Upload
doandiep
View
217
Download
3
Embed Size (px)
Citation preview
Report on the effective strategies for integrated geophysical, biogeochemical and hydrologic characterisation
Deliverable 6.3 S.E.A.T.M.van der Zee, T. Ghinda, E. Bloem, M. Wehrer, H.K. French, A. Godio, L. Pedersen, M. Bastani, G. Toscano, B. Biro, A. Søiland
Deliverable 6.3
Soil Contamination: Advanced integrated characterisation and time-lapse Monitoring
Title Report on the effective strategies for integrated geophysical,
biogeochemical and hydrologic characterisation
Author S.E.A.T.M.van der Zee, T. Ghinda, E. Bloem, M. Wehrer, H.K. French,
A. Godio, L. Pedersen, M. Bastani, G. Toscano, B. Biro, A. Søiland
Report No. Deliverable 6.3
ISBN
Organisation name of lead
contractor for this deliverable Bioforsk - The Norwegian Institute for Agricultural and Environmental
Research
No. of pages 71
Due date of deliverable: March 2012
Dissemination level Public
Key words
<Other>
Title of project: Soil Contamination: Advanced integrated characterisation and time
lapse Monitoring (SoilCAM)
Instrument: 6.3 Environmental technologies, Topic ENV.2007.3.1.2.2, Development of
technologies and tools for soil contamination assessment and site
Contract number: 212663
Start date of project: June 2008, Duration: 48 months
Project co-funded by the European Commission within the Seventh Framework Programme (2007-2013)
Disclaimer
The information provided and the opinions given in this publication are not necessarily those of the authors or the
EC. The authors and publisher assume no liability for any loss resulting from the use of this report.
ii
Summary
To effectively combine invasive and non-invasive methods for site characterisation and monitoring
and modelling, the joint interpretation of these methods at the two sites, i.e., Gardermoen and Trecate
has been carried out. The main attention was given to address the limitations with regard to combining
methods successfully and identification of promising options.
The integrated strategy aims at an integrated site-characterisation and monitoring design, with a
variety of methods and strategies. Thus, guidelines are aimed at for ‘full system’ characterisation
integrating geophysical, pedological and biogeochemical data with a 3D numerical transport model for
each contaminated site. Whereas the effort is aimed at, and illustrated for the two SoilCAM sites, it
should be possible to more generally define an integrated strategy for monitoring and predicting future
trends in shallowly contaminated field sites that contain aerobically biodegradable contaminants.
Together with other deliverables, the integrated strategy leads to recommendations for the decision
making stakeholders.
As the illustrative presentation for the two quite different sites reveal, the three main methodologies,
i.e., invasive and non-invasive (geophysical) measuring and the modelling, are needed in the
integrative approach, because each of these methodologies alone suffer from significant shortcomings.
Hence, they serve as ground truth for each other, albeit on different aspects. Therefore, the integration
as reported in this report has resulted in an appreciation of the contamination and its future
development, for each of the sites, that is both more ambitious and better justified.
iii
Content
SUMMARY .............................................................................................................................. II
1. FRAMEWORK FOR EFFECTIVE METHODS INTEGRATION ............................ 1
2. RESEARCH COMPONENTS THAT ARE IMPORTANT FOR THE
INTEGRATION OF STRATEGIES ...................................................................................... 2
2.1 Site specific aspects .......................................................................................................................... 2
2.2 Flow and transport modelling ........................................................................................................ 2 2.2.1 Site specific models .................................................................................................................... 2
2.3 Geophysical assessment of spatial variability of subsoil .............................................................. 5
2.4 Geochemical studies for remediation monitoring ........................................................................ 9
2.5 Water flow and contaminant transport modelling ..................................................................... 13 2.5.1 Water flow modelling, hydrogeological aspects ...................................................................... 13 2.5.2 Contaminant transport modelling ............................................................................................. 16
3. COMBINING NON-INVASIVE GEOPHYSICAL AND INVASIVE METHODS
FOR MONITORING CONTAMINANT AND REMEDIATION PROCESSES ............ 19
3.1 Introduction ................................................................................................................................... 19
3.2 Combining geophysical data ......................................................................................................... 20 3.2.1 ERT and RMT measurements at Trecate site ........................................................................... 20 3.2.2 Joint inversion of cross-hole ERT and GPR data from Moreppen ........................................... 22 3.2.3 Electrical resistivity measurement with Syscal and Ohmmapper equipment at OSL .............. 23 3.2.4 MRS and GPR measurements at Trecate ................................................................................. 24
3.3 Combining geophysical, geochemical and hydrogeological information at Trecate ............... 25
3.4 Combining geophysical, geochemical and hydrogeological information at OSL .................... 30
3.5 Combining geophysical, geochemical and hydrogeological information at Moreppen .......... 31 3.5.1 Electrical resistivity tomography .............................................................................................. 31 3.5.2 Relations between geophysical and soil parameters................................................................. 32 3.5.3 Trench study: Observation of a de-icing chemical breakthrough using electrical resistivity
tomography at Moreppen research station ........................................................................................ 32 3.5.4 Lysimeter study: Observation of a de-icing chemical breakthrough using electrical resistivity
tomography in closed system lysimeters ........................................................................................... 46 3.5.5 Conclusions .............................................................................................................................. 52
3.6 Remediation monitoring by time-lapse geophysics .................................................................... 53
4. INTEGRATION OF NON-INVASIVE AND INVASIVE METHODS FOR
REMEDIATION MONITORING ........................................................................................ 54
4.1 The monitoring-modelling-monitoring loop ............................................................................... 54
iv
4.2 Effective strategies for combination of methods ......................................................................... 56
5. CONCLUDING REMARKS ......................................................................................... 58
6. REFERENCES ................................................................................................................ 60
1
1. Framework for effective methods integration
In the SoilCAM project, two sites (Gardermoen and Trecate) that have similar geology but different
contaminant situation – driven by different weather conditions and histories have been considered with
a range of invasive and non-invasive techniques, and with site specific approaches to modelling. For
an effective, or even optimal, integration of approaches, the original scope in the SoilCAM proposal is
leading.
This implies:
1) Align the results of all work packages at the level of abstraction at which the integrated strategy will
be conducted. Define, within the existing guidelines, a more precise and reliable site-characterisation
and monitoring design, including the implementation of state of the art of biological, chemical and
geophysical methods and combined strategies (i.e. use of different methodologies).
2) Description of guidelines for ‘full system’ characterisation integrating geophysical, pedological and
biogeochemical data with a 3D numerical transport model for each contaminated site.
3) To define an integrated strategy for monitoring and predicting future trends in shallowly
contaminated field sites that contain aerobically biodegradable contaminants
4) To provide a methodology for translating the integrated strategy regarding both type and level of
detail into recommendations for the decision making stakeholders.
2
2. Research components that are important for the integration of strategies
2.1 Site specific aspects
The information that we have received from different sources must be integrated into an interpretation
of the site, that can be tested with respect to its consistency, both internally and with respect to the data
at hand. Several aspects need to be reconciled:
Stratigraphy: different layers and horizons till the basis of the site; an argumentation for
this basis; textural & porosity, physical, and chemical properties of the strata
Environmental conditions: climate/meteorological data (rainfall, snow, temperature
patterns and statistics), vegetation/land use
Hydrology: variation of water fluxes (infiltration, runoff, evapotranspiration, leaching,
capillary up-flow/seepage, groundwater level pattern
Transport: chemical boundaries at top and bottom, important chemical concentration,
direction of fluxes in water and gas phases, to atmosphere, groundwater, in
groundwater.
Contaminant distribution: load/concentration present in each subsurface compartment
(aqueous-, gas-, solid- and possibly nonaqueous phase); spatial extent of the
contaminant distribution in each phase; time- and spatial scale of vertical and horizontal
transport; persistance and attenuation processes
Biogeochemical interactions: equilibrium and non-equilibrium chemical and
transformation processes, plus involved native chemicals.
Additional to these descriptors, also the modelling approaches need to be aligned to each specific site.
For Gardermoen as well as for Trecate, a host of models was used that complement each other, as
detailed in the next section.
2.2 Flow and transport modelling
2.2.1 Site specific models
In view of the quite different types of contamination and the different geohydrologies and raised
questions of end-users, stake holders and involved decision makers, the models used at Gardermoen
and Trecate had to be differently focussed.
Gardermoen: The annually occurring contamination with easily aerobically biodegradable
contaminants is predominantly an unsaturated soil problem, because the contaminants need to be
degraded prior to reaching groundwater level. This poses several main problem areas:
(i) how is water infiltration during the (contaminated) snow and ice melting period in early spring, and
its effects on spatially variable flow in the (say) 4-5 m thick unsaturated zone, towards groundwater.
(ii) how does spatiotemporal flow affect contaminant transport and dispersional mixing, including
uncertainties,
(iii) what does the complex transport process imply for the availability of electron acceptors that are
necessary for aerobic biodegradation.
3
To this end, use was made of the following models:
(1) SWAP: a transient, unsaturated flow model with high root water uptake and atmospheric forcing
functionality (Van Dam et al., 1997). J.C. van Dam, J. Huygen, J.G. Wesseling, R.A. Feddes, P. Kabat, P.E.V.
Walsum, P. Groenendijk, and C.A. Diepen. SWAP version 2.0, theory; simulation of water flow, solute transport and plant
growth in the soil-water-atmosphere-plant environment. Technical Report Technical document 45, report 71, Winand Staring
Centre, Wageningen, 1997
(2) PEARL: this is principally a pesticide leaching model, that is one of the official EU pesticide
screening models used at this time. It has been used to investigate the complexity of spatiotemporal
variability of contaminant leaching. M. Leistra, A.M.A. van der Linden, J.J.T.I. Boesten, A. Tiktak, and F. van der
Berg. PEARL model for pesticide behaviour and emissions in soil-plant systems; descriptions of the processes in FOCUS
PEARL v. 1.1.1. Technical Report Alterra-rapport 013; RIVM report 711401009, Alterra; RIVM, Wageningen; Bilthoven,
2001.
(3) ORCHESTRA: this is a multicomponent transport model, that has been linked to SWAP
specifically for this project SoilCAM, to efficiently combine complex chemistry with transport.
(4) SUTRA-3D-DENS: is one of the most advanced 3D flow and solute models, that accommodates
density dependent flow. It has been adjusted in the context of this project SoilCAM, to enable flow
and transport simulations in spatially variable soil environments.
Trecate:
The problem areas for Trecate, that was contaminated two decades ago with oil-type of contaminants,
are slightly different: the contaminant have already been in place for a significant period of time, its
spreading to the broader environment is mainly dissolved in flowing groundwater, or by volatilization
to the atmosphere, but much evidence points towards very slow losses of contaminants. Hence, the
main questions were:
(i) how is it possible that the light oil fractions that were spilled, do not seem to disperse into the
broader environment, apparently,
(ii) to what extent, is the contaminant being spread to the environment, by passing groundwater, and
possibly leaching soil water.
To address these questions, use was mad of two predominant models, i.e.,
(1) MODFLOW: this is probably the most well-known model for flow of water in 3D through the
shallow and deeper groundwater, and has been linked, also in this project, with models such as MT3D
and such, for solute transport,
(2) STOMP: which is probably the most versatile multiphase flow model available (Wipfler et al.,
2003), to quantify the redistribution of light oil that is present in an oil lens.
4
Figure 1: Conceptual modelling framework
Figure 2: Conceptual model structure for a contaminated site
Site conditions
Description of conditions
Quantities for characterisation
Description of changes in time
Contaminant source
Location, size
Composition, properties
Changes in time
Site processes
Description of processes
Equations
Parameters for
characterisation
Description of changes in time
Transport vectors
Groundwater
Water from infiltration
Air
Moving contaminant /
emulsion
Input
Values of quantities
Values of parameters
Output
Changes in conditions
Effects and changes of processes
Main components of the site model
5
2.3 Geophysical assessment of spatial variability of subsoil
A detailed characterization of the site conditions is very important for providing an adequate basic for
assessment of potential impact and remediation processes. A host of techniques has been employed,
such as ERT and GPR on local areas, RMT was used in areas with ERT measurements, MRS was
applied for larger surfaces. Very localized data were obtained using IP, TEM and TDR. In addition,
invasive methods were used to characterize spatial variability, such as the MCS, for which the
SoilCAM project was its second application area.
Electrical Resistivity Tomography (ERT) was used as a rather central approach to elucidate the
variability of the subsoil, leading to e.g. ERT estimated resistivity distributions in several vertical
profiles at Gardermoen and Trecate and which show that detailed results that can be obtained by
applying this method.
Figure 3: Example of ERT results for resistivity distribution at Gardermoen (partial results from Figure 4.7 of
D1.2: 2D reference model of ERT resistivity data along line 2 measured 11/03/2009.)
Figure 4: Example of ERT results for resistivity distribution at Trecate (results from Figure 3.1 of D1.1)
The Radio Magneto-Telluric (RMT) method provides distributions of resistivity values along vertical
profiles or in a soil volume that is discretized by means of several vertical profiles. The measured
components of the electric field and magnetic field for a number of frequencies are used in an
inversion algorithm for a discretization of the spatial domain [D1.4] and give results about electrical
resistivity. The 3D inversion of measured RMT data includes lateral variations that cannot be taken
into account in 2D inversion, and avoids some distortions [D1.4]. However, the inversion algorithm
still includes some operations that can cause local effects in the results. The RMT results can give a
good image of layers and local structures indicated by higher or lower resistivity values.
6
Such images of the subsurface structures were obtained at Trecate for an area of 100 m x 100 m, and
from top layer down to about 50 m depth.
Figure 5: Example of RMT results for resistivity distribution along a profile in the Trecate site (partial results
from Figure 2.3 of D1.4, that includes results of individual 3D inversion of RMT data.)
A highly conductive clay layer below the surface may screen the RMT signal and limit the penetration
depth, as it was remarked for the Gardermoen site [D1.1].
Ground Penetrating Radar (GPR) gives an insight in the structural differences of soil layers or
volumes consisting of materials with different grain size. Several cross-hole and single hole georadar
investigations were performed in the Trecate site. Snow melt at Moreppen was monitored by means of
time lapse geo-radar data in spring 2009.
This method enables to detect the limits between local layers of different materials. Cross-hole GPR
measurements provide information on electromagnetic wave velocities in soil at different depths,
which are influenced by the rock type, the air content and especially by the water content in various
soil layers. Electrical permittivity and dielectric constant are very high for water in comparison to the
values for usually encountered rock materials, and much lower for air. Therefore, the wave velocities
are much influenced by the water content in soil. Using cross-hole GPR data, the measured wave
propagation velocity can be converted in soil water content on the basis of Topp’s equation or another
relationship between water content and dielectric constant.
7
Figure 6: Example of results of cross-hole GPR in the Gardermoen – Moreppen site (partial results from Figure
4.5 of D1.3: Inversion results of data at borehole pair CD)
The Magnetic Resonance Sounding (MRS) method provides data on the water content, by inversion
of the MRS signal data. These results show also the porosity in the saturated zone. The MRS method
can provide results for aquifers in non-magnetic rocks. For example, the MRS method application in
the Trecate site (November 2010) provided estimations of the water content from topsoil down to
about 50 m depth. This survey included 4 MRS soundings, allowing to perform an inversion along a
200 m long line. An image of water contents in a vertical profile of 100 m length and 50 m depth
shows rather large longitudinal and vertical differences [D1.3]. The MRS method gives an image on
the heterogeneity of an aquifer, contributing to the understanding of its behaviour and effects in case
of a contaminant plume.
8
Figure 7: Example of MRS results in the Trecate site (partial results from Figure 6.7 of D1.3)
The various geophysical methods cover the field situation at different scales and can provide, as
summarized above, information about: main structures, layers, local volumes with different soil
properties, water content, porosity. Considerable preparatory work may be needed for some methods,
such as cross-hole measurements, or require large numbers of measurements.
9
2.4 Geochemical studies for remediation monitoring
Invasive analyses of groundwater samples and soil samples provide the ground truth regarding on site
conditions and remediation processes. Their limitation is that these data are generally point
measurements, that are in need of additional knowledge, to be translated to the larger, field scale.
Figure 8: Example of contaminant concentrations represented on the basis of point data from the Trecate site
[Bolognino&Godio 2008]
The example of Figure 8 is a nice indication of how the spatial pattern of contamination, and in
particular the isoconcentration lines that are inferred from a few data points are predominantly an
interpretation. Indeed, research of the past decades has established that the uncertainty is large in this
kind of representations.
Therefore, also interpretations based on point measurements to interpret bioremediation and other
chemical changes are limited by the small support basis, and profound uncertainties in between
measurement points. In principle, such limitations are also appropriate for geophysical observations,
although the spatial support in those cases is much better.
10
Figure 9: Example of dissolved organic carbon concentrations showed by analyses of ground water samples
from several depths in a point of the Trecate site (data obtained by FSU).
How to integrate holistically techniques on organically contaminated sites – the hydrogeochemical
point of view
The required level of knowledge, expenses and combination of methods is likely to be very different
for attaining the objectives of the above mentioned stages (see 2.2) [Wehrer 2012].
