques for installing
minipiezometers and estimating hydraulic conductivity. Trans
Ash 132(3):493502
Becker MW, Georgian T, Ambrose H, Siniscalchi J, Fredrick K
(2004) Estimating ow and ux of groundwater discharge using
water temperature and velocity. J Hydrol 296(14):221233.
Boel S (2008) Hydrogeologische studie van de browneld-site
Vilvoorde-Machelen: grondwaterstromingen transport
[Hydrogeological study of the Vilvoorde-Machelen browneld:
groundwater uxes and transport]. MSc Thesis, Katholieke
Universiteit Leuven, Belgium, 125 pp
Bracken LJ, Croke J (2007) The concept of hydrological connec-
tivity and its contribution to understanding runoff-dominated
geomorphic systems. Hydrol Process 21(13):17491763.
Bronders J, Touchant K, Van Keer I, Patyn J, Provoost J (2007) The
characterization of contamination and the role of hydrogeology
in the risk management of a mega browneld site. Proceedings
of IAH XXXV Congress: Groundwater and Ecosystems,
Lisbon, September 2007, 270 pp
Bronders J, Touchant K, Van Keer I, Patyn J (2008) A risk
management plan for the redevelopment of a browneld site: an
example. ConSoil 2008: proceedings of the 10th International
UFZ-Deltares TNO Conf. on Soil Water Systems, Milan, June
2008, pp 4754
Brouyère S (2003) Modeling tracer injection and well-aquifer
interactions: a new mathematical and numerical approach.
Water Resour Res 39(3):15. doi:10.1029/2002WR001813
Brouyère S, Carabin G, Dassargues A (2005) Inuence of injection
conditions on eld tracer experiments. Ground Water
43(3):389400. doi:10.1111/j.1745-6584.2005.0041.x
Brouyère S, Batlle-Aguilar J, Goderniaux P, Dassargues A (2008) A
new tracer technique for monitoring groundwater uxes: the nite
volume point dilution method. J Contam Hydrol 95:121140
Butler JJ (1998) The design, performance and analysis of slug tests.
Lewis, Boca Raton, FL, 262 pp
Cardenas MB, Zlotnik VA (2003) Three-dimensional model of
modern channel bend deposits. Water Resour Res 39(6):1141.
Carey SK, Quinton WL (2005) Evaluating runoff generation during
summer using hydrometric, stable isotope and hydrochemical
methods in a discontinuous permafrost alpine catchment.
Hydrol Process 19(1):95114
Cey EE, Rudolph DL, Parkin GW, Aravena R (1998) Quantifying
groundwater discharge to a small perennial stream in southern
Ontario, Canada. J Hydrol 210:2137
Chen XH (2000) Measurement of streambed hydraulic conductivity
and its anisotropy. Environ Geol 39(12):13171324
Cirone PA, Duncan PB (2000) Integrating human health and
ecological concerns in risk assessments. J Hazard Mater 78(1
Constantz J (2008) Heat as a tracer to determine streambed water
exchanges. Water Resour Res 44, W00D10. doi: 10.1029/
DOV (2008) Databank Ondergrond Vlaanderen [Database on the
subsurface in Flanders]. Departement Mobiliteit en Openbare
Werken, Vlaamse Gemeenschap.
Accessed 18 Jan 2011
Drost W, Klotz D, Arnd K, Heribet M, Neumaier F, Rauert W (1968)
Point dilution methods of investigation ground water ow by
means of radioisotopes. Water Resour Res 4(1):125146
Dujardin J, Batelaan O, Canters F, Boel S, Anibas C, Bronders J
(2011) Improving surface-subsurface water budgeting using
high resolution satellite imagery applied on a browneld. Sci
Total Environ 409:800809. doi:10.1016/
EC (2006) Soil protection: the story behind the strategy. European
Commission, Brussels, 28 pp
EEA (2007) CSI 015: progress in management of contaminated
sitesassessment published Aug 2007. European Environment
Agency, Copenhagen.
ress-in-management-of-contaminated-1. Accessed 24 Sept 2008
EU (2000) Directive 2000/60/EC of the European Parliament and of
the Council establishing a framework for Community action in
the eld of water policy. OJEC L327, 72 pp
Fleckenstein J, Anderson M, Fogg G, Mount J (2004) Managing
surface watergroundwater to restore fall ows in the Con-
sumnes River. J Water Resour Plann Management-ASCE 130
(4):301310. doi: 10.1016/(ASCE)0733-9496(2004)130:4(301)
Freeze RA, Cherry JA (1979) Groundwater. Prentice-Hall, Engle-
wood Cliffs, NJ
Harbaugh A, Banta E, Hill M, McDonald M (2000) MODFLOW-
2000, The U.S. Geological Survey modular groundwater model:
user guide to modularization concepts and the groundwater ow
process. US Geol Surv Open-File Rep 00-92
Haveley E, Moser H, Zellhofer O, Zuber E (1967) Borehole dilution
techniques: a critical review. Isotopes in Hydrology, IAEA,
Hayashi M, Rosenberry DO (2002) Effects of groundwater
exchange on the hydrology and ecology of surface water.
Ground Water 40(3):309316
Hyun Y, Kim H, Lee SS, Lee KK (2011) Characterizing streambed
water uxes using temperature and head data on multiple scales
in Munsan stream, South Korea. J Hydrol 402:377387
Kalbus E, Reinstorf F, Schirmer M (2006) Measuring methods for
groundwater-surface water interactions: a review. Hydrol Earth
Syst Sci 10:873887
Keery J, Binley A, Crook N, Smith JWN (2007) Temporal and
spatial variability of groundwatersurface water uxes: devel-
opment and application of an analytical model using tempera-
ture time series. J Hydrol 336:116. doi:10.1016/
Kikuchi CP, Ferré TPA, Welker JM (2012) Spatially telescoping
measurements for improved characterization of ground water
surface water interactions. J Hydrol 446447:112.
KMI (2013) Koninklijk Meteorologisch Instituut (RMI Royal
Meteorological Institute), Brussels, Belgium. http:// Accessed 8 Oct 2013
Lapham WM (1989) Use of temperature proles beneath streams to
determine rates of vertical ground-water ow and vertical
hydraulic conductivity. US Geol Surv Water-Suppl Pap 2337
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
Lautz LK (2010) Impacts of nonideal eld conditions on vertical
water velocity estimates from streambed temperature time
series. Water Resour Res 46, W01509. doi:10.1029/
Lexartza-Artza I, Wainwright J (2009) Hydrological connectivity:
linking concepts with practical implications. Catena 79:146
152. doi:10.1016/j.catena.2009.07.001
Loheide SP, Gorelick SM (2006) Quantifying stream-aquifer
interactions through the analysis of remotely sensed thermo-
graphic proles and in situ temperature histories. Environ Sci
Technol 40(10):33363341. doi:10.1021/es0522074
McClain ME, Boyer EW, Dent CL, Gergel SE, Grimm NB,
Groffmann PM, Hart SC, Harvey JW, Johnston CA, Mayorga
E, McDowell WH, Pinay G (2003) Biogeochemical hot spots
and hot moments at the interface of terrestrial and aquatic
ecosystems. Ecosystems 6(4):301312
Morris DA, Johnson AI (1967) Summary of hydrological and
physical properties of rock and soil materials as analyzed by the
Hydrological Laboratory of the U.S. Geological Survey 1948
1960. USGS, Reston, VA
OVAM (20032006) Veiligheids: en voorzorgsmaatregelen te
Machelen, voormalige Biolux site [Security and precautionary
measures for the formal Biolux site in Machelen]. Website
OVAM. Accessed
27 March 2009
Rosenberry DO, LaBaugh JW (2008) Field techniques for estimat-
ing water uxes between surface water and groundwater. US
Geoll Surv Tech Methods 4-D2, 128 pp
Schmidt C, Conant B Jr, Bayer-Raich M, Schirmer M (2007)
Evaluation and eld-scale application of an analytical method to
quantify groundwater discharge using mapped streambed
temperatures. J Hydrol 347:292307. doi:10.1016/
Schmidt C, Kalbus E, Krieg R, Bayer-Raich M, Leschik S, Reinstorf F,
Martienssen M, Schirmer M (2008) Schadstoffmassenströme
zwischen Grundwasser, Flussbettsedimenten und Oberächen-
wasser am regional kontaminierten Standort Bitterfeld [Contami-
nant mass ows between groundwater, riverbed sediments and
surface water at the regional contaminated site in Bitterfeld].