1. Stage of aiming for immediate action: the thread is obvious and needs to be treated immediately.
Quick answers are required on concentrations, loads and location of hazardous substances as well as
the anticipated short term distribution in the environment. Process knowledge about the long term fate
is of minor importance.
Decisions are most likely based on organo-leptic analysis, on-site analytical methods and a
limited number of lab samples for chemical analytical investigations. Pre-existing site
knowledge from maps, prior investigations and permanent installations like wells needs to
be included.
Most chemical analytical methods for aqueous samples are fast. Yet, trace analysis for
hydrophobic organic contaminants (aqueous and solid phase samples) have already a time
frame of several days. Fast surrogate methods are IR-spectroscopy or total organic carbon
content. The latter is a good surrogate for TPH in organic poor environments.
Fast microbiological screening methods might play a role at this stage, if either the
contamination is of microbiological origin or in the form of ecotoxicological tests.
Sophisticated approaches of subsurface investigation may include direct push techniques
with online hydrocarbon analysis with IR and electrical resistivity for the detection of a
-14.00
-12.00
-10.00
-8.00
-6.00
-4.00
-2.00
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
C [mg/l]
z [
m]
11
plume. These are in particular helpful with a limited number of organic trace analyses in
the lab.
Experimental long term approaches, considering flow, transport and biochemical processes
and geophysical methods requiring time intensive data processing will most likely not be
applied.
2. Stage of aiming for an investigation strategy: the site is poorly characterized and little background
information is available. The first objective approaching such a site could be to decide, which type,
quantity and spatial distribution of measurements are of benefit. In addition to information gained in
stage 1), non-invasive methods should add information to the existing knowledge.
Non-invasive methods with a high sensitivity for stratigraphical units (georadar, electrical
resistivity) should help to define the borders of different sets of stratigraphical subunits
and/or potential contaminant plume locations, which would provide a basis for the layout
of a detailed field sampling campaign
Preliminary numerical case studies based largely on assumptions can help to develop
hypotheses on the contaminant fate, define future points of interest and required parameters
to be measured in hydrogeochemical- and hydrogeological field campaigns
The knowledge base of stage 1) should be broadened based on the same techniques or a
revised set of techniques
3. Stage of aiming for a monitoring strategy: the basic characterization is done but dimensions and
development of the pollution and processes in the polluted area are not well enough understood to be
able to suppose a remediation strategy. Building on the information form stages 1) and 2), additional
investigations are required aiming for a more detailed characterization of the plume and the processes
determining the fate of contaminants. Unsaturated zone and saturated zone investigations require
different approaches for hydrogeochemical investigations.
In the saturated zone multilevel piezometers provide a great level of detail for the depth
resolution of the contaminant plume. This great level of detail aids to interpret cross-
borehole geophysical data and provides information on the level of heterogeneity. Areal
investigations with multilevel piezometers are likely to be incapable due to the large
amount of samples per spot. Yet, geoelectrical and electromagnetic methods in turn can be
carried out with less effort in standard wells. This way, an areal and depth resolved
characterization of the contaminant plume can be received by combining both approaches.
Characterization of the unsaturated zone is most commonly connected with more
difficulties than the saturated zone. The sampling of soil solutions may prove impossible in
deep unsaturated zones. Geophysical methods, probably assisted by coring analysis, may
be the only way to get hold on the processes at such sites. Yet, due to the presence of a gas
phase and the time variability of phase properties and extents, geophysical methods
incorporate a great deal of ambiguity. Thus, areal investigations in the unsaturated zone
must rely on fuzzy approaches momentarily.
In the surface near unsaturated zone, field lysimeters and suction devices of various kinds are powerful
methods to identify and quantify processes, though such investigations may require more effort than
groundwater sampling in wells. Yet, such investigations can only provide spot measurements. The
lysimeter scale possesses already a considerable heterogeneity due to variability of material properties
and preferential flow phenomena. Thus, equipping such devices with geophysical probes appears as a
promising tool to reveal the effect of such characteristics inside the lysimeter.
12
Hydrochemical data provide the signature of microbial processes in the subsurface. Due to
interpretation ambiguities microbial investigations are required. Many highly contaminated
subsurface sites are at least in parts anaerobic, which makes authentic sampling of
microbial communities a challenging task. Thus, culture independent methods should be
favoured.
Numerical modeling of organically contaminated sites is particularly challenging due to the
large amount of processes and parameters acting on different scales. The hydrochemical
data can serve as validation for numerical models of the site to be able to develop realistic
scenarios of the interplay of hydrological, geochemical and biological processes. Yet, it
must be assumed that a crowd of valid models is able to explain the experimental data
equally well. Therefore, particular care has to be taken that worst- and best-case scenarios
embrace the valid site models.
4. Stage of aiming for a remediation strategy: processes, dimensions and development are assumed to
be well known and decision on the further treatment of the site is required. Building on information
from previous stages, a selection of remediation strategies can be tested in the lab and in the field
Testing remediation strategies generally involve various stages. In a first stage, batch
techniques can be recommended for a rapid screening of a range of potential strategies.
Subsequently, laboratory column experiments offer the opportunity to test a selection of
strategies under more natural flow conditions, while a high level of control is still exerted.
In pilot scale tests in the field obviously the most natural boundary conditions control the
situation. The level of natural complexity is high due to inherent heterogeneity and lack of
control of the boundary conditions. In this context, geophysical tools provide valuable
support to interpret the response of the system to the remedy. Provided sufficient
hydrogeochemical data is collected, the geophysical tools applied in the pilot study can be
calibrated to monitor the areal remediation effect on a larger scale. This again will most
likely result in soft data and requires additional bio-geochemical information collected with
lysimeters and multilevel-piezometers.
13
2.5 Water flow and contaminant transport modelling
The models of the water flow and contaminant transport at the Gardermoen site integrates information
from geophysical and and ground trothing data and provided good results in comparison with the
observations in the field (for more details see D3.2 and D3.3).
2.5.1 Water flow modelling, hydrogeological aspects
Water flow modelling combines the invasively and non-invasively obtained data in a systematic and
consistent way. Since water flow is generally the motor of contaminant transport, either because
contaminant is dissolved in water that is flowing, or because contaminant volatilizes into the gas phase
and is pushed by the flowing water, geohydrological modelling is a crucial step in the interpretation of
site dynamics. In the case of Trecate, water flow was also recognized to play an important role in the
alternating stages of capillary entrapment and release of light oil phases at the capillary fringe. This
entrapment is of great importance as it is responsible for the persistence of the oil contamination at the
Trecate site. Since the groundwater fluctuations are significant, a relatively large oil volume can be
trapped in this way, to be removed from the site only through slow volatilization fluxes, and by slow
and incomplete dissolution in bypassing water.
Whereas the invasive measurements provided information of an oil enrichment zone in the vicinity of
the Trecate site, the geophysical methods enabled the assessment of this zone in its aerial extent. This
information is crucial for the contaminant transport modelling.
At the Gardermoen site, the geophysical methods were predominantly focussed on the determination
of the variability of the subsoil hydraulic properties and the assessment of flow as a function of time.
Combining the geophysically sensed patterns with modelling required that the geophysical output be
translated first into soil hydraulic properties and their variation. This is where one of the main
challenges lie. The gain that is possible is, that not only the variability (in terms of variances,
contrasts, autocorrelation scale) can be quantified statistically, but also the actual realization of the
subsoil can be determined: which parameter values are found where? Such knowledge surpasses that
of the description given by French and Van der Zee (2012).
In Deliverable D3.3, simulations of flow and transport in a two dimensional vertical profile of the
unsaturated zone at Oslo airport using layering information from the georadar measurements are
shown, one example is shown in Figure 10 and Figure 11 below.
Figure 10: Changes in saturation levels simulated with SUTRA for a profile based on layering structure
interpreted from GPR measurements.
Saturation
14
Figure 11: Changes in electrical resistivity in the same vertical profile as that of the SUTRA simulations (Fig.
10). Blue and purple show more conductive areas, red yellow and green show areas becoming less conductive
areas.
The actual realization is of importance, because it enables time-lapse assessments of changes, e.g. with
regard to contaminants and their biodegradation products, and because they enable the determination
of optimal positions for monitoring processes and changes. Hence, a major factor is to what degree the
geophysical observations can be translated into soil properties, which are compatible with model
input.
Pedotransfer functions
Modelling of unsaturated zone flow and transport requires a description of the hydraulic pressure-
water saturation relationship (=water retention characteristic, WRC) and the unsaturated conductivity
function of the textural subunits. This is frequently done using parametric functions like the van-
Genuchten-Mualem equations (Van Genuchten, 1980). The parameters of these functions can be
received by fitting to an experimentally derived WRC.
The experimental approach requires undisturbed soil samples in core cutting rings and time consuming
laboratory analysis. As undisturbed samples are difficult to obtain, in particular in depths >1m, one has
often to rely on pedotransfer functions, which predict the parameters of a water retention function on
the basis of more straightforwardly derived soil parameters (Schaap et al., 1998, 2001; Vereecken et
al., 2010). A review on the use of pedotranfer functions was provided by (Vereecken et al., 2010).
Mostly, these functions provide a relationship between textural properties of the soil and van
Genuchten-Mualem parameters. Textural properties are more straightforward to derive than a WRC
and thus, this procedure offers the possibility to estimate the WRC for samples where no such
measurements are available. An alternative to experimentally derived WRC and pedotransfer function
is the simultaneous observation of pressure and water content in the field (Gribb et al., 2009).
Also the saturated hydraulic conductivity Ks is an important parameter for modelling of flow and
transport. Like the WRC, it can be determined experimentally in the laboratory on undisturbed core
cutting rings. Also in this case, it is easier to derive the saturated hydraulic conductivity using
pedotransfer functions and regressions, which are based on textural properties of soil samples (Beyer,
1964; Kozeny, 1927; Vereecken et al., 2010).
The use of pedotransfer functions should be carried out with care. The problem of pedotransfer
functions is that they often do not deliver reliable estimates (Stumpp et al., 2009) and should be
evaluated for their applicability on a specific field site. There are several reasons for this. Firstly, most
pedotransfer functions predict van Genuchten parameters. Yet, this is a unimodal function and can not
represent dual porosity systems. Soils with a considerable contribution of a second porosity system
can be modelled using a bimodal function (Durner, 1994). Also, non-capillary flow in large pores can
not be modelled using the van-Genuchten function (Mohanty et al., 1997). Currently, the improvement
15
of pedotransfer-functions for the prediction of water flow characteristics near saturation range is
hampered by lack of sufficient data (Vereecken et al., 2010).
Secondly, the hydraulic characteristics of soils are not only determined by texture and bulk density,
which are mostly used as predictors for pedotransfer-functions (Vereecken et al., 2010). Aggregation
and formation of structures influence the pore structure of soils. These originate from weathering,
microbial- and plant activity as well as human activities. Important agents of aggregation are, for
example, iron- and manganese oxides, carbonates and organic carbon contents. Also other caking
agents may influence the role of texture for the hydraulic properties. Consequently, pedotransfer-
functions are likely to be more successful, if these drivers are considered in the calibration. Yet, while
organic matter has already been included into pedotransfer-functions, there are no general objective
descriptors of soil structure. This prevented the inclusion of this important information into
pedotransfer-functions so far (Vereecken et al., 2010).
So far, for practical applications it is recommended to improve estimates of standard pedotransfer-
functions by establishing site-specific relationships between measured water retention characteristics
and the estimates of the standard pedotransfer-functions for each required parameter. Such site specific
models help to improve the results of pedotransfer estimates, like shown in SOILCAM deliverable
D3.2.
In D3.2, the pedotransfer function ROSETTA (Schaap et al., 2001) (Schaap et al., 2001) and the Ks-
relationships by Hazen/Beyer (Beyer, 1964; Hazen, 1892) and Kozeny/Carman (Carman, 1938;
Kozeny, 1927) were evaluated for the Gardermoen airfield soil. It was found that there is considerable
deviation between measured water retention characteristics and ROSETTA estimates. These estimates
could be improved by site specific models. The estimates for the saturated hydraulic conductivity
using the Kozeny/Carman approximation proved to be quite reliable. Yet, the amount of samples
investigated in both procedures was only very limited and should be confirmed by further
experimental investigations.
Infiltration
At the Gardermoen site, water infiltration is the driving force for contaminant transport towards the
primary threatened target: groundwater, particularly the phreatic aquifer. At the Trecate site, water
infiltration is important in view of its entrapment effect on the contaminant at the phreatic level and
because it is a major transport factor of oil to the broader environment. For these reasons, it is clear
that infiltration is of primary interest in this research. This is the more so, if the infiltration rate is not
well known. Whereas in Trecate, the assumptions are reasonably valid that uniform atmospheric
inputs and irrigation rates are established for each field, this is not so in Gardermoen. For Gardermoen,
the main infiltration processes occur in early spring, the snow and ice melt period. Particularly in this
period, the melting is highly variable, and the strongly spatiotemporally varying fluxes through the
snow pack and partly frozen topsoil is difficult to monitor, because of the highly preferential nature of
flow.
Water infiltration can be determined using geophysical methods. In this research, such an investigation
has been carried out for the Gardermoen site (at the experimental site Moreppen) for characterisation
of soil conditions that influence water infiltration. The results of GPR survey from October 2008 were
analyzed by comparing them with stratigraphic data and reported in deliverable [D1.2]. The equipped
trench and also lysimeters for point measurements in the field were used for studying the natural
process and for experiments regarding water infiltration and contaminant transport in field conditions.
A flow model was developed for the unsaturated zone and the saturated zone in the Gardermoen site
taking into acount the specific conditions during the winter and snow melting and reported in
deliverables [D3.3] and [D3.2].
For the Trecate site, taking into consideration the observations, several layers are included in
calculations. Different sources of data show various ranges of values for hydraulic conductivity. An
overall estimation specified a high value of hydraulic conductivity in the range 0.7 – 1.2 * 10-2
m/s
16
[Bolognino&Godio 2008]. Some measured hydraulic conductivities show more permeable sub-layers
at 10 – 11 m depth and 13 m depth [D4.4]. Less permeable sub-layers are at 9 m, 12 m and 15 m
depth. For less deep layers, vertical hydraulic conductivity was estimated to be in the range 0.5 – 3.4
m/d, in alternating layers with different permeability. Average horizontal hydraulic conductivity was
estimated in a range of values around 56.5 m/d [Cassiani et al 2004].
However, hydraulic conductivity values are constrained by compatibility with basic field observations
regarding the time variation of groundwater levels in several points during many years. Integrating the
groundwater flow equation over the thickness of a sub-layer with an irregular distribution of hydraulic
conductivity values (as may be partially shown by measurements) leads to a term that includes a sum
written as an integral or in a discretized form. Several studies addressed the soil heterogeneity problem
for areas near the Ticino river which flows at a small distance from the Trecate site [Felletti et al 2006]
[Zappa et al 2006].
Instead of using only directly measured values, additional values had to be included for successful
groundwater flow simulation. The combined hydraulic conductivities have to be realistic
representations of the soil properties so that to obtain time series of simulated groundwater levels in
agreement with the observed levels. According to the weighted sum, a small hydraulic conductivity
over most part of the sub-layer thickness can be hydraulically dominated by a large hydraulic
conductivity over a smaller fraction of the sub-layer thickness, in agreement with theoretical analyses
in the literature [e.g. Dagan, 1987]. The numerical simulation results are in agreement with the
observed time series of groundwater levels in many points of the studied site when certain average
values are used for the hydraulic properties.
Two dimensional simulations in report D3.3 serve to illustrate the use of geophysical data for the
characterisation of the geological structures in numerical models for water and solute transport. These
simulations indicate that contaminant transport in the natural soils next to the runway is gravity
dominated, i.e. mainly vertical and a function of infiltration. This is consistent with the assumption
made for the complex geochemical modelling of propylene glycol in D3.3. The simulations also
indicate that the layering structure has larger influence on the vertical spreading of the plume than on
the vertical centre of mass, which is fairly similar for hydraulic permeability fields in the same order
of magnitude.
2.5.2 Contaminant transport modelling
Contaminant transport modelling can have various aims. For the present research, the main aim of the
transport modelling is to assess to what degree the combination of invasive and non-invasive
measurements in transport modelling leads to a better result. A better result is usually that the
observed phenomena agree well with the modelled phenomena. For this assessment, many different
optimization strategies, both formal and informal, have been developed.
Integration of bio-geochemical and hydrogeological results with contaminant reactive transport
modelling leads to the need for adjustments of flow velocities and degradation rates to obtain
concentrations in monitored locations and plume dimensions in agreement with observations.