Grundwasser 13:133146. doi:10.1007/s00767-008-0076-7
Schornberg C, Schmidt C, Kalbus E, Fleckenstein JH (2010)
Simulating the effects of geologic heterogeneity and transient
boundary conditions on streambed temperatures: implications
for temperature-based water ux calculations. Adv Water
Resour 33(11):13091319. doi:10.1016/
Smith JWN (2005) Groundwater-surface water interactions in the
hyporheic zone. Science report SC030155/SR1, Environment
Agency, Bristol, UK
Soetaert K, De Clippele V, Herman P (2002) FEMME, a exible
environment for mathematically modeling the environment.
Ecol Model 151(23):177193
Sophocleous M (2002) Interactions between groundwater and
surface water: the state of the science. Hydrogeol J 10:5267
Stonestrom DA, Constantz J (2003) Heat as a tool for studying the
movement of ground water near streams. US Geol Surv Circ
1260, 105 pp
Touchant K, Bronders J, Patyn J, Seuntjens P, Lookman R, Joris I
(2007) Browneld problematiek Vilvoorde-Machelen: risicoma-
nagement regionale grondwaterverontreiniging [The problemat-
ic Vilvoorde-Machelen browneld: risk management and
regional groundwater pollution]. Report 2007/IMS/43, VITO,
Mol, Belgium
Triska FJ, Duff JH, Avanzino RJ (1993) The role of water exchange
between a stream channel and its hyporheic zone in nitrogen
cycling at the terrestrial aquatic interface. Hydrobiologica
Unland NP, Cartwright I, Andersen MS, Rau GC, Reed J, Gilfedder
BS, Atkinson AP, Hofmann H (2013) Investigating the spatio-
temporal variability in groundwater and surface water inter-
actions: a multi-technique approach. Hydrol Earth Syst Sci
17:34373453. doi:10.5194/hess-17-3437-2013
Van Keer I, Lookman R, Bronders J, Touchant K, Patyn J, Joris I,
Wilczek D, Vos J, Dewilde J, Van de Wiele K, Maebe P, De
Naeyer F (2009) Examples of risk management in Flanders for
large scale groundwater contamination. In: Hlavinek P, Popov-
ska C, Marsalek J, Mahrikova I, Kukharchyk T (eds) Risk
management of water supply and sanitation systems impaired
by operational failures, natural disasters and war conicts.
NATO Security through Science Series, Sub-Series C: Environ-
mental Security, Rowan and Littleeld, Lanham, MD, pp 241
Woldeamlak ST (2007) Spatio-temporal impacts of climate and
land-use changes on the groundwater and surface water
resources of a lowland catchment, PhD Thesis, Vrije Universi-
teit Brussels, Belgium
Hydrogeology Journal (2014) 22: 16571668 DOI 10.1007/s10040-014-1159-4
... D'un point de vue quantitatif, l'intégration des rivières dans les modèles d'échange nappe-rivière en intégrant les hétérogénéités géologiques et les processus sur différentes échelles est un autre défit majeur (Fleckenstein et al., 2006;Flipo, 2013). D'un point de vue appliqué, de nombreuses recherches ont été menées avec comme but la gestion qualitative ou quantitative de la ressource en eau du continuum surface-souterrain (Dujardin et al., 2014;Engeler et al., 2011;Goderniaux et al., 2009;Kurtz et al., 2014;Leaf et al., 2015;Peyrard et al., 2008) et notamment en France, où une forte proportion de l'eau potable consommée provient de nappes alluviales d'accompagnement en relation directe avec les entités de surfaces (Labarthe, 2016;Lalot, 2014;Loizeau, 2013;Pryet et al., 2015b). ...
... Cette échelle d'étude répond donc à certains processus locaux comme régionaux, spatialement restreints comme étendus et temporellement rapides comme très lents. Bien que plutôt rare comparé à l'échelle locale, il existe un certain nombre d'études sur les échanges nappe-rivière à l'échelle intermédiaire (Doppler et al., 2007;Dujardin et al., 2014;Engeler et al., 2011;Hunt et al., 2006;Loizeau, 2013). Généralement traités sous l'angle de la modélisation, ces études utilisent des mesures locales de charges (Doppler et al., 2007;Loizeau, 2013), température (Dujardin et al., 2014;Engeler et al., 2011) pour calibrer un modèle numérique à l'échelle intermédiaire. ...
... Bien que plutôt rare comparé à l'échelle locale, il existe un certain nombre d'études sur les échanges nappe-rivière à l'échelle intermédiaire (Doppler et al., 2007;Dujardin et al., 2014;Engeler et al., 2011;Hunt et al., 2006;Loizeau, 2013). Généralement traités sous l'angle de la modélisation, ces études utilisent des mesures locales de charges (Doppler et al., 2007;Loizeau, 2013), température (Dujardin et al., 2014;Engeler et al., 2011) pour calibrer un modèle numérique à l'échelle intermédiaire. Le but étant souvent la nécessité d'utiliser différents types de données pour limiter les erreurs d'estimations (Fleckenstein et al., 2010;Hunt et al., 2006;Sophocleous, 2002) ...
Full-text available
Les interactions entre les eaux souterraines et les eaux de surface sont complexes et jouent un rôle prépondérant dans le fonctionnement des hydrosystèmes, tant en termes quantitatifs (soutien des étiages, évènements de crues) que qualitatifs (transport de polluants d’origines agricole ou industrielle). Ces problématiques sont abordées à travers la caractérisation et la modélisation des échanges nappe-rivière à l’échelle intermédiaire [1-10km], avec pour objet d’étude le champ captant de Thil-Gamarde qui alimente la métropole de Bordeaux en eau potable. Ce champ captant est traversé par un cours d’eau, la Jalle de Saint-Médard, vecteur historique de pollutions. Une première phase est constituée d’un volet théorique, visant à améliorer les modalités d’intégration des rivières dans les modèles hydrogéologiques de l’échelle intermédiaire à régionale, à travers une condition de conductance (type Cauchy). Un outil a été développé permettant leur intégration à partir d’informations sur des paramètres hydrodynamiques et géométriques de la nappe et de la rivière, jusqu’à présent souvent négligés. L’étude du champ captant de Thil-Gamarde, débute avec l’établissement d’un modèle conceptuel des écoulements et du transport en nappe. L’approche multidisciplinaire envisagée est constituée de travaux hydrogéologiques complétés par des travaux hydro-chimiques et géophysiques réalisés en parallèle, permettant d’acquérir des données a priori sur le fonctionnement du système. Suite à cela, un modèle 2D horizontal d’écoulement et de transport en nappe, en régime transitoire est élaboré. La calibration du modèle est réalisée avec l’algorithme de Gauss-Levenberg Marquardt implémenté dans la suite PEST++. Une régularisation de type Tikhonov sur les valeurs a priori de chaque paramètre est effectuée pour contraindre la calibration. Elle a également été accompagnée par une régularisation mathématique (SVD). L’étape de calibration du transport nappe-rivière résolu par l’équation d’advection-dispersion nécessitant des temps de calculs trop importants, un modèle équivalent, rapide, de transport advectif avec le suivi de particules a été développé. A l’issue de cette phase de calibration l’estimation des incertitudes paramétriques et prédictives du modèle est conduite par analyse linéaire et méthode du Null Space Monte Carlo. Enfin, quelques scénarios prédictifs de gestion sont évalués afin de répondre à la contrainte majeure de la réduction des proportions d’eau de rivière dans les ouvrages de production.
... Radon and heat tracers can be used independently to investigate groundwater-surface water interactions (Navarro-Martinez et al., 2017). However, researchers advise that multiple lines of evidence, by combining different techniques, can validate results and overcome limitations of a single method (Dujardin et al., 2014;Gilfedder et al., 2015). The traditional technique to investigate subsurface flow is by hydraulic gradients, which may underestimate subsurface flow compared to other techniques (Mulligan and Charette, 2006;Su et al., 2016). ...