The experimental data of this project are considerable, but they have their limitations. These
limitations are inevitable, because the project duration, as well as the at best approximate nature of our
prior understanding of the subsurface (based predominantly on invasive, point scale observations)
leads to an investigation that is a learning experience. A clear example is the research using lysimeters
that is aimed to provide data with reduced complexity (in terms of initial and boundary conditions) to
provide an intermediate to local quantification of soil properties, and a site-scale overall view of all
complexity. The results of this research [D3.3 among others], still appears to have a profound
complexity. Another clear example is the parameterization of the biodegradation biogeochemistry and
kinetics. The biodegradation focussed subprojects provided both a wealth of data and a clear
illustration of the significant site specificness of the parameterization. For this reason, the
17
parameterization as such of the biodegradation model differs between the Trecate and the Gardermoen
site, and both are hampered by their uncertainty bounds.
Unfortunately, at this moment, the project consortium is still working on the reconciliation of the
different local, lysimeter, and non-invasive measurements, towards an integrated modelling exercise,
and for this reason, this final assessment is still imperfect.
For an illustration of the regional modelling, the Trecate site is considered. For this site, the invasive
measurements and monitoring have provided for the physical (soil hydraulic) properties such as the
porosity, hydraulic conductivity, and retention properties for water in the vadose zone. Whereas the
geophysical methods resulted in areal impressions of how the subsurface looks, such data were still
limited in their areal extent and therefore had to be interpreted in a statistical sense. Although this may
be regarded as a severe limitation, considering that the project aimed at quantifying ‘the actual
realization’ rather than the statistical properties, for this purpose this is not a major concern. The
compelling reason for this assertion is that the Trecate areal extent of the ‘initial’ contamination is
quite large and, in fact, much larger than the autocorrelation scales that are important for the flow and
transport behaviour through the contaminant ‘hotspot’. In statistical jargon, the Trecate contamination
site is ergodic. For this reason, local variations of properties affect the local dissolution, transport, and
biodegradation processes, but for the entire site, these variations are of secondary importance. A clear
example is given in Figure 12. This figure, illustrating the ‘plumes’ of several compounds that interact
with each other, and as the different panels reveal, the variability is not such, that the overall pattern is
difficult to convey to managers! Instead, we observe almost circular chemical iso-concentration
contours, that are clearly related (using information of the deliverables of WP1) to the Trecate site
size.
Contaminant concentrations (detailed modelling of degradation reactions)
Concentrations of generated ions (manganese)
18
Sulfate concentrations
Figure 12: Two-dimensional areal modelling of the contaminant transport, subject to the presence of electron
acceptors as Mn, and of sulfate, at Trecate.
The implication of Figure 12 is that the various sources of information have resulted in a rather simple
interpretation (almost circular contaminated site, releasing contaminant in bypassing water and subject
to biodegradation kinetics), that may even be compatible to analytical treatment.
19
3. Combining non-invasive geophysical and invasive methods for monitoring contaminant and remediation processes
3.1 Introduction
Over the recent years the use of geophysical techniques has become more widespread in order to
monitor hydrogeological (hydrogeophysics) and biological processes (biogeophysics) at the field
scale. These techniques can provide information about the physical properties of larger volumes of the
subsurface than traditional soil/soil-water sampling techniques and are cost-efficient.
Geophysical methods such as ground penetration radar (GPR), electrical resistivity (ER and ERT), are
well established methods for characterising subsurface properties and spatial structure (e.g. Hubbard
and Rubin, 2000). Electrical and electromagnetic methods are widely applied for soil mapping and
detecting of contaminated plume. Electrical resistivity (ER) permits a preliminary assessment of the
hydrogeological make-up of the areas. Low frequency electromagnetic (EM) methods are usually
adopted for fast mapping and preliminary assessment of the aerial extent of the potentially
contaminated land. Time-lapse measurements have become a common method to characterize changes
in water saturation and solute transport in the unsaturated zone (French and Binley, 2004; French et al.
2002). The non-uniqueness of the interpretation techniques (inversion) can be reduced by constraining
the inversion through the addition of independent measurements along the same profile. Such
measurements include soil physical properties, soil suction, contaminant concentration and
temperatures.
In previous deliverables we have shown the results of different techniques applied at our study sites
(Gardermoen, Moreppen, and Trecate), such as ERT and GPR on local areas, RMT was used in areas
together with ERT measurements, MRS was applied for larger surfaces. Very localized data were
obtained using IP, TEM and TDR. In addition, invasive methods were used to characterize spatial
variability, such as the MCS, for which the SoilCAM project was its second application area.
20
3.2 Combining geophysical data
In order to reduce the non-uniqueness of the data interpretation several combined analyses have been
carried out:
Joint inversion of ERT and RMT measurements at Trecate
Joint inversion of ERT and GPR measurements at Moreppen
Electrical resistivity measurements with Syscal and Ohmmapper equipment at Gardermoen
MRS and GPR measurements at Trecate
3.2.1 ERT and RMT measurements at Trecate site
The ERT data measured along a profile of about 200 m in the Trecate site were used for obtaining
estimated electrical resistivity down to almost 40 m depth [D1.1]. Electrical resistivity down to about
50 m was also estimated on the basis of RMT measured data along the same line, by 1D inversion.
This profile is located from south to north of the borehole B-I. The electrical resistivity results indicate
three main structures [D1.1] below the topsoil:
a resistive layer below the topsoil, indicating coarser material and/or low water saturation
a more conductive layer below about 12 – 14 m depth, possibly a layer of finer texture
a resistive structure at about 30 m depth, indicates a deeper layer of coarser material since this
part is definitely in the saturated zone, possibly a deeper aquifer system, though not sufficient
coverage of area with the RMT method to be certain of this.
There are some almost similar structures especially in the top and middle of the figures, indicating the
unsaturated zone and the saturated zone. Differences can be seen in the lower part of the figures,
where the RMT results cover a wider and deeper area.
21
Figure 13: Comparative results of ERT and RMT (1D inverted) methods along a S-N line in the Trecate site
(results from Figure 3.29 of D1.1). Notice different in colour scale, ERT profile shows low resistivity in blue ,
RMT shows low resistivity in red. Similar horizontal scale.
The possibility of enhancing the use of different geophysical primary data can be analysed by
performing joint inversion. This option was studied for the Trecate site [D1.4]. The ERT and RMT
data were measured along the same profiles in an area of the Trecate site, but with different intervals
along the profiles. The inversion was carried out using a non-uniform vertical discretization. A
smoothness-constrained scheme was used for 2D joint inversion [D1.4]. Weights of ERT and RMT
data and smoothing during the inversion were specified.
22
Figure 14: Example of 2D joint inversion results of ERT and RMT data along a vertical profile in the Trecate
site (partial results from Figure 2.3 of D1.4)
This joint inversion method was evaluated by means of comparison with 3D individual RMT inversion
results, and the conclusions were favourable [D1.4].
3.2.2 Joint inversion of cross-hole ERT and GPR data from Moreppen
Additional constraints may be applied in order to obtain results that match a priori information [D1.4].
Joint inversion with cross-gradient constraints was performed for ERT and GPR cross-hole data from
Moreppen, taking into consideration soil resistivities and radar velocities [D1.4]. The results indicated
well the vadose zone above about 3 m depth, the saturated zone below about 4 m depth, and the
capillary fring between 3 and 4 m depth.
23
Figure 15: Example of results of joint inversion with cross-gradient constraints, for the Gardermoen –
Moreppen site (partial results from Figure 3.7 of D1.4), that indicates the phreatic aquifer (blue) and the
overlying vadose zone.
3.2.3 Electrical resistivity measurement with Syscal and Ohmmapper equipment at OSL
The electrical resistivity results obtained on the basis of the measurements with two different
equipments were compared for the Gardermoen site [D1.1]. The two types of equipment, Syscal and
Ohmmapper, work differently. The distributions of electrical resistivity obtained by inversion were
compared by means of graphical representations with the same scale of values.
Figure 16: Example of Syscal and Ohmmapper results in the Gardermoen site (results from Figure 3.43 of D1.1)
Even if they were obtained using different measurements, the results are in comparable ranges of
values and both indicate some of the particular areas.
24
3.2.4 MRS and GPR measurements at Trecate
Main layers
The MRS method was employed at the Trecate site [D1.3]. The results below the depth of about 13 m
show two main aquifers and a layer with lower water content between them. The first saturated layer is
found between about 13 m and 28 m, with average water content of about 20 % in the area of MRS
measurements and high permeability. The layer below the first aquifer has laterally variable lower
values of water content. The second aquifer is found below 45-50 m depth, its average water content is
higher, about 25% , and it has lower permeability. The estimated characteristics of these saturated
layers are introduced in the flow model and are compatible with the information from stratigraphic
data below the top saturated zone.
Figure 17: Water content in a vertical profile, from MRS measurements (results from Figure 6.8 of D1.3)
Figure 18: Water content at different depths estimated by MRS measurements (results from Figure 6.7 of D1.3,
blue colour for higher permeability, and red colours for lower permeability)
With cross-hole GPR measurements in the Trecate site (performed by POLITO and Uppsala on
19.10.2010) a separate layer below 11 – 12 m depth was observed (see Figure 19).
25
Figure 19: Cross-hole GPR measurements in the Trecate site 19.10.2010
3.3 Combining geophysical, geochemical and hydrogeological information at Trecate
Local structures
Local results of geophysical, geochemical and hydrogeological investigations point out small scale
heterogeneities or local structures that may distort the average flow and contaminant transport over
medium distances in the site. The resistivity images of some geophysical methods can give an
impression of a section at the site, with such local structures, although due to measurement processing,
the impression has significant uncertainty bands. The geophysical information can be interpreted more
adequately by combining such graphical results with data from other research methods applied in the
site. As an example, iso-surfaces of low resistivity values (Figure 20) as identified by RMT at the
Trecate site are shown, that reveal such local structures.
Measurements of the hydraulic conductivity beneath the ground water table are shown in Figure 21.
These values can be considered to be large and this was expected because groundwater levels at
different locations at the Trecate site were closely correlated (Figure 22). In other words, the very
coarse textured soil layers, with a high contents of coarse grains of 2 – 6.3 mm (Figure 23) act as
large-permeability layers, that prevent significant gradients in the water levels in the horizontal
directions.
Such highly permeable layers can be detected as the low resistivity layers bounded by the iso-
resistivity surfaces of 50 and 100 Ωm of the RMT method (Figure 20). They are characterized by high
water content and a high permeability, due to the high contents of coarse sand
26
The implication is that with non-invasive techniques (here RMT), layers can be detected that are
hydrologically very meaningful (preventing significant horizontal head gradients; transmission of
large volumes of water horizontally in a short time) if they are continuous/not interrupted, and that
these techniques also indicate whether or not such layers are continuous. For hydrological and
contaminant modeling, this potential of geophysical methods is very advantageous.
Figure 20: Low resistivity sub-layers identified by RMT method application in the Trecate site (Partial results
from Figure 2.9 of D1.4. Iso-resistivity surface for 50 and 100 Ωm to image the morphology of the saturated
layer/zone in 3D using 3D RMT resistivity model.)
27
Figure 21: Sub-layer large values of hydraulic conductivity (values from POLITO [D4.4])
Figure 22: Close correlation between series of ground water depth values measured in the points P19 and P16
of the site during one year (data from POLITO), also obtained by modelling
-16
-14
-12
-10
-8
-6
-4
-2
0
0 0.0005 0.001 0.0015 0.002 0.0025
K [m/s]
z [
m]
5
6
7
8
9
10
11
12
13
5 6 7 8 9 10 11 12 13 14
P16
P1
9
28
Figure 23: Content of grain fraction of 2 – 6.3 mm (values measured by FSU)
Contamination
The contamination of the subsoil influences some porous medium properties that may have a local
effect on ground water flow, but that also make it possible that such a contaminated volume is detected
geophysically. Since the NAPL has higher resistivity than water, the smear zone at the phreatic water
level has a measurable higher resistivity (Figure 24 and Figure 25) than the water saturated layer
below. Its resistivity is even higher when the ground water table is high and the contaminant is
entrapped capillary.
Soil core analyses carried out by FSU show that water content in this smear zone under the ground
water table in the contaminated area (Figure 26) is considerably smaller than the measured porosity
values of about 0.35 – 0.40. Apparently, the higher resistivity smear zone does not have the entire
porosity available for storing water, as the light oil contaminant occupies part of the pores. From the
integrated modelling [D3.3], we know that this is indeed the case and that the contaminant is trapped
when the groundwater level is higher than the smear zone.
Because the contaminant leads to a higher resistivity, the use of only the non-invasive methods might
have resulted in the erroneous conclusion, that the phreatic water level is located at depths of 10-12 m.
Due to the soil core analysis, it is clear that the contaminant is responsible for this effect.
-14
-12
-10
-8
-6
-4
-2
0
0 10 20 30 40 50 60 70 80 90 100
Contents [%]
z [
m] S2
BS 7
BS 5
29
Figure 24: Resistivity values in the smear zone, individual 3D inversion of RMT data and joint 2D inversion of
ERT and RMT at Trecate site (Partial results from Figure 2.3 of D1.4.)
Figure 25: Resistivity values in the smear zone a) crosshole ERT, b) cross GPR data (results from Figure 3.17 of
D1.4)
30
Figure 26: Water content under the ground water table (depths over 7 - 8 m) in the contaminated area
(measurements from FSU)
At the Gardermoen site, the de-icing chemicals do not form a separate liquid phase (as oil does in
Trecate), and the impact of dissolved de-icing chemical on the resistivity is much smaller. Therefore,
detecting these contaminants non-invasively is much more difficult, and the obtained data
predominantly refer to texture and the related soil hydraulic properties, such as porosity and hydraulic
conductivity [D1.2]. The range of resistivity values for some grain size distributions is large even in
the unaffected area [D1.2] and therefore derivation of soil physical properties from resistivity is not
straightforward.
3.4 Combining geophysical, geochemical and hydrogeological information at OSL
Main layers
Whereas the invasive approaches enable the assessment of soil properties and their statistical
description (mean, variance, autocorrelation structure), the benefit of combining this with geophysical
methods is that also the layering can be determined. The stratigraphy of the Gardermoen site as shown
in Figure 37 is the result of the GPR measurements (another feasible technique would be MRS).
Indicated in this figure is also the location of three soil cores obtained invasively. With the measured
grain size distributions of the soil cores, the layers are characterized with regard to their contents of
clay, silt, sand, gravel at different depths, and of the differences that may be expected in a soil
hydraulic sense within and between the different layers.
-14.00
-12.00
-10.00
-8.00
-6.00
-4.00
-2.00
0.00
0 10 20 30 40 50 60 70 80 90 100
Water content [%]
z [
m]
BS 7
BS 5
31
Figure 27: GPR results in the Gardermoen site, compared with stratigraphic data (results from Figure 4.5 of
D1.2)
3.5 Combining geophysical, geochemical and hydrogeological information at Moreppen
3.5.1 Electrical resistivity tomography
The resistivity of geological materials exhibits one of the largest ranges of all physical properties. In
sedimentary rocks, the resistivity of the interstitial fluid is more important than that of the rock. This is
due to conduction in rocks occurs by pore fluids acting as electrolytes with the actual mineral grains
contributing very little to the overall EC of the rock. Resistivity is influenced by factors such as soil
type, porosity, connectivity of the pores and their tortuosity, the saturation level and temperature
(Reynolds 1997). The three phases present in soil are air, fluid and solids and they affect the resistivity
differently: air is an insulator, the water solution resistivity is a function of the ionic concentration and
the resistivity of the solid grains is related to the electrical charges density at the surface of the
constituents. The geometry of the pores (void distribution and form) determines the proportion of air
and fluid present in the sediment. Clay particles conduct electricity not only through free pore-water
but also through adsorbed water at the surface of the clay particles; hence, the resistivity of the solid
matrix cannot be neglected in fine-textured soils. The soil at Moreppen does not contain clay and
adsorbed water is therefore not an issue. Different ions, at the same concentration, in the soil solution
do not affect the conductivity in the same way due to variations in ion mobility (Samouëlian et al.
2005).
In this study, cross-borehole resistivity surveys (ERT) were carried out. Cross-borehole resistivity
surveying is an extension of the conventional surface resistivity imaging. By using measurements from
electrodes in two or more boreholes, an image of the resistivity between the boreholes is obtained. The
same arrays of electrodes in boreholes can be used to obtain a resistivity profile as with surface
surveys. This method offers improved sensitivity to variations in electrical properties with depth
compared to surface-applied surveys (Binley & Kemna 2005). There are various examples of this
method and one of the first to demonstrate how this technique can be applied in hydrogeophysics is
Dailey, et al. (1992). A wide range of applications for the use of cross-borehole resistivity in
hydrogeophysical problems has developed, some include: vadose zone studies (e.g. Binely et al.
(2002)), characterizing the transport of tracer in the subsurface (e.g. Kemna et al. (2002)), and
monitoring leakage from underground storage tanks (Ramirez et al. 1996).