Subsurface flow plays an important role in the functioning of wetlands and in the maintenance of their ecosystem services. Specifically, the transport and exchange of dissolved matter between sediments and surface waters is regulated by subsurface flow, which can strongly affect ecological zonation and productivity. Having a quantitative understanding of this subsurface flow is therefore important. Field techniques based on Darcy’s equation or natural tracers are often used separately to assess flows. Here, radon and heat (both natural groundwater tracers) and Darcy’s equation are used simultaneously to quantify the subsurface flow in a tidal wetland (Kooragang Island, Newcastle, Australia) and the results of the independent methods are compared. A steady-state radon mass balance model indicated an overall net subsurface exfiltration of 10.2 ± 4.2 cm/d while a 1D, vertical fluid heat transport model indicated a net exfiltration of 4.3 ± 2.9 cm/d. Flow estimated from analysis of hydraulic heads indicated an exfiltration rate of 3.2 ± 1.8 cm/d. The difference in flow rates is likely due to the localised measurement of the heat and head methods relative to radon, and therefore, these methods are less likely to capture zones of preferential subsurface flow. The main advantage of radon is that it provides the total subsurface flow regardless of the driving force. While head gradient or heat tracer method have the advantage of temporally quantify infiltration and exfiltration, we highlight that these methods may underestimate subsurface flows in highly dynamic coastal systems, such as tidal wetlands where a large portion of the subsurface flow is recirculated seawater. This could potentially lead to errors in solute flux estimates. This study highlights the importance of employing a multi-tracer approach and has implications towards quantifying the hydrological export of dissolved constituents (e.g., carbon and nitrogen) in coastal wetlands.
... If no flow across the boundaries is considered arbitrary, it may affect the simulated model results leading to incorrect groundwater simulated behavior. Dujardin et al. (2014) also emphasized the use of specified flux in groundwater flow modelling of the improvement in simulation accuracy to a satisfactory level. Therefore, it can be assumed that estimated specified flux was nearly close to the real. ...
Groundwater flow modeling is an important tool for understanding and computing hydrology and water availability of an aquifer zone. However, an accurate representation of boundaries and their initial conditions are vital for simulation of the groundwater flow phenomena. In this study, efforts have been made to develop a GIS based methodology for estimating flux across boundaries of the study area using Darcy flow tool. The spatial maps of topography, bore log, transmissivity, hydraulic conductivity, porosity and groundwater levels for the study area were created in ArcGIS 9.3.1 using krigging method. A buffer zone of 1×1 km 2 cell size was created on inner and outer side of the boundaries and Darcy flow model was used to estimate specified flux across boundaries. The groundwater behavior of the study area was simulated with specified flux boundary condition (Neumann boundary condition) and no flow boundary condition to assess importance and estimation accuracy of estimated flux. Darcy model output indicates that flux across the boundaries contributed about 36.20 mm in the average annual change in groundwater table depth. With estimated specified flux, simulation accuracy of groundwater flow model (R 2) increased to 0.97 from 0.90. The satisfactory level (R 2 =0.97) of simulation accuracy reveals that developed methodology can be used for estimating flux across boundaries in the absence of physical boundaries.
... If no flow across the boundaries is considered arbitrary, it may affect the simulated model results leading to incorrect groundwater simulated behavior. Dujardin et al. (2014) also emphasized the use of specified flux in groundwater flow modelling at regional scale. The estimated flux across boundaries (2690m 3 /day) Therefore, it can be assumed that estimated specified flux was nearly close to the real. ...
Full-text available
Groundwater flow modeling is an important tool for understanding and computing hydrology and water availability of an aquifer zone. However, an accurate representation of boundaries and their initial conditions are vital for simulation of the groundwater flow phenomena. In this study, efforts have been made to develop a GIS based methodology for estimating flux across boundaries of the study area using Darcy flow tool. The spatial maps of topography, bore log, transmissivity, hydraulic conductivity, porosity and groundwater levels for the study area were created in ArcGIS 9.3.1 using krigging method. A buffer zone of 1×1 km2 cell size was created on inner and outer side of the boundaries and Darcy flow model was used to estimate specified flux across boundaries. The groundwater behavior of the study area was simulated with specified flux boundary condition (Neumann boundary condition) and no flow boundary condition to assess importance and estimation accuracy of estimated flux. Darcy model output indicates that flux across the boundaries contributed about 36.20 mm in average annual change in groundwater table depth. With estimated specified flux, simulation accuracy of groundwater flow model (R2) increased to 0.97 from 0.90. The satisfactory level (R2=0.97) of simulation accuracy reveals that developed methodology can be used for estimating flux across boundaries in the absence of physical boundaries.