The main advantages of using cross-borehole compared to surface imaging are that this method offers
higher resolution with depth and investigation can be made without the need for access to the surface
(e.g. surveys under building). In comparison with surface surveys, cross-borehole method has been
shown to provide high-resolution images of hydrogeological structures and, in some cases detailed
assessment of dynamic processes in the subsurface environment (Binley et al. 2002). There are also
some disadvantages and these include the fact that boreholes are required, data sensitivity is
constrained to the region between the boreholes, more sophisticated instrumentation might be required
for data acquisition, the noise level may be much higher for surveys in the vadose zone than using
32
surface electrodes due to weaker electrical contact, and data processing is more complex (Binley &
Kemna 2005). The conditions for cross-borehole imaging are extremely variable and the acquisition
geometry should be considered on a case-by-case basis. The contrast in electrode contact and
influence of backfill or any borehole water column will vary, and so will the separation between
boreholes and the instrumentation resolution and measurement rates (Binley & Kemna 2005).
3.5.2 Relations between geophysical and soil parameters
In order to translate the geophysical parameters like the bulk resistivity measured with the ERT
method to soil properties (e.g. soil organic matter, porosity, saturation) we need to apply site specific
petro-physical relations.
Forquet (2009) explored different empirical ways to relate water content to bulk resistivity and
estimated fitting parameters for soil samples from Moreppen. He recommended using an approach
published by Yeh et al. 2002 for predicting water content due to the fact that they studied relative
changes rather than absolute ones. They used a simple model to relate water content to soil bulk
resistivity, where (Ω m) is the bulk resistivity of the soil from the ERT survey, is the water
content, (Ω m) and m are fitting parameters.
Equation 3.5-1
Looking at the difference in natural log resistivity before and after infiltration the equation becomes
linear:
Equation 3.5-2
The relationship between bulk resistivity, fluid resistivity, and saturation is found by Archie: Archie’s
law is given by (Archie, 1942):
Equation 3.5-3
where σbulk is the bulk conductivity of earth, ρbulk is the bulk resistivity of earth, F is the formation
factor, σw is the pore fluid conductivity, φm is porosity raised to Archie’s cementation factor m, and Sw
is saturation raised to Archie’s saturation factor n.
It is an empirical formula for the effective resistivity, which takes into account porosity, the saturation
level, and the resistivity of the fluid, where m and n are fitting parameters. Looking at the change in
saturation over time one can assume the porosity not to change, hence the uncertainty of the estimated
porosity is removed.
3.5.3 Trench study: Observation of a de-icing chemical breakthrough using electrical resistivity tomography at Moreppen research station
In this study we show results of the time-lapse ER measurements, suction cups data, and tensiometer
data. We have used the data of pore water electrical conductivity from suction cups and derived
saturation from the tensiometer suction measurements to combine with the time-lapse ERT
33
measurements. In this study we will analyse the impact of tracer and saturation of the porous media for
the ERT measurement results. For the ER measurement a protocol has been developed which will
discussed here as well.
Materials and methods
Moreppen soil parameters measured in the laboratory
To relate soil water content to bulk electrical conductivity Forquet tested soil samples collected at 2.4
m depth at Moreppen. Solutions of different electrical conductivity were combined with different
saturation levels S (Figure 28). He then normalized the results for standard water to gain a single
empirical relation (Forquet pers. communications) for the Moreppen soil samples given by:
Equation 3.5-4
For Moreppen, Forquet (2011) also found the fitting parameters for Archie’s law. Archie’s law has
been verified down to low saturation levels, especially in fine-textured materials. It is well-verified for
coarse sand which is representative of Moreppen sediments which have insignificant amounts of clay.
The linear form of Archie’s law for unsaturated porous medium is as follows (Archie, 1942):
Equation 3.5-5
where (Ω m) is the bulk resistvity from the ERT survey, (Ω m) is the pore water resistivity
measured by the suction cups, is the porosity, S is the water saturation and m (-) and n (-) are fitting
parameters (m is the cementation factor and n is the saturation factor). Forquet (2011) found m and n
to be 0.8793 and 1.8851 respectively. Literature values for these constants are 1.5 and 2.5 for m and n
in Archie’s law and Yeh et al. (2002) obtained a value of 1.336 for m (Forquet 2011).
The value for porosity is needed to calculated saturation from Archie’s law. There are some suggested
porosity values for the sediments at Moreppen and they differ slightly. Kitterød (2008) estimated the
porosity of the three units at Moreppen using grain size distribution based on Gustafson’s analytical
equations, where he got a porosity of 0.22 for the top set, 0.23 for the fore set and 0.14 for the silt
layer in the fore set unit. Pedersen (1994) estimated porosity in the laboratory from sediment samples
from Moreppen, where he found the porosity to range from 0.3 to 0.4.
34
Figure 28: Relation between saturation and electrical resistivity for soils from Moreppen (Forquet, 2011)
Tracer experiment
During the snowmelt of 2010 a tracer experiment was carried out. The added solution (using a
pesticide sprayer on the snow cover) consisted of 100 g PG/m2
and 10 g Br/m2 and was applied on 26
March 2010, DAY 00 (Figure 29). The specifications are given in Table 1. The application was done
approximately 6 days before the snowmelt started. Previous experiments (French and van der Zee,
1999) showed that the de-icing chemicals penetrate the snow cover and infiltrates the soil as a pulse
during the initial phase of the snowmelt. During the tracer experiment, snowmelt measurement,
temperature measurements, groundwater level measurements, ERT measurement, water samples from
the suction cups and tensiometer reading were performed and will be discussed in this study.
Table 1: Specifications for the tracer experiment in Moreppen.
South wall
Date of applications 26 March 2010
Amount of applied de-
icing chemicals 1000 g PG/m2
Commercial name of
de-icing chemical
applied
Kilfrost type II
Applied inactive tracer 10 g Br/m2
Area which de-icing
chemicals were applied
4.2 * 3 m
= 12.6 m2
35
Figure 29: Application of de-icing tracer on 26 March 2010, day 00.
Snowmelt measurements
The cumulative snowmelt per day was estimated from measured snow water equivalent (SWE). This
was done by placing a cylinder with a diameter of 50 mm into the snow cover at three different
locations every day at Moreppen and weighing the snow which filled the cylinder at each place.
Assuming 1 g snow = 1 ml water, the SWE (mm) was found and averaged for the three locations using
Equation 3.5-6
where SWE (mm/day) is the water equivalent of snow cover, V is average volume of collected snow
where the weight of snow collected is converted (g snow= ml water = 1000 mm3
water) and r is the
radius of the cylinder (mm).
The precipitation equivalent was added to the SWE to form cumulative snowmelt. The reason why it
is the weight of snow which is used is that measuring the depth of snow would not give the
precipitation equivalent, as snow will compact over time. The total precipitation for the period of
interest here, staring 26th of March 2010 (day zero when tracer experiment started), measured at the
weather station at OSL was added to obtain a cumulative infiltration of total snowmelt and
precipitation for the period. Figure 30 gives the measured cumulative snowmelt. As there is no runoff
at Moreppen and evaporation is assumed to be insignificant, the cumulative precipitation is assumed
the same as the infiltration. Considering cumulative infiltration rather than time has been shown to be
useful when considering dispersion and concentration during infiltration (Wierenga 1977).
Figure 30: Snowmelt since tracer application day 0, 26 March 2010.
36
Temperature measurements
Temperature of the soil profile at Moreppen was constantly measured with a thermistor every hour at
depths; 0.05, 0.1, 0.15, 0.2, 0.3, 0.4, 0.9, 1.4, 1.9, and 2.4 m using a Campbell-logger. The thermistors
are placed 100 cm into the north wall of the lysimeter trench. The temperature values used in the
temperature correction of the ERT data were averaged for the two hours in which the ERT
measurements were carried out.
Air temperatures were also measured on an hourly basis at Moreppen, using a Campbell-logger.
ERT measurements
In the South wall of the trench at Moreppen, two boreholes were installed for cross borehole resistivity
measurements. The boreholes have a separation of 3.2 m, a depth of 4.95 m, and 34 electrodes with
0.15 m separation in each borehole. The boreholes are located 1.4 m from the trench wall.
A Syscal Pro Switch (Iris instrument) was used to obtain the ERT measurements. This is a versatile
electrical resistivity meter, supplied by a 12 V battery, and combines a transmitter, a receiver and a
switching unit in one single casing. The measurements are carried out automatically and stored in the
internal memory (quality factor, output voltage, stacking number) after operator has selected limit
values. The output specifications are 800 V (power switch), 1 000 V (manual mode) for the voltage,
2.5 A for the current and 250 W for the power using the internal DC/DC converter and battery.
A dipole-dipole configuration was used with a fixed spacing of 0.45 m (three electrode-pair spacings)
between both the current electrode pairs and the potential electrode pairs. One great advantage of
using dipole-dipole configuration is that the acquisition time is reduced due to the possibility of multi-
measurements. Data collection of one dataset took approximately 1.5 hrs. It is argued by Winship et al.
(2006) the data capture time is critical and is therefore recommended to be short as each image should
reflect a “snapshot” of the infiltration through the subsurface. The reason for using a spacing of 0.45 m
is that the resolution towards the centre of the profile improves with larger electrode spacing (Slater et
al., 2000). In previous experiments (French et al., 2002), the boreholes were closer and therefore a
resolution in the centre of the profile was sufficient with 0.30 m electrode-pair spacing. With this
electrode spacing, if current is applied between electrode 1 and 4, the potential difference is measured
between electrodes 2 and 5 then 3 and 6, then 5 and 8 and so on until every possible combination of
electrode pairs with this spacing has been measured. Current is then applied to electrode pair 2 and 5
and the procedure continues.
For each couple of boreholes, 2074 measurements were programmed. The normal and the reciprocal
dipole-dipole measurement were measured, hence the total of 4148 measurements were recorded.
Rather than repeating a measurement to ensure data quality, reciprocal measurements are measured
and this is part of the program run. This is done by swapping the potential and current electrodes. The
reason to use reciprocal measurements to estimate errors is that the repeatability of resistivity
measurements does not provide an adequate error estimate. Anomalies, such as poor electrode contact
will give the same reading if a measurement was repeated; hence the error would not show. The theory
behind using reciprocal measurements is that the resistance measured should be the same even when
swapping current and potential electrodes. This permits removal of outliers prior to data inversion and
also allows characterization of data weights for the inversion process (Binley et al., 1995).
Normal and reciprocal error
From the normal and reciprocal measurement collected in the field, the error estimate was calculated
for each data point by comparing reciprocal and normal measurements:
| |
|
| Equation 1
37
where is error estimate in percent, is the normal resistivity measurement (Ω m) and is the
reciprocal resistivity measurement (Ω m). Data used in the inversions are average values of reciprocal
and normal values.
R3t code, mesh and model error
The inversions of the measured resistivity datasets were done with the code R3t (v1.6; Binley &
Kemna, 2005). A tetrahedral prism mesh was generated using the software gmsh (v2.5; Geuzaine &
Remacle, 2009) using a characteristic length of 0.3 m for the area around the boreholes (dimension of
6.4 x 2.8 m, boreholes at (1.6, 1.4) and (4.8, 1.4)) with finer characteristic length for the electrodes
(0.05 m), (63168 elements). For this specific model we calculated the model error by forward a
homogeneous soil with resistivity of 100, the result should show 100 ohmm for each separate
measurement, the difference between the measurement and the true absolute value of 100, gives the
error for each measurement. This error is the model error and has been used in the inversions.
Temperature corrections
Temperature has a strong influence on the EC of the subsurface. EC of pore water increases with
temperature due to increase in ion agitation as a result of decreasing viscosity of the fluid while the
change in the surface EC of rocks and sediments due to temperature variations are caused by changes
in the surface ionic mobility. These two mechanisms have different dependence on temperature
(Hayley et al. 2007). A study carried out by Rein et al. (2004) showed that even diurnal temperature
variations can have a relatively large effect on the EC. Campbell et al. (1948) found that the EC of
alkaline and saline soils increased exponentially by about 2.02% per °C between 15 and 35
°C. It was
suggested by Aaltonen (2001) that coarse-grained material presents a wider range in seasonal
resistivity variations due to temperature than fine-grained material.
There are two proposed types of temperature dependence relationships concerning EC variations in
sediments: one linear and one exponential. Campbell (1948) suggested a linear relationship explaining
the dependence of temperature on EC, where (S/m) is the recorded EC at temperature (°C) and
(S/m) is the EC at a conventional reference temperature of 25oC, and m (
-°C) is the fractional
change in EC per °C and varies for different materials and fluids.
[ ( )] Equation 3.5-8
Sen & Goode (1992) found supporting evidence to this relationship in shaly sands containing varying
amounts of clays.
Llera et al (1990) proposed an exponential relationship between EC and temperature where (J/mol)
is the activation energy of conduction, (J/kg K) is the universal gas constant, (K) is temperature
and (S/m) and (S/m) are the EC recorded at temperature and at the reference temperature
25oC, respectively.
(
) Equation 3.5-9
Hayley et al. (2007) tested these two theoretical models. Their first observation was that equation 3.5-
9, suggesting an exponential temperature relationship provides better results over a larger temperature
range. Using equation 3.5-8 [ ( )], they found that linear relationships for the
temperature range 0-25°C for approximating temperature dependence of surface and fluid conductivity
have similar slopes, and their data is consistent with the theoretical models. For the temperature range
0-25°C, also the one of interest in this study, the temperature dependence is not well described with
38
petro-physical models calibrated to the range 25 – 200 °C. They suggest using the linear
approximation which is specific to the temperature range of interest, equation 3.5-8. Hayley et al.
(2007) suggested a method of correcting for temperature variations in time-lapse resistivity surveys.
This method is based on equation 3.5-8. They found that for a temperature scale from 0-25°C,
theoretical linear relationships between temperature and surface and fluid conductivity of sediments
are similar to that observed in their laboratory data. The average estimated porosity of the samples in
their study was 0.3 and the samples were glacial tills with sporadic pebbles and cobbles in the sandy
clay matrix. The linear relationship is described by:
[ ( )
( ) ] Equation 3.5-10
where (S/m) is the resistivity at the standard temperature, (°C) is the standard temperature, (Ω m) is the in situ resistivity, (°C) is the in situ temperature and (°C) is the fractional change in
EC per °C for 25°C. Hayley et al. (2007) found to be 0.0183.
As temperature dependence experiments on the EC of soil from Moreppen have not been carried out
yet, the suggested by Hayley et al. (2007) was used in this work. Equation 3.5-10 was applied to the
pixel values of the individual inversions, using 25°C as the standard temperature. Measurements in
this study are carried out in mainly medium to coarse grained material. For conductivity meters,
temperature is automatically corrected for as temperature is also measured by a thermistor, where the
reference temperature today is 25°C. Resistivity measurements are not corrected for temperature
variation (Hayley et al. 2007).
The EC measurement data from the suction cups are automatically temperature corrected to 25°C by a
conductivity meter and by correcting the ERT measurements to 25°C the two datasets can more easily
be compared. The average soil temperatures measured during the same two hours as the ERT surveys
were used. The temperature measurements were used for the intervals between the upper and lower
depths of the upper and lower measurements. Due to the lack of deeper thermistors, the measurement
at 2.4 m was applied down to 5 m.
Protocol ERT analysis
Each data was analysed following the same protocol following these 6 steps:
1) Electrical resistance values were filtered and measurements with a reciprocal error higher than
30% were excluded, also repeatability errors in which the deviations within the normal or
reciprocal repetition higher than 10 were excluded, and measurements which had a geometric
factor higher than 10000 were excluded from further analysis.
2) For each dataset obtained in step 1 a measurement error model was calculated subdividing
resistance values into bins according to their values, creating averages of errors and
resistances for each bin and calculating a regression equation thereof (Koestel et al., 2008).
The model error has been added to the measurement error specific for each measurement in
the sequence. This total error model has been used in the inversions.
3) For each dataset obtained in step 1 together with the error model obtained in step 2 was
inverted using the code R3t (v1.6; Binley & Kemna, 2005). The GMSH model described
above has been used in the program.
4) The inversion results were corrected for temperature as described above.
5) The temperature corrected model obtained in step 4 was then forward and corrected for the
difference in background to measurement date.
6) Of the datasets from step 5, one common datasets was created. This was to reduce bias
occurring from comparing datasets based on different number of measurement points. The common dataset contained 1099 measurements. With these datasets, finally a standard
inversion as well as a difference inversion was carried out using the method of Labrecque &
39
Yang (2001). Using this method, the difference of measured resistances between 01/04/2012
and the subsequent dates was added to forward modelled resistances based on the inversion of
the data from 01/04/2010. This resistance data was then inverted for each of the subsequent
dates. For the difference inversion the error model was based on the additive error of
01/04/2010 and each individual subsequent date.
Suction cup data
The aim of using a lysimeter trench is to monitor water flow and pollutant transport by extracting soil
water from the different depths of the soil profile. In the south, north and west wall of the trench
suction cups, made of Teflon and quartz (Teflon avoids ion sorption compared to using ceramics) were
installed in the period 1993-4. In this study, only water samples from the south wall are used. The
Teflon suction cups have a pore size of 2 µm and a porous area of 33 cm2 (Prenart) The suction cups
are placed in the whole profile and to ensure that the soil remained undisturbed above the suction cups,
the distance from the wall increases with 10 cm for each deeper layer of suction cups. A vacuum pump
ensures the designated constant suction of the set up (French et al. 1994). The area monitored between
the boreholes both for GPR measurements and ERT are the same as that monitored by the suction
cups, hence the same infiltration of snow and chemicals (Table 1). A diagram showing the distribution
of the suction cups in the south and north wall is seen in Figure 31 and this illustrates the resolution
limits of the pore water EC data down the profile. The suction cup placed at 4.5 m depth is to allow
groundwater to be sampled (French 1999).
Depth (m) South wall suction cups Distance from
wall (m)
-0.4 1 2 3 4 0.7
-0.9 5 6 7 8 9 0.8
-1.4 10 11 12 13 14 15 0.9
-1.9 16 17 18 19 20 21 22 1
-2.4 23 24 25 26 27 28 29 30 1.1
-2.9 31 32 33 1.1
-3.2 34 35 36 37 1.05
-4.5 38 0.5
2.5 m
Figure 31: Diagram showing the distribution of suction cups in the south wall of the lysimeter trench at
Moreppen. One number represents one suction cup in the soil profile (Adapted from French (1999)).
A closed system of PVC pipes connects each suction cup in the soil profile to its respective Prenart
collecting bottle inside the lysimeter trench. The applied suction should be as small as possible to limit
the influence on the natural flow patterns of water in the soil profile. Suction is needed due to the low
potential head in the unsaturated zone; it is required to collect water from the soil. The vacuum pump
was set to a constant suction 0.15 bar in December 2009. Before this the vacuum pump was coupled
up to tension meters in the profile so that the suction applied varied with the natural variations, with a
minimum suction of 0.03 bars. However, it was found that not all tension meters worked properly a
fixed suction was applied (Elkin 2011). Soil water first runs into a test tube before running into the
collecting bottle. The water sample in the test tubes were collected when filled with water and frozen
for later measurements in the laboratory using a conductivity meter.
The GMSH model created to analyse the ERT results consisted of layers which coincide with the
suction cup layers. Each boundary is in between two suction cup depths. Please note that from a depth
40
of 3.2 m no more suction cups were installed so we have used the suction cup results from 3.2 m to up
to 5 m depth.
For each layer at each time step the EC values obtained in the laboratory were averaged to get one
value for each layer and then forwarded with the R3t program using the GMSH model and inverted
again with the R3t program. During the inversion the model error was used to weight the ‘created’
measurement points. The inversion was done assuming 2074 measurement point as is the case with the
ERT, as well as with the created common dataset of 1099 measurements from the ERT to see if there
would be any difference in the results.
Tensiometer data
At the south wall tensiometers were installed at depths of: 0.40m, 0.56m, 0.72m, 1.90 and 2.40m.
Readings were automatically taken every 5 minutes. The suction values measured at 12:00 by the
tensiometers were then converted to saturation values using the parameters from the pF curves
produced by Pedersen (1994) and Forquet (2009). For each tensiometer layer (boundary between the
depths of the tensiometers the pF values in this layer has been averaged and used to calculate the
saturation.
We used the empirical relationship between saturation and bulk resistivity from Forquet (Equation 3.5-
4) to calculate the bulk resistivity from the measured saturation. Note that this is the bulk resistivity as
obtained from normalized water, so only the saturation affects the resistivity here.
The GMSH model created to analyse the ERT results consisted of layers which coincide with the
tensiometer layers. Each boundary is in between two tensiometer depths. Please note that from a depth
of 2.40 m no more tensiometers were installed so we have used the suction cup results from 2.4 m to
up to 5 m depth.
For each layer at each time step the calculated bulk resistivity was then forwarded with the R3t
program using the GMSH model and inverted again with the R3t program. During the inversion the
model error was used to weight the ‘created’ measurement points. The inversion was done assuming
2074 measurement point as is the case with the ERT, as well as with the created common dataset of
1099 measurements from the ERT to see if there would be any difference in the results.
Results and discussion
ERT data
Temperature corrections
Throughout the period monitored, there are great variations both spatially through the profile and in
time and illustrates the importance of temperature correction of ERT surveys were the focus is on
changes due to infiltration. As can be seen in Figure 32 the temperature near the surface ranges from -
0.4°C to 4.4°C during the 25 day melting period. Within the profile temperature varies between -0.40
to 2.60°C at day 6. The temperature variations combined with the stable soil temperatures prior to
snow melt is the same as seen in the snowmelt period 2001 (French & Binley 2004). As the soil
temperature for the top 0.4 m until the 14th of April were below freezing, the same assumption is made
here, as by French and Binley (2004) that the top soil at Moreppen was frozen until the 14th of April
and that prior to this date the percolation took place through a frozen layer. The ice layer on the
ground surface is mostly responsible for the redistribution of melt water and cause preferential flow. It
has been found that the typical seasonal soil frost at Moreppen is about 0.4 m deep during the winter
season (French et al. 2002). This depth corresponds to the soil temperatures measured in 2010 as well.
41
Figure 32: Temperature profiles measured at Moreppen research station during ERT measurements at day 6,
day 12, day 19 and day 31.
Snowmelt and groundwater pattern
The response of the groundwater level and pattern and amount of snowmelt is of interest as these are
part of what is potentially the signal seen in the ERT data. The cumulative snowmelt calculated is
assumed to equal the amount of infiltrating water during snowmelt as there is no runoff in the area.
Groundwater level responds rapidly to the first snowmelt event (Figure 33), approximately at 31st
April. This illustrates the short response time of the system to infiltration through the highly permeable
sediments at Moreppen. The significant increase in groundwater level, from 4.94 m depth to 4.56 m, is
seen to change in about 20 days, about the same time as the majority of the snowmelt infiltration,
which levels off after the 15th of April. The cumulative snowmelt is seen to increase quite steady, with
a short stagnation between 6th and 8
th of April. The same pattern in rapid response from groundwater
as a consequence of snowmelt was found by French and Binley (2004).
Figure 33: Graph showing the calculated cumulative snowmelt in mm from Moreppen and the measured daily
averages groundwater level (m) at Moreppen during the time of interest 2010.
42
ERT profiles
Individual inversions and difference inversions.
The individual inversions shown in Figure 34 first row are independent inversions of day 6, day 12,
day 19, and day 31. The difference inversions in the same figure second row show the inversion of day
12, day 19 and day 31 in which day 6 is used as the background resistivity.
On day 6 low resistivities are already shown at the top of the profile and increase rapidly towards a
larger depths. Especially in the difference inversions, it can be seen that the profile wets up quickly on
day 12. It is also visible that the groundwater responds to the snowmelt. This is in agreement with the
rapid groundwater rise observed in Figure 33 and previous studies (French and Binley, 2004).
From these figures alone it is difficult to decipher whether the change in conductivity is due to water
saturation increase or the de-icing chemical and tracer (causing an increased electrical conductivity of
the soil water) or a combination of these. Therefore we explore how the saturation and EC affect the
bulk electrical conductivity based on independent datasets namely from the tension meters and the
suction cups providing EC measurements.
On day 19 the profile is wet and most likely the tracer is reaching greater depths. On day 31 the soil
profile becomes closer to the initial conditions. The topsoil shows increased resistivity, while the water
table rise stays intact (low resistivities at the bottom of the profile).
Day 06 Day 12 Day 19 Day 31
Figure 34: First row shows the ERT results using individual inversions of day 6, day 12, day 19 and day 31. The
second row shows the results for the ERT difference inversions of day 12, day 19, and day 31 with background
day 6. On the x-axis is the distance and on the z-axis is the depth.
43
Figure 35: Depth avareaged ERT results using individual inversions of day 6, day 12, day 19 and day 31.
Suction cup data (electrical conductivity of the soil water)
The EC data from the samples taken with the suction cups show the tracer distribution. Figure 36
shows the resistivity calculated from the measured EC values for the profile at days 6, 12, 19 and 31.
From day 6 to day 12 there is a strong transition from high to low fluid resistivity near the surface. The
snow starts melting together with the tracer, therefore the resistivity drops rapidly, after the tracer has
past, the top layer receives only fresh rainwater and the resistivity increases.
At 0.9 m depth the resistivity first increases from day 6 to day 12, indicating more fresh water is
received, but without the tracer. After that the tracer enters the profile and the resistivity drops at day
19. At day 31 the tracer has passed and the resistivity increases again. The same trend can be observed
at 1.4 m and 1.9 m depth. Here the resistivity increases until day 12 and then drops at day 19.
At the deeper layers the difference in resistivity is less prominent. This might be due to the fact that
the soil is wetter at higher depth, and the tracer becomes more diluted.
44
Figure 36: Resistivity profiles of pore water taken with the suction cups during day 6, day 12, day 19, and day
31 to show the contribution to bulk resistivity from EC of pore water.
Tensiometer data (Saturation levels)
Figure 37 shows the resistivities derived from the tensiometer readings along the profile for day 6, 12,
19 and 31. From day 6 to day 12 there is a decrease in bulk resistivity near the surface, due to an
increase in saturation. The snow melts, rapidly wetting the top soil, resistivity drops down rapidly,
after snow melt the top layer slowly becomes dryer again and the resistivity increases. At day 6 the
soil tension is reduced. Therefore the pattern at 0.6 and 0.7 m depth is even sharper. The same patterns
as at the top surface can be observed at 0.6 m depth. The resistivity first decreases from day 6 to day
12, indicating that the soil is getting moister. At 0.6 m depth and even better at 0.7 m depth, the delay
in soil wetting can be nicely seen. The wetting of the soil at 0.7 m is slower and takes longer time until
day 19 to wet up completely before it starts to dry again. At day 31 the soil dries out and the resistivity
increases again through the whole profile. At the deeper layers the difference in resistivity is less
prominent. This might be due to the fact that the soil is naturally wetter at higher depth.
45
Figure 37 Resistivity profiles derived from soil saturation levels, which are based on the suction measured by
tensiometers (equation by Forquet, 2011), during day 6, day 12, day 19, and day 31 to show the contribution to
bulk resistivity from soil saturation levels in the Moreppen soil.
ERT, suction cup and tensiometer data trends
By combining the ERT data with the water samples from the suction cups and the water content data
derived from the tensiometer reading a complete overview of the profiles are obtained (Figure 38).
Now we can clearly distinguish the processes going on, the wetting of the snow due to snowmelt
infiltration as well as the groundwater rise due to the snowmelt. And we can analyse the movement of
the tracer.
The response of the suction cup data is slower and also shallower than the response of the tensiometer
data which is at day 12 and day 19 much deeper. At day 31 all measurement techniques show
increasing resistivities over the whole profile.
Due to the much larger density in measurement points with the ERT method is would be easier to
detect resistivity boundaries than with the suction cup and tensiometer data. Those profiles are made
only with minimum data points and therefore cannot show the exact resistivity transitions.
Be aware that both suction cups and tensiometers are only installed up to 3.2 and 2.4 m maximum,
therefore the results at the deeper part up to 5 m cannot be compared to the ERT (Figure 38). The
groundwater level shown in the ERT data is therefore not visible in the suction cup and tensiometer
data. This is one of the advantages of the ERT measurement. It is easier to install electrodes deeper
and the ERT method can also be used in saturated soil.
46
Figure 38: Resistivity profiles of pore water taken with the suction cups, profiles derived from soil saturation
levels, which are based on the suction measured by tensiometers (formulae by Forquet, 2011) and profiles from
the ERT inversions, during day 6, day 12, day 19, and day 31 to show the contribution to bulk resistivity from the
tracer and soil saturation levels in the Moreppen soil.
3.5.4 Lysimeter study: Observation of a de-icing chemical breakthrough using electrical resistivity tomography in closed system lysimeters
The subsequent paragraphs aim for groundtruthing of time lapse electrical resistivity tomography
using monitoring data of water contents and pore water electrical conductivity in FSU lysimeters. It is
intended to unravel the parameters of influence on electrical resistivity changes during infiltration of
melt water loaded with de-icing chemicals to be able to delineate the potential of ERT to monitor such
events. A primary control of electrical resistivity changes by de-icing-chemical infiltration or
degradation would be desirable, if this method should be used in monitoring systems for airfield
management. If water content changes are dominating the electrical response, such data can serve as
soft data for unsaturated flow model validation and calibration.
Materials and Methods
Lysimeter experiment
Data of the 2010 snowmelt-de-icing chemical infiltration experiment in the FSU lysimeters was used
to compare results of the electrical resistivity tomography (ERT) and hydrogeochemical data of the
reactive tracer breakthrough. The time series was chosen to start on 01/04/2010 (2 days after snowmelt
started; 8 days after tracer application) to end of May (28/05/2010; 66 days after tracer application),
when the drainage before summer cessation was completed. The conduction and data of the snowmelt
experiment and its hydrogeochemical effects is described in detail in D2.4. ERT was carried out in
four lysimeters but only lysimeter 2 is treated in the subsequent example. The infiltrating solution (5l,
2080 µS/cm) contained 0.2 g/l Br, 4.9 g/l PG and 1.0 g/l Formate, where the potassium formate
contributed most to the electrical conductivity of the solution (1950 µS/cm).
The setup of the lysimeter 2 is described in detail in D 2.4, chapter 2.1.2. It contained 18 electrodes for
ERT, installed in two opposite rows (depths 10cm, 20cm, 30cm, 40cm, 50cm, 60cm, 70cm, 80cm and
90cm). Suction cups and FDR probes were installed on opposite sides of a vertical plane perpendicular
to the electrodes (3 FDR Probes at 20cm, 50cm and 80 cm depth, suction cups at 20cm, 40cm, 60cm,
80cm and 90cm depth) and seepage water was collected at the lower boundary. The seepage water was
analysed for electrical conductivity, pH, bromide, propylene glycol, formate, organic carbon, anions
and cations. Suction cup samples were analysed for bromide, pH and electrical conductivity.
The ERT measurements were carried out using a SYSCAL Pro switch (Iris Instruments, Orleans,
France). A dipole-dipole configuration was used with zero skipped, one skipped and two skipped
47
electrodes. Including reciprocals, 548 measurements were conducted for each lysimeter. The output
specifications are 800 V (power switch), 1 000 V (manual mode) for the voltage, 2.5 A for the current
and 250 W for the power using the internal DC/DC converter and battery.
Column experiment
A column experiment was conducted to analyse electrical properties of the soil material on the small
scale. Soil material from the lysimeter extraction site was refilled into 2 PVC columns (area A=12.4
cm2, length l=15cm) and a solution of NaNO3 47.144mg/l, CaSO4 185.279 mg/l and KHCO3 11.27
mg/l (318 µS/cm @ 15.5°C) was percolated under saturated conditions until a stable fluid electrical
conductivity was observed in the column outflow. Current in the 2 soil columns was measured while
applying 20V over flat circular (4cm diameter) stainless steel meshes on top and bottom of the
column. Bulk electrical conductivity was calculated from the measured resistances using the
geometrical factor A/l. The ratio of bulk conductivity and fluid conductivity was used to calculate the
formation factor (see below). The water filled porosity of the material was 0.33 and 0.32 and the bulk
density 1.73 g/cm3 and 1.76 g/cm
3.
Data Analysis-ERT
Electrical resistance values were filtered and measurements with reciprocals deviating by more than
10% were excluded from further analysis. The remaining data was inverted using the code R3t (v1.6;
Binley & Kemna, 2005). An error model was calculated subdividing resistance values into bins
according to their values, creating averages of errors and resistances for each bin and calculating a
regression equation thereof (Koestel et al., 2008). A tetrahedral prism mesh was generated using the
software gmsh (v2.5; Geuzaine & Remacle, 2009) using a characteristic length of 1.5cm on a cylinder
of 1m length and 0.3m diameter (180448 elements). A difference inversion was carried out using the
method of Labrecque & Yang (2001). Using this method, first an inverse model of the data from
01/04/2012 is created and this is used to forward model resistances of the applied electrode
configuration. Then, the difference of measured resistances between 01/04/2012 and the subsequent
dates is added to these forward model resistances. This resistance data is then inverted for each of the
subsequent dates. For the difference inversion the error model was based on the additive error of
01/04/2010 and each individual subsequent date to account for error propagation.
Temperature correction
The bulk conductivity σbulk (S/m) (inverse of the bulk resistivity ρbulk (Ωm)) resulting from the ERT
data was corrected for soil temperature to derive the conductivity at 25 ◦C (Campbell et al., 1948):
( ) ( ( ))
Data Analysis-Pore water
Pore fluid conductivity was analysed with a conductivity meter on site (WTW, Weilheim, Germany).
Anions were analysed in the lab with ion chromatography (IonPac AS11-HC,AS50/EG50/IC20/CD20
Dionex, Idstein). Samples were filtered in the lab (0.45µm filter with supor membrane). Cations were
analysed by ICP-OES (725 ES, Varian). The samples were filtered and acidified immediately after
sampling.
Pore fluid conductivity analysed in suction cup and seepage water and water content changes from
frequency domain reflectometry (FDR) probes were used to predict a bulk electrical resistivity change.