... Mass flux estimates can be improved substantially, if groundwater fluxes are measured locally and explicitly considered instead of simple calculations of Darcy' law based on the local hydraulic conductivity and gradient (Rein et al. 2009;Milosevic et al. 2012). Smart complementary methods allowing groundwater discharge quantification can be used but the key point in such transient environments is to use of high frequency observations enabling a fine characterization of the transient nature of groundwater flow conditions (Hatch et al. 2006;Kalbus et al. 2006;Ellis et al. 2007;Hyun et al. 2011;Dujardin et al. 2014). ...
Full-text available
Exposure from groundwater contamination to aquatic receptors residing in receiving surface water is dependent upon the rate of contaminated groundwater discharge. Characterization of groundwater fluxes is challenging, especially in coastal environments where tidal fluctuations result in transient groundwater flows towards these receptors. This can also be further complicated by the high spatial heterogeneity of subsurface deposits enhanced by anthropogenic influences such as the mixing of natural sediments and backfill materials, the presence of subsurface built structures such as sheet pile walls or even occurrence of other sources of contaminant discharge. In this study, the Finite Volume Point Dilution Method (FVPDM) was successfully used to characterize highly transient groundwater flows and contaminant mass fluxes within a coastal groundwater flow system influenced by marked tides. FVPDM tests were undertaken continuously for more than 48 hours at 6 groundwater monitoring wells, in order to evaluate groundwater flow dynamics during several tide cycles. Contaminant concentrations were measured simultaneously which allowed calculating contaminant mass fluxes. The study highlighted the importance of the aquifer heterogeneity, with groundwater fluxes ranging from 10‐7 to 10‐3 m s‐1. Groundwater flux monitoring enabled a significant refinement of the conceptual site model, including the fact that inversion of groundwater fluxes was not observed at high tide. Results indicated that contaminant mass fluxes were particularly higher at a specific monitoring well, by more than 3 orders of magnitude, than at other wells of the investigated aquifer. This study provided crucial information for optimizing further field investigations and risk mitigation measures. This article is protected by copyright. All rights reserved.
... River-aquifer exchanges mostly occur as vertical fluxes through the riverbed if the river is flat and shallow. However, in deep and incised rivers, lateral water exchanges with the adjacent aquifer can largely exceed vertical ones [Dujardin et al., 2014]. Under such environment, groundwater head is often the most common parameter used to estimate exchange flux between rivers and aquifers [Barlow et al., 2000;Batlle-Aguilar et al., 2014;Fritz and Arntzen, 2007]. ...
Full-text available
The application of heat as a tracer for assessing river-aquifer exchanges has been mainly limited to vertical flow through the riverbed. Lateral river-aquifer exchanges become more important than vertical riverbed exchanges if the river is deeply incised into an aquifer. Few studies have examined lateral river-aquifer exchanges and the ability of heat to constrain such exchanges. This study aims to perform a robust assessment of the limits of heat as a tracer to quantify lateral river-aquifer exchanges. The study is based on a section of the Meuse River in Belgium, a river predominantly gaining in the studied area and becoming intermittently losing in the winter time. A calibrated transect model shows that river temperature can affect groundwater temperature up to 9 m into the aquifer. An accompanying synthetic modelling investigation using Monte Carlo simulation shows that heat data for distances between 4 and 9 m from the river can reduce the uncertainty of river-aquifer exchanges for conditions similar to those of the transect model. The ability of heat to reduce the river-aquifer exchange uncertainty improves with distance from the river because of the reduction in the number of acceptable model realizations. The optimal distance is 8 m from the river where the groundwater temperature is no longer affected by the river temperature. The synthetic modelling also indicates that heat alone cannot constrain river-aquifer exchanges better than the commonly used hydraulic head. However, when combined with hydraulic head, heat can significantly reduce the uncertainty of river-aquifer lateral exchanges under gaining conditions.