Prediction of bulk resistivity was done applying Archie's law (Archie, 1942):
where σbulk is the bulk conductivity of earth (S/m), ρbulk is the bulk resistivity of earth (Ωm), F is the
formation factor, φm is porosity raised to Archie’s cementation factor m, and Sw is saturation raised to
Archie’s saturation factor n. Values for m were derived in the column experiment described above.
The cementation factor was calculated as m=1.44. The saturation exponent was set to n=m (Revil,
2012). Predictions of bulk resistivity using pore fluid conductivity were available for the depths 20cm,
48
80cm and 90cm, where suction cup data was available and 100cm, where seepage water data was
available. Saturation data was calculated using the water content estimates from the nearest FDR
probe. Porosity was derived from cutting core rings and set to 0.3 in 20 cm depth, 0.38 in 80 cm, 0.34
in 90 cm and 100 cm. Because the focus is on time lapse changes of resistivity, it is assumed that the
contribution of surface conductivity can be neglected. Such predicted changes of ρbulk were compared
to median and range of the electrical resistivity tomograph in the according depth.
Results and Discussion
Resistivity versus soil properties
Figure 39 shows the ER tomograph on 01/04/2010 (temperature corrected), the day of the snowmelt.
Three horizons of different resistivity can be discerned: a low resistive horizon 1 in 0-0.1 m depth, a
higher resistive horizon 2 in 0.1-0.7 m depth and a low resistive horizon 3 in 0.7-1.0 m depth. This
subdivision into three horizons is reflected in the bulk density and porosity values. In particular,
horizon 2 with its high resistivity, low porosity and high bulk density is remarkable, because it
contradicts the expectation of a continuous increase of bulk density and decrease of porosity with
depth. The reason for this anomaly is former construction activities near the runway. L2 was extracted
at a spot, where a dirt road for construction machines used to be, which could have resulted in
compaction of surface near layers. It seems that this anomaly can be sensed with ERT, which is
probably due to the low water content of this compacted layer.
Figure 39: a) ERT compared to b) porosity and c) bulk density in cutting core rings for lysimeter 2. Resistivity
bulk in a) was divided by (m) before logarithmization. Values of porosity and bulk density for the depths 0.025
m, 0.33 m, 0.84 m were derived in a soil pit next to the extraction site of L2 the values at 0.88 m directly in the
pit of L2.
Time lapse resistivity
The time lapse resistivity results for 6 dates are presented in Figure 40. Generally, they show a
reduction of resistivity with time. 4 days after start of snowmelt (05/04/2012), all horizons decreased
in resistivity. Another five days later in particular horizon 2 is affected. Towards the end of the time
series resistivity increases again.
-1
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
20 40 60
de
pth
[m] porosity Vol%
1 1.5 2
bulk density g/cm3
a) b) c)
log bulk
49
The changes are either an effect of the soil moisture increase of the profile due to infiltrating melt
water or of the relatively large electrical conductivity of this melt water (containing de-icing chemical
solution) or of a combination of both. As will be explained later on, the resistivity decrease is unlikely
to be attributable to water content changes. Generally one would expect a decrease of resistivity
progressing continuously from the top to the bottom. The fact that the decrease of resistivity is
observed in horizon 3 already at the beginning and then later in horizon 2 hints on tunnelling of melt
water through preferential flow paths. The increase of resistivity towards the end of the observation
period is due to the drying out of the soil as summer approaches or the dilution of the pore water with
low conductivity rain water.
01/04/ 05/04 12/04 16/04 21/04 28/05
Figure 40: Time lapse ER tomographs starting on the day of snowmelt 01/04/2012. Resistivity bulk was divided
by (m) before logarithmization.
Ground truth of time lapse resistivity
The ground truthing of the ERT is shown in Figure 40 and Figure 42, where the predicted bulk
electrical resistivity of the medium using pore water conductivity from aqueous samples and water
content changes by FDR is compared to the median of the ER tomographs in the according depth.
Figure 41 shows a comparison of the absolute values in 90 cm depth. The resistivity is strongly over-
predicted by Archie's law. In Figure 42, the ratio of bulk resistivity changes in relation to 01/04/2010,
the start of snow melt, are shown. In general, predictions and measurements show a decreasing trend
of about the same order of magnitude. Yet, the dynamics are quite different in detail. In 90cm, the
drop of resistivity is observed earlier in the ERT than in the suction cup data. Also, a rise of resistivity
is observed in the ERT data towards the end of the period in 80 cm depth. This leads to a considerable
disagreement of prediction and observed resistivity.
The disagreement of the absolute values (Figure 41) is probably due to neglecting the contribution of
surface conductivity to bulk conductivity in the model. Due to the lack of measurements of the surface
conductivity, it was not possible to estimate this contribution. Current models based on the use of
cation exchange capacity to predict the surface contribution (Sen et al., 1988; Günzel, 1994; Revil,
2013) failed likewise to predict the resistivity observed by ERT.
When comparing time lapse ratios of ERT and EC in extracted water from suction cups, it has to be
considered that the suction cup data represents point measurements or some volume around each
suction cup while the ERT method integrates over a larger volume. This can lead to the effect
log bulk
50
observed for the depth 90cm - part of the core in this depth is reacting quite early on the input of
conductive melt water, while in the suction cup this is observed much later. This observation was
already pointed out in the discussion on the tomographs (Figure 40). In Figure 40, it appears that this
early breakthrough in the deeper region on 05/04/2012 redistributes and the high conductivity gets
diluted. The effect is also observed in Figure 42 in 80cm depth. Consequently, while time lapse
resistivity predictions and observations agree in the sense that they show a reduction of resistivity with
time, details of the time series differ due to different sampling volumes of the medium.
Figure 41: Comparison of predicted bulk resistivity, based on suction cup fluid conductivity in 90cm depth and
water content data versus median, minimum and maximum of ERT in the according depth.
0
50
100
150
200
250
300
350
400
28
.3.
2.4
.
7.4
.
12
.4.
17
.4.
22
.4.
27
.4.
2.5
.
bu
lk r
esis
tivi
ty (
oh
m m
)
suction cup 90
median ERT
51
Figure 42: predicted bulk resistivity using pore water conductivity data from aqueous samples and FDR data
with median of the ERT in four depths. The error bars denote minimum and maximum of resistivity in the
according depth.
Figure 43a shows the influence of water content changes on the prediction of the bulk resistivity in
100cm depth using Archie's law based on fluid conductivity data from aqueous samples and FDR
probes. Archie's law was used including water content changes calculated with FDR data in 80cm
depth and also neglecting them by keeping the saturation constant. It becomes obvious that the
predicted bulk resistivity is hardly influenced by the water content changes. Therefore, it must be
assumed that also the decrease of resistivity observed in the ERT is mainly driven by the high
electrical conductivity of the melt water.
Figure 43b shows the increase of electrical conductivity in the seepage water over the observed period.
A large proportion, maybe around 50%, of the electrical conductivity originates from bromide,
potassium and formate. This can be estimated using the rule over thumb that 1 meq/l≈1µS/cm.
Inorganic carbon plays only a minor role. A considerable proportion of ions contributing to electrical
conductivity was not identified, although major anions, cations and metabolites of de-icing chemical
were analysed. It must be assumed that the electrical conductivity of the seepage water is to some
degree determined by unidentified ionic metabolic products. Thus, the ERT response is mainly due to
infiltration of the de-icing chemical potassium formate, bromide and unidentified metabolic products
of de-icing chemical degradation.
0
0.5
1
1.5
28.3
.
2.4
.
7.4
.
12
.4.
17.4
.
22.4
.
27
.4.
2.5
.
tim
e la
pse
rati
o (0
1/04
/201
0) suction cup 20 cm
average ERT
0
0.5
1
1.5
28.3
.
2.4.
7.4.
12.4
.
17.4
.
22.4
.
27.4
.
2.5.
tim
e la
pse
rati
o (0
1/04
/201
0)
suction cup 80 cm
average ERT
0
0.5
1
1.5
28.3
.
2.4.
7.4.
12.4
.
17.4
.
22.4
.
27.4
.
2.5.
tim
e la
pse
rati
o (0
1/04
/201
0)
suction cup 90 cm
average ERT
0
0.5
1
1.5
28.3
.
2.4.
7.4.
12.4
.
17.4
.
22.4
.
27.4
.
2.5.
tim
e la
pse
rati
o (0
1/04
/201
0)
seepage water 100 cm
average ERT
52
Figure 43: a) bulk resistivity prediction based on FDR and fluid conductivity data from aqueous samples in
100cm depth using Archie's law considering water content changes and neglecting them b) contributions of
different ions to electrical conductivity of the pore water.
3.5.5 Conclusions
These studies show that ERT is sensitive to soil structures such as increased bulk density and soil
texture. Time lapse investigations can be used to monitor infiltration of melt water, in particular, if it is
highly conductive. In our experiment, the conductivity resulted mainly from potassium formate and
products of the degradation. During infiltration at the airport site, no bromide would be present, but
since there Potassium formate is applied on the runways, this will always be present and contribute to
increased electrical conductivity, hence still visible. Propylene glycol alone would be not sensed with
ERT as it is not conductive. Also, it is unknown, whether its degradation products are present in
sufficiently large amounts to be sensed, because in the demonstrated study, formate and propylene
glycol were added concurrently.
It is important to point out, that for other soil materials with a large change in water content during
infiltration the ERT signal would be affected twofold: by the change of water contents and the
infiltration of a conductive solution. Keeping these implications in mind, time lapse ERT can be
recommended as a tool for monitoring de-icing chemical infiltration on the airfield. In particular, it is
capable of direct observation of preferential transport processes, which are difficult to evaluate with
other methods. It is recommended to use an ERT monitoring tool to cover a larger spatial extent and
support the interpretation with a point wise method of ground truthing, such as suction cups or plates.
If a lower spatial resolution of the ERT is implemented, further validation steps are recommended.
0
50
100
150
200
250
300
350
400
30.03. 04.04. 09.04. 14.04. 19.04.
bulk
res
isti
vity
(oh
m m
)
date
Forward model neglecting soil moisture changes
Bulk electrical resistivity forward model
0
20
40
60
80
100
120
0
100
200
300
400
500
600
700
30/03 09/04 19/04 29/04 09/05
C (
mg
/l)
EC (
µS/
cm)
date
el. conductivity
formate
K+
TIC
Br-
53
3.6 Remediation monitoring by time-lapse geophysics
As geophysical methods yield information about the subsoil and in some cases its contamination, also
changes may be measurable. For this potential, obviously the same constraints hold with regard to the
accuracy, resolution, depth interval, horizontal scale of each method, as well as on the local field
conditions [D5.3].
Several experimental schemes were used for different geophysical investigations [D1.1] [D1.2] [D1.3]
[D1.4]. Some geophysical measurements were performed using several equipments with different
capabilities and output, for comparison. The primary geophysical data processing steps and results
interpretation were given the most attention, are used to propose improvements [D1.2] [D1.3] [D1.4]
[D4.4], [D5.3].
Time-lapse geophysical results such as electrical resistivity data, or georadar data, can be prepared and
interpreted separately for each time, just like other monitoring results. They can also be inverted
jointly [D1.4] to determine changes in time. The results have general applicability for a wide range of
contaminated sites. Examples that appear promising are:
Monitoring infiltration experiments and phenomena: Time-lapse ERT measurements of infiltration
may provide images of preferential flow, which is a major complication in assessing contaminant
transport in the soil. Because of their much better spatial resolution, fast pathways can be detected and
the distribution of travel times, e.g. towards groundwater, can be quantified. At
Moreppen/Gardermoen, ERT was in good agreement with direct monitoring methods.
In the case of biogeochemically reactive chemicals and contaminants, time-lapse measurements can be
aimed at quantifying the rate of transport or the rate of depletion in case of degradation. In the first
case retardation has to be accounted for: how much slower than the transport carrier water will the
contaminant move. In the case of Trecate, the main body of light oil contaminant at the capillary
fringe depletes very slowly because of the entrapment mechanism. Time-lapse monitoring will then
only lead to good results, if a long time between measurements, i.e., order of years, is considered. The
same is the case if the contaminant is persistent, i.e., slow to degrade. In that case, changes in
contaminant mass may also be slow enough to warrant periods between measurements of years.
The possibility of monitoring by time-lapse measurements was studied in the Trecate site at the scale
of the zone between the boreholes B-S3 and B-S4. Monthly cross-hole ERT measurements at several
frequencies were performed starting from December 2009 and provided data sets regarding the field
situations under yearly variable ground water levels in a contaminated area of the site. 2D resistivity
distributions were obtained by inversion in the vertical plane between the two boreholes, being
constrained by consideration of three subsurface layers (top layer, vadose zone layer, saturated layer).
Whereas the results revealed a coherent variation as a function of time of the electrical resistivity
[D1.2], but the interpretation for the vadose zone appeared to be complicated. After time-lapse ERT
data processing, direct conversion of resistivity values to saturation was not found to be a realistic
option, and the integration with borehole data was considered necessary [D1.2]. In particular, the
interpretation of the resistivity of the control water (not contaminated) was difficult, because the
background values changed seasonally due to diverse processes upstream and unrelated to Trecate, as
well as the impact of degradation on the downstream water quality: released degradation products
were affecting the non-invasive measurements. Whereas this supports the potential of such
methods to detect degradation, hence changes in the contaminants, as well as chemical transport
in groundwater, the limited duration of the SoilCAM project did not enable a matured way to
deal with these complications.
54
4. Integration of non-invasive and invasive methods for remediation monitoring
4.1 The monitoring-modelling-monitoring loop
With geophysics, we obtain a better impression of the subsoil, and of the location and quantity of
(some) contaminants. In case of degrading contaminants, also the impact of this process on the broader
environment can be detected. The potential of invasive and non-invasive methods in combination is
therefore related with:
improvements of data support, quality, and interpretation
improvements in the conceptual model
updating the conceptual model to comply with changes after a long period
extension of the conceptual model with added parameters & processes
Such improvements are typically related with a cyclic approach to site management, where among
others, the monitoring leads to a conceptual model, which is used for modelling. The modelling may
then lead to the awareness that insufficient data are available, in general or at a specific location,
which may give rise to new measurements.
At the Trecate site, the combination of data from invasive conventional (chemical, physical,
(micro)biological) and non-invasive geophysical measurements has resulted in the identification of the
capillary fringe (at the boundary of vadose and water saturated zones) as the location where the oil
contaminant was concentrated, even decades after the spill. This necessitated the modelling of the
capillary fringe, which was done in two ways: (i) a laboratory experiment was conducted, and
literature data from another laboratory experiment was also taken into consideration, and (ii) these
experiments were modelled, to establish the processes if a light oil spill at the capillary fringe is
subjected to water level variations. This modelling effort, that required that completely different
(multiphase physics flow and transport) software was introduced into the SoilCAM project, explained
the persistence of the oil due to capillary entrapment.
This awareness had a great impact on the conceptual model of Trecate. It appeared that the
contaminant’s position was now well known, and that the gradual, slow release into groundwater was
understood, and enabled the definition of a contaminant source term in the regional modelling with
confidence. Also, for determining the degradation of the contaminants, it was clear that this occurred
at the capillary fringe and involved pure phase oil. Regarding the environmental risks, it was clear that
large emissions to the surrounding areas are not very probable. Geochemical results indicated a new
component (manganese) as having an important influence during the contaminant degradation, and the
conceptual model was extended to include the corresponding substances in the monitoring program.
This is of course also the case for the effects of such substances, both environmentally and for the
monitoring methods. For further monitoring, it was essential that the invasive observations of
groundwater levels were correct and that the depth of the groundwater level interpreted from the
geophysical methods is distorted due to the presence of a pure oil phase.
Interpretation of resistivity has to be done with care also. Resistivity results derived from geophysical
measurements give much information that has to be further improved especially as regards its
accuracy and proper interpretation. For instance, ground water conductivity upstream the Trecate site
has values that range between about 300 – 400 μS/cm [D1.2] [Bolognino&Godio 2008] [D2.4 FSU
data]. Every set of resistivity results and also time-lapse results should be interpreted taking into
consideration the upstream ground water conductivity and the situation of the contaminant plume at
the respective times. The presence and concentrations of various ions in ground water change its
55
conductivity and consequently resistivity. An example has been shown for the resistivity in relation
with water electric conductivity and porosity in [D5.3].
Resistivity estimations in combination with measured water conductivity showed two distinct layers in
the upper part of the aquifer in the contaminated area. This is confirmed by the vertical profile
observations on water conductivity.
Figure 44: Vertical section in the contaminant plume
Figure 45: Estimated resistivity influenced by the contaminant plume if almost homogeneous soil layer is
considered in the plume zone
Differences between resistivity values inside the plume and outside may be rather small in comparison
with the large range of measured geophysical results.
56
4.2 Effective strategies for combination of methods
Subsoil characterization
The geophysical methods have been used at both Gardermoen and Trecate for the characterization of
the subsoil structures. In both cases, this has been successful: contrasting layers and sublayers could be
distinguished and these different structures were in agreement with the information that was gained
non-invasively.