This paper investigated and demonstrated the fundamental role of riparian and terrestrial zone heat balances on the use of thermal gradients to characterize and parameterize groundwater systems. The riparian area aquifer (RAA) includes vegetative cover, comprising a forest of lush Acacia karroo together with some reed, sedges, and forbs. On the other hand, terrestrial area aquifer (TAA) is non-forested. The result illustrated the riparian area suppresses groundwater temperature by an average of 2 °C compared to TAA during summer. These signals were found to uniformly dampen temperature anomalies resulting from lateral groundwater inflow from TAA in summer. The impact of lateral groundwater flow from the TAA was only observed in winter, where the effect of latent heat was cancelled out because plants had undergone dormancy. Consequently, the study concluded that the riparian groundwater temperature is more sensitive to the riparian energy balance. To this effect, riparian vegetative cover can considerably complicate the interpretation of the shallow groundwater thermal regime across hydrologic landscapes with varying surface vegetative architecture. Therefore, analyzing thermals signals may require correcting the “false” groundwater geothermal signals due to vegetative cover. The recommendation is that isolating the contributions of these signals should be independently constrained (e.g., introducing thermal data offset) to strengthen the plausibility of heat in parameterizing groundwater systems.
Full-text available
Stream-groundwater exchange has been investigated in a wide range of hydrologic settings, though very few studies have focused on fine-sediment streambeds. Well-established thermal methods (i.e., analytical and numerical solution of time-series temperature depth-profiles) in combination with Darcy’s and electrical resistivity (ER) evaluations were implemented to improve understanding of processes dominating flow and transport in a low permeability and low-flow coastal stream such as Oso Creek, Texas. The seasonal-trend decomposition using Loess (STL) is tested as a potential means to differentiate between advection and conduction and is validated against groundwater fluxes derived from the other well-established thermal methods. The numerical and analytical solutions indicate groundwater upward discharge was 9 mm d-1 for summer and 3.5 mm d-1 for winter, corresponding to the region's extreme drought conditions. These types of low flow conditions are usually accompanied by hyporheic flow, limiting the vertical flow assumption. While the numerical and analytical methods provide good insight into streambed hydrology for a low-permeability and low-flow stream in a semiarid coastal area, there are limitations associated with the STL method. The analytical and numerical thermal methods employed herein confirm that conduction and diffusion are the dominant processes of heat and solute transfer in fine-sediment streambeds, providing an improved understanding of process-based groundwater-stream interaction and water resources in this type of settings.
Full-text available
Groundwater seepage influences the temperature of streams and rivers by providing a relatively cool input in summer and warm input in winter. Because of this, groundwater seepage can be a determining factor in the provision of suitable water temperatures for aquatic biota. Climate warming affects stream and groundwater temperatures, and changes the thermal characteristics of streams leading to the potential disappearance of habitats. In this study the importance of groundwater for the temperature of two Dutch lowland streams and its possible role in mitigating the effects of climate change was determined by combining field measurements and a modelling experiment. Stream temperature measurements using fibre optic cables (FO-DTS) and sampling of ²²² Rn were done to map localized groundwater inflow. Several springs and seepage ‘hot-spots’ were located which buffered the water temperature in summer and winter. A stream temperature model was constructed and calibrated using the FO-DTS-measurements to quantify the energy fluxes acting on stream water. This way, the contribution to the stream thermal budget of direct solar radiation, air temperature and seepage were separated. The model was then used to simulate the effects of changes in shading, groundwater seepage and climate. Shading was shown to be an important control on summer temperature maxima. Groundwater seepage seemed to buffer the effect of climate warming, potentially making groundwater dominated streams more climate robust. Protecting groundwater resources in a changing climate is important for the survival of aquatic species in groundwater-fed systems, as groundwater seepage both sustains flow and buffers temperature extremes.
Classic estimates of groundwater fluxes are usually based on the application of Darcy's law, which can lead to large imprecisions in transient groundwater flow cases. There is a need for direct, in situ measurement techniques able to monitor time-variable groundwater fluxes. The investigation presented here demonstrates that the Finite Volume Point Dilution Method (FVPDM) is a promising technique for the continuous monitoring of groundwater fluxes. The experimental configuration consisted of monitoring transient groundwater fluxes generated by a multiple step pumping test, which was undertaken in the alluvial aquifer of the River Meuse, Liège (Belgium). Additionally, two FVPDM tests were simultaneously performed in two piezometers screened at two different depths in the alluvial aquifer. Tracer concentration changes during the FVPDM tests were interpreted as the consequences of Darcy flux changes in the alluvial aquifer, which was related to changes in the applied pumping rate. Piezometric levels were also monitored in piezometers located around the pumping well. The pumping test was interpreted using classical analytical solutions, and the FVPDM tests were interpreted using a new mathematical solution, which allows for calculating changes in Darcy fluxes based on the FVPDM tracer concentration evolution during transient groundwater flow conditions. The experiment demonstrated the FVPDM's ability to monitor, as well as be sensitive to changes in transient groundwater fluxes. The FVPDM interpretation also showed contrasting results between the upper part of the aquifer, which is made of loam and sand and slow groundwater flows prevail, and the lower part of the aquifer, which is made of gravels and pebbles and intense groundwater flows prevail.