The precise output variables for the various geophysical methods may differ. Whatever the outcome of
the inversions, an image is obtained that suffers from two aspects: (i) the image is subject to error and
uncertainty, due to different sources, and (ii) the image is generally not in a form that is directly of use
in the geosciences modelling.
These aspects are something that is continuously subject of research and gradually the inversion
process will lead to better and more useful results. For instance, to translate the images into useful soil
properties suffer from the non-uniqueness of the relationships between what is measured (e.g.
resistivity) and what it stands for: water content, type of mineral, texture, chemical concentrations in
solution, etc. To address this non-uniqueness, joint inversion methods have been developed, and will
be further developed. In this respect, the consideration of conventional approaches based on soil cores
and water samples have contributed to the interpretation of the subsoil. However, to be realistic, it has
to be admitted that an automated inversion of geophysical measurements into soil parameter fields
remains a problem, if nuances of parameter changes are the scope of research. As the measurements at
Trecate indicated, coarser features such as decidedly higher (or lower) permeability layers, and the
position of the contaminated zone (in case of oil spill) can be identified.
The complexity to derive soil parameter fields of the soil surface does not mean that with the current
state-of-the-art, these methods do not contribute. In this respect, the Gardermoen area is a good
example, as the textural variability of this coarse sand material are limited, from the scope of
geophysical detectability. Images with structures were obtained, but these cannot directly be translated
into texture fields without access to ground truthing data along the same lines.
The complexity is, that the texture affects water retention as well as hydraulic conductivity, and that
neither of these are fixed as continuously water is percolating downward, or moving upward, to be
evapotranspired. In the context of the SoilCAM project, the image was therefore translated into an
approximate texture field. This field was translated using pedo transfer functions (PTF) into parameter
fields of the Van Genuchten/Mualem hydraulic functions. Introduced in modelling, it proved possible
to simulate the water flow problem and to adjust the parameter statistics until a good agreement is
found. However, the transientness of water flow implies immediately, that to conduct such an analysis
successfully, a time-lapse approach is needed.
For the situation at Gardermoen, the smaller scale sedimentary layers and structures appeared to have
a modest effect on the water flow: due to the large hydraulic conductivity and the poor water retention
of the sandy material, flow was predominantly vertical, with only small horizontal components as
affected by these structures. Therefore, it appeared feasible to adopt a so-called parallel stream tube
model in the modelling of the fate of de-icing chemicals in deliverable [D3.3].
Detection of contaminant biodegradation and transport
If the subsoil structure can be interpreted well from geophysical methods, it may become feasible to
detect also how the mass of contaminants behave, particularly whether or not large quantities are
emitted to the broader environment and whether the contaminant levels decrease by natural causes
such as biodegradation. This, however, requires that measurements are done at different times and that
the subsoil characterization is good enough to appreciate relatively small changes. Commonly, it will
be necessary to confront geophysical measurements with conventional ones.
57
At Trecate, the measured values of TPH and other indicators show the evolution of the contaminant
concentrations in monitoring points and further downstream, measured concentrations of TPH and
other indicators reveal that biodegradation reduces TPH concentrations to low values, at distances less
than 500 m.
Contaminant transport occurs at a relatively small rate in comparison to the existing quantity of crude
oil of the spill. Taking into account measured concentrations (source of data: POLITO), the mass
outflow per year is small through ground water downstream from the contaminated subsurface. Since
the spill event, the cumulated removed volume of hydrocarbons is rather low. Even though, transport
could be detected geophysically, particularly through the effect of degradation products on the water
properties.
For the Trecate site, different remediation strategies have been considered. For instance, the potential
of air sparging to strip the contaminant or aerate the subsoil has been tested. However, in view of the
poor way that this technique can be engineered (it is difficult, if not impossible, te steer where the air
flow moves to, even at small hydraulic contrasts in the subsoil), and the apparently limited risks
associated with the contamination, it is questionable whether such an effort is worthwhile.
58
5. Concluding remarks
In this report, results of different methodologies; conventional invasive, the geophysical non-invasive
experimental approaches and the modelling, are analysed and combined to illustrate how pedotransfer
functions can be used to make quantitative links between one dataset to another.
As the illustrative presentation for the two quite different sites reveal, these three main methodologies
are needed in the integrative approach, because each of these methodologies alone suffers from
significant shortcomings. Stepping away from the detail, these shortcomings are predominantly:
1. conventional invasive approaches are only able to give data sets that reflect properties at different
points in space and time, that require interpolation techniques for obtaining a site comprehensive
interpretation. Since the interpolations are usually very uncertain, so is the site-covering assessment.
2. single geophysical non-invasive approaches may give a site (or transect) covering interpretation of
properties, but usually only at a relative scale: these approaches indicate how different properties are
from a spatiotemporal perspective, but do not reveal the magnitude of the properties, only the changes.
Since the geology and total porosity remains constant, time-lapse measurements reveal relative
changes in geophysical properties, which then are a combination of saturation, electrical conductivity
of the soil water and temperature over time.
3. the modelling is site covering, but is context specific, and if the values of properties from place to
place and through time are not known, the modelling may weigh the various processes incorrectly.
Hence, the three approaches serve as validation for each other, albeit on different aspects. Invasive
methods provide local data on hydrogeological properties and contamination concentration, non-
invasive space integrated data provide knowledge in between point measurements, and time-lapse
measurements provide dynamic changes, models integrate all information and explain dynamic
changes due to changes in boundary conditions. This has the consequence that by the combination of
approaches, the assessment of the current contamination as well as the anticipation of future
developments at these sites can be done in a more justified and therefore more confident way. By
implication, this is also the case for other sites, that are integratively investigated with these three
approaches.
Examples of the added value of the integrated approach are diverse. For both sites, the conceptual
models depended much on the information that was acquired with the different approaches. With
additional measurements and modelling, the conceptual models could be tested and this has resulted in
an interpretation of the contamination (as it is, Trecate, or as it occurs annually, in Gardermoen) in
which we have confidence. Based on this interpretation, we safely predict for Trecate that
contamination of the broader environment is unlikely to be large, if current management is continued.
However, if for instance, rice cultivation, with large seasonal leaching fluxes is discontinued, our
interpretation may help in assessing whether contaminant emissions at Trecate may indeed increase to
hazardous levels. For Gardermoen, the balance between subsoil stratifications as conceived
geophysically, with the material properties found invasively, is crucial for the modelling result that
nevertheless, gravity flow is so dominant, that transversal interactions will remain limited. This gives
confidence, that the one-dimensional modelling of the fate of de-icing chemicals does not ignore
important multidimensional processes. For this reason, the prediction that it will be very difficult, if at
all possible, to apply nitrate for bioremediating de-icing chemicals without creating a groundwater-
nitrate problem, has become better founded. Whether increased degradation can be measured by
geophysical methods, is still uncertain based on the measurements in the SoilCAM project and
required further research.
The integration, as reported here, has resulted in an appreciation of the contamination and its future
development for each of the sites, i.e. they are both more ambitious and better justified. The time lapse
approaches and results are better understood. Better understanding of what the geophysical changes
59
mean and how they are influenced by temperature, water saturation and chemical changes in the pore
water. It is evident, that time-lapse analysis is promising only if changes are sufficiently large during
the measurement intervals. For the Trecate site, with current management, time-lapse analysis may
only be of interest for monitoring remediation processes if the interval is significantly larger than the
SoilCAM project duration. On the other hand larger time resolution may give issues concerning the
lifetime of the elctrodes which was not part of this study. A priori, this could not have been
anticipated, as so many of the processes were unknown, and had not yet been parameterized. By the
same token, the much faster and annually occurring processes of de-icing chemical release to the soil
at Gardermoen, and movement towards groundwater and degradation that occurs meanwhile, may well
be suited for time-lapse monitoring.
60
6. References
Allen, J. P., Atekwana, E. A., Atekwana, E. A., Duris, J. W., Werkema, D. D., Rossbach, S. (2007).
The microbial community structure in petroleum-contaminated sediments corresponds to
geophysical signatures. Applied and environmental microbiology, Vol. 73, No. 9, 2007.
Alvarez, P. J. J., Vogel, T. M. (1991). Substrate interactions of benzene, toluene, and para-xylene
during microbial degradation by pure cultures and mixed culture aquifer slurries. Applied And
Environmental Microbiology, Vol. 57, No. 10, p. 2981-2985, 1991.
Amirbahmana, A., Schönenbergerb, R., Furrerc, G., Zobrist, J. (2003). Experimental study and steady-
state simulation of biogeochemical processes in laboratory columns with aquifer material.
Journal of Contaminant Hydrology 64, p. 169– 190, 2003.
Appelo, C.A.J., Becht, R., Van der Griend, A.A., Spierings, T. C. M. (1983). Buildup of discharge
along the course of a mountain stream deduced from water-quality routings (EC routings).
Journal of Hydrology, 66, p. 305-318, 1983.
Appelo, C.A.J., Drijver, B., Hekkenberg, R., de Jonge, M. (1999). Modeling in situ iron removal from
ground water. Ground Water, Vol. 37, No. 6, November – December 1999.
Archie, G.E. (1942). The electrical resistivity log as an aid in determining some reservoir
characteristics. . Petroleum Development and Technology, Proc. Amer. Inst. Min. Met. Eng.
146, 54–62.
Arthurs, P., Stiver, W. H., Zytner, R. G. (1995). Passive volatilization of gasoline from soil. Journal of
Soil Contamination, 4(2), 1995.
Atekwana, E. A., Werkema, D. D., Duris, Jr., J. W., Rossbach, S., Atekwana, E. A., Sauck, W. A.,
Cassidy, D. P., Means, J., Legall, F. D. (2004). In-situ apparent conductivity measurements and
microbial population distribution at a hydrocarbon-contaminated site. Geophysics, Vol. 69, No.
1; p. 56–63, 2004.
Atekwana, E., Atekwana, E., Rowe, R., Werkema, D., Legall, F. (2004). The relationship of total
dissolved solids measurements to bulk electrical conductivity in an aquifer contaminated with
hydrocarbon. J. Appl. Geophys., 56, 281– 294, 2004.
Balci, O., Ormsby, F. (2007). Conceptual modeling for designing large-scale simulations. Journal of
Simulation, 1, 3, 2007, 175-186.
Bardi, L., Mattei, A., Steffan, S., Marzona, M. (2000). Hydrocarbon degradation by a soil microbial
population with b-cyclodextrin as surfactant to enhance bioavailability. Enzyme and Microbial
Technology 27,p. 709–713, 2000.
Bastani, M., Bloem, E., Malehmir, A., French, H. K., Pedersen, L. B., Godio, A., Wehrer, M., Kamm,
J. (2011). 3D imaging of a contamination zone at Oslo airport from joint interpretation of
surface GPR and electrical resistivity data. Geophysical Research Abstracts Vol. 13, EGU2011-
1523, EGU General Assembly 2011.
Bernard, J. (2003). Short note on the principles of geophysical methods for groundwater
investigations, Definition of main hydrogeological parameters, Electrical methods for
groundwater, Magnetic resonance method for groundwater. www.Terraplus.com, 2003.
Beyer, W. (1964). Zur Bestimmung der Wasserdurchlässigkeit von Kiesen und Sanden aus der
Kornverteilung. Wasserwirtschaft - Wassertechnik (WWT), 165-169.
Binley, A., & Kemna, A. (2005). Electrical Methods. p. 129–156. In Hubbard, S., Rubin, Y. (eds.),
Hydrogeophysics. Springer.
61
Binley, A., Ramirez, A., Daily, W. (1995). Regularised image reconstruction of noisy electrical
resistance tomography data. In: Beck, M.S., Hoyle, B.S., Morris, M.A., Waterfall, R.C.,
Williams, R.A. (Eds.), Process Tomography—1995, Proceedings of the Fourth Workshop of the
European Concerted Action on Process Tomography, Bergen, 6–8 April 1995, pp. 401–410.
Bloem, E., Schotanus, D., French, H. K., Binley, A. (2011). Time-lapse electrical resistivity and multi-
compartment sampler measurements for monitoring flow and transport at Oslo airport,
Gardermoen. Geophysical Research Abstracts Vol. 13, EGU2011-12664, EGU General
Assembly 2011.
Bortolami, C., Braga , G., Colombetti, A., Dal Prà, A., Francani, V., Francavilla, F., Giuliano, G.,
Manfredini, M., Pellegrini, M., Petrucci, F., Pozzi, R., Stefanini, S. (1976). Hydrogeologïcal
features of the Po valley (northern Italy). IAH Redbook. 1976.
Botros, F. E., Harter, T., Onsoy, Y. S., Tuli, A., Hopmans, J. W. (2009). Spatial variability of
hydraulic properties and sediment characteristics in a deep alluvial unsaturated zone. Vadose
Zone Journal, Vol. 8, No.2, p.276–289, 2009.
Brooks, R.J., Tobias, A. M. (1996). Choosing the best model: level of detail, complexity, and model
performance. Math. Comput. Modelling, 24, 4, 1996, p. 1-14.
Caputo, M. C., De Carlo, L., Cassiani, G., Deiana, R. (2011). Electrical methods for monitoring a site
potentially contaminated by landfill leachate. Geophysical Research Abstracts Vol. 13,
EGU2011-13332, EGU General Assembly 2011
Carman, P.C. (1938). Fundamental principles of industrial filtration—a critical review of present
knowledge. Transactions of Institution of Chemical Engineering 16, 168-188.
Carmody, o., Frost, R., Xi, Y., Kokot, S. (2007). Adsorption of hydrocarbons on organo-clays -
implications for oil spill remediation. Journal of Colloid and Interface Science 305(1):pp. 17-24,
2007.
Cary, L., Trolard, F. (2006). Effects of irrigation on geochemical processes in a paddy soil and in
ground waters in Camargue (France). Journal of Geochemical Exploration 88, p. 177–180,
2006.
Cassiani G. and the ModelPROBE Team: Minimally invasive characterization of a hydrocarbon
contaminated site: the Trecate example. Geophysical Research Abstracts Vol. 13, EGU2011-
4795, EGU General Assembly 2011.
Cassiani, G., Godio, A., Arato, A., Sambuelli, L., Stocco, S., French, H. K., Kaestner, A., Binley, A.
M., Kemna, A., Flores, A., Rizzo, E., Deiana, R., Bruno, V., Lapenna, V. (2009). ModelPROBE
and SoilCAM two EU FP7 projects aimed at a minimally invasive characterization of
contaminated sites. GNGTS, Trieste, 2009.
Cassiani, G., Strobbia, C., Gallotti, L. (2004). Vertical radar profiles for the characterization of deep
vadose zones. Vadose Zone Journal, 3, 2004.
Cassidy. D. P., Werkema. D. D., Sauck, Jr. W., Atekwana, E., Rossbach, S., Duris, J. (2001). The
Effects of LNAPL biodegradation products on electrical conductivity measurements. Journal of
Environmental and Engineering Geophysics, Vol.6, Issue 1, p.47-52, 2001.
Catapano, I., Soldovieri, F., Crocco, L. (2011). On the feasibility of the linear sampling method for 3D
GPR surveys. Progress In Electromagnetics Research, Vol. 118, 185-203, 2011.
Chiang, C. Y., Salanitro, J. P., Chai, E. Y., Colthart, J. D., Klein, C. L. (1989). Aerobic biodegradation
of benzene, toluene, and xylene in a sandy aquifer-data analysis and computer modeling.
Ground Water, Vol. 27, No. 6, 1989.
62
Clement, T. P. (1997). A modular computer code for simulating reactive multispecies transport in 3-
dimensional groundwater systems. U.S. Department of Energy, Battelle Memorial Institute,
Pacific Northwest National Laboratory Richland, Washington, 1997.
Clements, L., Palaia, T., Davis, J. (2009). Characterisation of sites impacted by petroleum
hydrocarbons, National guideline document. Cooperative Research Center for Contamination
Assessment and Remediation of the Environment, Technical Report series no. 11, 2009.
Colarieti, M. L., Toscano, G., Scelza, R.A., Rao, M. A., Greco, G. (2011). Biodegradation of
propylene glycol by soil bacteria. Geophysical Research Abstracts Vol. 13, EGU2011-8999,
EGU General Assembly 2011.
Cozzarelli, I. M., Bekins, B. A., Baedecker, M. Jo., Aiken, G. R., Eganhouse, R. P., Tuccillo, M. E.
(2001). Progression of natural attenuation processes at a crude-oil spill site: I. Geochemical
evolution of the plume. Journal of Contaminant Hydrology, Vol.53, p.369– 385, 2001.
Cunningham, C. R. (2004). Biodegradation rates of weathered hydrocarbons in controlled laboratory
microcosms and soil columns simulating natural attenuation field conditions. A Master’s Thesis
Presented to the Faculty of California Polytechnic State University San Luis Obispo, 2004.
Dahan, O., Ronen, Z. (2011). Continuous monitoring of the vadose zone hydraulic and chemical
properties as a tool for optimization of remediation strategies. Geophysical Research Abstracts
Vol. 13, EGU2011-475, EGU General Assembly 2011.