Full-text available
The interaction between groundwater and surface water along the Tambo and Nicholson Rivers, southeast Australia, was investigated using 222Rn, Cl, differential flow gauging, head gradients, electrical conductivity (EC) and temperature profiling. Head gradients, temperature profiles, Cl concentrations and 222Rn activities all indicate higher groundwater fluxes to the Tambo River in areas of increased topographic variation where the potential to form large groundwater–surface water gradients is greater. Groundwater discharge to the Tambo River calculated by Cl mass balance was significantly lower (1.48 × 104 to 1.41 × 103 m3 day−1) than discharge estimated by 222Rn mass balance (5.35 × 105 to 9.56 × 103 m3 day−1) and differential flow gauging (5.41 × 105 to 6.30 × 103 m3 day−1). While groundwater sampling from the bank of the Tambo River was intended to account for the variability in groundwater chemistry associated with river-bank interaction, the spatial variability under which these interactions occurs remained unaccounted for, limiting the use of Cl as an effective tracer. Groundwater discharge to both the Tambo and Nicholson Rivers was the highest under high flow conditions in the days to weeks following significant rainfall, indicating that the rivers are well connected to a groundwater system that is responsive to rainfall. Groundwater constituted the lowest proportion of river discharge during times of increased rainfall that followed dry periods, while groundwater constituted the highest proportion of river discharge under baseflow conditions (21.4% of the Tambo in April 2010 and 18.9% of the Nicholson in September 2010).
Conference Paper
Full-text available
Brownfield; risk management plan; groundwater contamination; volatile organic hydrocarbons; natural attenuation; risk based
The interactions between groundwater and surface water are complex. To understand these interactions in relation to climate, landform, geology, and biotic factors, a sound hydrogeoecological framework is needed. All these aspects are synthesized and exemplified in this overview. In addition, the mechanisms of interactions between groundwater and surface water (GW–SW) as they affect recharge–discharge processes are comprehensively outlined, and the ecological significance and the human impacts of such interactions are emphasized. Surface-water and groundwater ecosystems are viewed as linked components of a hydrologic continuum leading to related sustainability issues. This overview concludes with a discussion of research needs and challenges facing this evolving field. The biogeochemical processes within the upper few centimeters of sediments beneath nearly all surface-water bodies (hyporheic zone) have a profound effect on the chemistry of the water interchange, and here is where most of the recent research has been focusing. However, to advance conceptual and other modeling of GW–SW systems, a broader perspective of such interactions across and between surface-water bodies is needed, including multidimensional analyses, interface hydraulic characterization and spatial variability, site-to-region regionalization approaches, as well as cross-disciplinary collaborations.
The nature of ground water discharge to a stream has important implications for nearby ground water flow, especially with respect to contaminant transport and well-head protection. Measurements of ground water discharge were accomplished in this study using (1) differences between current meter measurements, (2) stream temperature surveys combined with streamflow estimates, and (3) heat transport modeling of measured temperature gradients below the streambed. The first two techniques produced an area-averaged estimate of ground water flow, while the last produced a point estimate of ground water flux. Point measurements differed from area-averaged methods by 1 or 2 orders of magnitude. We hypothesize that discharge to the study creek is spatially heterogeneous, and is dominated by springs and seeps. Thermal gradient measurement did not quantify these local sources of stream inflow. Point measurements of inflow from temperature gradients or seepage meters, therefore, may not represent ground water inflow in some streams. Stream temperature and streamflow surveys were combined using a simple heat-balance to yield a higher-resolution estimate of streamflow than could have practically been obtained with current meters alone. This approach has potential as a cost-effective method of quantifying ground water discharge in streams where stream inflow is highly heterogeneous.