Daniels, J. J. (2000). Ground penetrating radar fundamentals. Appendix to a report to the U.S.EPA,
Region V Nov. 25, 2000
Davis, G. B., Johnston, C.D., Patterson, B.M., Barber, C., Bennett, M. (1998). Estimation of
biodegradation rates using respiration tests during in situ bioremediation of weathered Diesel
NAPL. GWMR, p. 123-132, 1998.
de Broissia, M. (1986). Selected mathematical models in environmental impact assessment in Canada.
Canadian Environmental Assessment Research Council, 1986.
Domonkos, M., Schmidt, B., Libisch, B., Polgári, M., Biró, B. (2010). Growth and mycorrhizal
colonization of four grasses in a MN-amended low quality sandy soil. Research Journal of
Agricultural Science, 42 (4), 2010.
Dreyer, M. G., Nelson, Y. M., Kitts, C. (2005). Weathering effects on biodegradation and toxicity of
hydrocarbons in groundwater. Proceedings of the Eighth International In Situ and On-Site
Bioremediation Symposium ,Baltimore, Maryland; June 6–9, 2005.
Durner, W. (1994). Hydraulic conductivity estimation for soils with heterogeneous pore structure.
Water Resources Research 30, 211-223. 1994.
Elkin, K. (2011). Research fellow; Institute for Plant and Environmental Sciences (Personal
communication, in person).
Essaid, H. I., Bekins, B. A., Herkelrath, W. N. and Delin, G. N. (2011). Crude Oil at the Bemidji Site:
25 Years of Monitoring, Modeling, and Understanding. Ground Water, 49: 706–726. doi:
10.1111/j.1745-6584.2009.00654.x
Essaid, H. I., Cozzarelli, I. M., Eganhouse, R. P., Herkelrath, W. N., Bekins, B. A., Delin, G. N.,
Butler, W. (2003). Inverse modeling of BTEX dissolution and biodegradation at the Bemidji,
MN crude-oil spill site. University of Nebraska – Lincoln, US Geological Survey, US
Geological Survey Staff - Published Research, 2003.
Esser, B., Hudson, B., Moran, J., Beller, H., Carlsen, T., Dooher, B., Krauter, P., Mcnab, W., Madrid,
V., Rice, D., Verce, M. (2002). Nitrate contamination in California groundwater: an integrated
approach to basin assessment and resource protection. U.S. Department of Energy Lawrence
Livermore National Laboratory, 2002.
63
Fank, J. Tracer investigations at the research station “Wagna” (Leibnitzer Feld, Austria) to detect the
role of the unsaturated zone for groundwater protection. Institute of Hydrogeology and
Geothermics, Joanneum Research, Graz, Austria
Fank, J., Rock, G. Tracer experiments on field scale for parameter estimation to calibrate numerical
transport models. Joanneum Research, Institute for Water Resources Management –
Hydrogeology and Geophysics Graz – Austria
Farmani, M. B., Kitterød, N.-O., Keers, H. (2008). Inverse modeling of unsaturated flow parameters
using dynamic geological structure conditioned by GPR tomography. Water Resources
Research, Vol. 44, 2008.
Felleti, F., Bersezio, R., Giudici, M. (2006). Geostatistical simulation and numerical upscaling, to
model ground-water flow in a sandy-gravel, braided river, aquifer analogue. Journal of
Sedimentary Research, Vol. 76, No. 11, November 2006
Fernández-Cirelli, A., Arumí, J. L., Rivera, D., Boochs, P. W. (2009). Environmental effects of
irrigation in arid and semi-arid regions. Chilean Journal Of Agricultural Research 69 (Suppl.
1):27-40, 2009.
Forquet, N. (2009). Two-phase flow modelling of vertical flow filters for wastewater treatment. Ph. D.
Thesis. Strasbourg: University of Strasbourg, ENGEES. 152 pp.
Forquet, N. (2011). Cemagref, Department of Waste Water Treatment (Personal communication, e-
mail).
Francaviglia, R., Capri, E., Klein, M., Hosang, J., Aden, K., Trevisan, M., Errera, G. (2000).
Comparing and evaluating pesticide leaching models: results for the Tor Mancina data set
(Italy). Agricultural Water Management 44, 135-151, 2000.
French, H. K. & Binley, A. (2004). Snowmelt infiltration: monitoring temporal and spatial variability
using time-lapse electrical resistivity. Journal of Hydrology, 297: 174-186.
French, H. K. (1999). Transport and degradation of deicing chemicals in a heterogeneous unsaturated
soil. Ph. D Thesis. Ås: Agricultural University of Norway, Department of Soil and Water
Sciences. 136 pp.
French, H. K., Godio, A., van der Zee, S. E.A.T.M., Wehrer, M., Totsche, K. U., Pedersen, L. B.
Greco, G. (2011). SoilCAM: Soil Contamination, Advanced integrated characterisation and
time-lapse Monitoring, an overview. Geophysical Research Abstracts Vol. 13, EGU2011-6105,
EGU General Assembly 2011.
French, H. K., Hardbattle, C., Bindley, A., Winship, P. & Jakobsen, L. (2002). Monitoring snowmelt
induced unsaturated flow and transport using electrical resistivity tomography. Journal of
Hydrology, 267 (3-4): 273-284.
French, H.K., Van der Zee, S.E.A.T.M., Leijnse, A., 1999.Differences in gravity dominated
unsaturated flow during autumn rains and snowmelt. Hydrological Processes 13 (17), 2783–
2800.
Frick, C.M., Farrell, R.E., Germida, J.J. (1999). Assessment of phytoremediation as an in-situ
technique for cleaning oil-contaminated sites. Petroleum Technology Alliance of Canada
(PTAC) Calgary, AB, 1999.
Gallardo, L. A., Meju, M. A. (2003). Characterization of heterogeneous near-surface materials by joint
2D inversion of dc resistivity and seismic data. Geophysical Research Letters, Vol. 30, No. 13,
1658, 2003.
Gawel, L. J. A guide for remediation of salt/hydrocarbon impacted soil. North Dakota Industrial
Commission Department of Mineral Resources Bismarck, ND 58505-0840
64
Gelhar, L. W., Welty, C., Rehfeldt, K. R. (1992). A critical review of data on field-scale dispersion in
aquifers. Water Resources Research, Vol. 28, No. 7, p. 1955-1974, 1992.
Geuzaine, C., Remacle, J.-F. (2009). Gmsh: A 3-D finite element mesh generator with built-in pre-
and post-processing facilities. International Journal for Numerical Methods in Engineering 79,
1309–1331. Available at http://doi.wiley.com/10.1002/nme.2579 (verified 4 December 2012).
Giampaolo, V., Lopez, F., Maineult, A., Rizzo, E., Votta, M., Lapenna, V. (2011). Self-potential
technique in the study of contamination of Trecate (Italy) and Zeitz (Germany) test sites.
Geophysical Research Abstracts Vol. 13, EGU2011-4527, EGU General Assembly 2011.
Glover, W. J. (2010). A generalized Archie’s law for n phases. Geophysics,Vol. 75, No. 6; p. E247–
E265, 2010.
Godio, A., Borsic, A., Arato, A., Sambuelli, L. (2011). On the inversion of cross-hole resistivity data.
Geophysical Research Abstracts Vol. 13, EGU2011-7950, EGU General Assembly 2011.
Gribb, M.M., Forkutsa, I., Hansen, A., Chandler, D.G., McNamara, J.P. (2009). The Effect of Various
Soil Hydraulic Property Estimates on Soil Moisture Simulations. Vadose Zone Journal 8, 321-
331.
Günzel, F. (1994). Die geoelektrische Untersuchung von Grundwasserkontaminationen unter
Berücksichtigung des Einflusses von Ton- und Wassergehalt auf die elektrische Leitfähigkeit
des Untergrundes.
Görlach, B., Interwies, E. (2003). Economic assessment of groundwater protection: a survey of the
literature, Final report. European Commission, 2003.
Hacini, Y., Martinez-Pagan, P., Chapellier, D., Aracil, E. (2006). Petrophysic and mineralogical
characterization of a perialpine gravel aquifer using geophysical logging methods, Kappelen test
site, Switzerland. 5a Asamblea Hispano-Portuguesa de Geodesia y Geofisica, Sevilla 2006.
Halihan, T., Puckette, J., Sample, M., Riley, M (2009). Electrical resistivity imaging of the Arbuckle-
Simpson aquifer. Oklahoma State University, 2009
Harter, T. (2003). Groundwater quality and groundwater pollution. ANR Publication 8084, University
of California, 2003.
Hayashi, M. (2004). Temperature-electrical conductivity relation of water for environmental
monitoring and geophysical data inversion. Environmental Monitoring and Assessment 96:
119–128, Kluwer Academic Publishers, 2004.
Hayley, K., Bentley, L. R., Gharibi, M. & Nightingale, M. (2007). Low temperature dependence of
electrical resistivity: Implications for near surface geophysical monitoring. Geophysical
Research Letters, 34: L18402.
Hayley, K., Bentley, L. R., Pidlisecky, A. (2010). Compensating for temperature variations in time-
lapse electrical resistivity difference imaging. Geophysics, 75 (4): WA51-WA59.
Hazen, A. (1892). Some physical properties of sands and gravels, with special reference to their use in
filtration, 24th Annual Report, Massachusetts State Board of Health, Publication Document, pp.
539-556.
Heinz, J., Aigner, T. (2003). Hierarchical dynamic stratigraphy in various quaternary gravel deposits,
Rhine glacier area (SW Germany): implications for hydrostratigraphy. Int J Earth Sci (Geol
Rundsch) 92:923–938, 2003.
Hess, K. M., Wolf, S. H., Cella, M. A., Garabedian, S.P. (1991). Macrodispersion and spatial
variability of hydraulic conductivity in a sand and gravel aquifer, Cape Cod, Massachusetts. US
EPA Environmental Research Brief, 1991.
65
Holch, J. (2008). Thermodynamic and kinetic degradation reactions of organic substances in
groundwater modeled with PHREEQC. Institut für Hydrologie Albert-Ludwigs Universität
Freiburg, 2008.
Holden, P.A., Halverson, L.J., Firestone, M.K. (1997). Water stress effects on toluene biodegradation
by Pseudomonas putida. Biodegradation, 8, 1997, 143-151. (from Holden 2005)
Holden, P. A., Fierer, N. (2005). Microbial processes in the vadose zone. Vadose Zone Journal, 4,
2005, 1-21.
Howe, A. (2000). Ground Penetrating Radar for the parameterisation of subsurface hydrological
properties. A thesis submitted to the University of London for the degree of Doctor of
Philosophy, Department of Geography King’s College London, September 2000.
Hunt, J. R., Holden, P.A., Firestone, M.K. (2005). Coupling transport and biodegradation of VOCs in
surface and subsurface soils. Environ. Health Perspect. 103, 1995, 75-78. (from Holden 2005)
Ipek, G. (2002). Log-derived cation exchange capacity of shaly sands: application to hydrocarbon
detection and drilling optimization. Louisiana State University, Department of Petroleum
Engineering, 2002.
Janssen, G. M. C. M., Cirpka, O. A., van der Zee, S. E. A. T. M. (2006). Stochastic analysis of
nonlinear biodegradation in regimes controlled by both chromatographic and dispersive mixing.
Water resources research, vol. 42, W01417, doi:10.1029/2005WR004042, 2006.
Janssen, G. M. C. M., Valstar, J. R., van der Zee, S. E. A. T. M. (2006). Inverse modeling of
multimodal conductivity distributions. Water resources research, vol. 42, W03410,
doi:10.1029/2005 WR004356, 2006.
Jin, G., Torres-Verdín, C., Devarajan, S., Toumelin, E., Thomas, E. C. (2007). Pore-scale analysis of
the Waxman-Smits shaly-sand conductivity model. Petrophysics, Vol. 48, No. 2; p. 104–120,
2007.
John, A. K. (2006). Dispersion in large scale permeable media. The Department of Petroleum and
Geosystems Engineering The University of Texas at Austin, 2006
Karvonen, T. (2002). Biodegradation/Biotransformation. Department of Civil and Environmental
Engineering, Helsinki University of Technology, 2002.
Katz, I., Ronen, Z., Adar E., Dahan, O. (2011). Enhanced in situ biodegradation of perchlorate in the
vadose zone. Geophysical Research Abstracts Vol. 13, EGU2011-10651, EGU General
Assembly 2011.
Kemna, A., Vanderborght, J., Kulessa, B., Vereecken, H. (2002). Imaging and characterisation of
subsurface solute transport using electrical resistivity tomography (ERT) and equivalent
transport models. Elsevier Journal of Hydrology 267 125–146, 2002.
Khalil1, M. A., Monterio Santos, F. A. (2011). Influence of degree of saturation in the electric
resistivity-hydraulic conductivity relationship. Developments in Hydraulic Conductivity
Research. INTECH, 2011. DOI: 10.5772/15667
Kieft, T.L., Murphy, E.M., Haldeman, D.L., Amy, P.S., Bjornstad, B.N., Donald, E.V. Mc, D.B.,
Ringelberg, D.C., White, J., Stair, R.P., Griffiths, T.C., Gsell, H.W.E., Boone, D.R. (1998).
Microbial transport, survival, and succession in a sequence of buried sediments. Microb. Ecol.
36:336–348, 1998.
Kitterød, N. O. (2008). Focused flow in the unsaturated zone after surface ponding of snowmelt. Cold
Regions Science and Technology, 53: 42-55.
66
Klipa, V., Sobotkova, M., Snehota, M. (2011). Evaluation of TDR water content measurements in
large undisturbed soil sample. Geophysical Research Abstracts ,Vol. 13, EGU2011-8348, EGU
General Assembly 2011.
Knight, R. (2001). Ground penetrating radar for environmental applications. Annual Reviews Earth
Planet. Sci. 29, 2001.
Koestel, J., Kemna, A., Javaux, M., Binley, A., Vereecken, H. (2008). Quantitative imaging of solute
transport in an unsaturated and undisturbed soil monolith with 3-D ERT and TDR. Water
Resources Research 44.
Konopka, A., Turco, R. (1991). Biodegradation of organic compounds in vadose zone and aquifer
sediments. Appl. Environ. Microbiol. 57:2260–2268, 1991.
Kotiadis, K., Robinson, S. (2008). Conceptual modelling: knowledge acquisition and model
abstraction. Proceedings of the 2008 Winter Simulation Conference (eds. S.J. Mason, R.R. Hill,
L. Monch, O. Rose, T. Jefferson, JW. Fowler).
Kozeny, J. (1927). Über kapillare Leitung des Wassers im Boden, Wien.
Labrecque, D.J., & Yang, X. (2001). Difference Inversion of ERT Data : a Fast Inversion Method for
3-D In Situ Monitoring. Journal of Environmental and Engineering Geophysics 6, 83–89.
Langenhoff, A. A. M., Brouwers-Ceiler, D. L., Engelberting, J. H. L., Quist, J. J., Wolkenfelt, J. G. P.
N., Zehnder, A. J. B., Schraa, G. (1997). Microbial reduction of manganese coupled to toluene
oxidation. FEMS Microbiology Ecology, Volume 22, Issue 2, p. 119–127, 1997.
Lee, M. W. (2011). Connectivity equation and shaly-sand correction for electrical resistivity. U.S.
Department of the Interior U.S. Geological Survey, Scientific Investigations Report 2011–5005
Lensing, H.J., Herrling, B. (1994). Simulation of the redox sequence of an infiltration passage by
direct numerical modelling of the mediating microorganisms. Hydrological, Chemical and
Biological Processes of Transformation and Transport of Contaminants in Aquatic
Environments (Proceedings of the Rostov-on-Don Symposium, May 1993). IAHS Publ. No.
219, 1994.
Li, H. (2002). Geostatistical shale models for a deltaic reservoir analog: from 3D GPR data to 3D flow
modeling. A Thesis submitted to the Graduate Faculty of the Louisiana State University and
Agriculture and Mechanical College, 2002.
Li, K. (2010). Determination of resistivity index, capillary pressure, and relative permeability.
Proceedings, Thirty-Fifth Workshop on Geothermal Reservoir Engineering Stanford University,
Stanford, California, February 1-3, 2010.
Libisch, B., French, H. K., Hartnik, T., Anton, A., Biró, B. (2011). Laboratory-scale evaluation of
selected remediation techniques for propylene glycol-based de-icing fluids. Geophysical
Research Abstracts Vol. 13, EGU2011-11127, EGU General Assembly 2011.
Light, T. S., Licht, S., Bevilacqua, A. C., Morash, K. R. (2005). The fundamental conductivity and
resistivity of water. Electrochemical and Solid-State Letters, 8 (1) E16-E19, 2005.
Lißner, H., Wehrer, M., Bloem, E., Totsche, K. U., (2011). Soil heterogeneity strongly affects fate and
transport of deicing chemicals. Geophysical Research Abstracts Vol. 13, EGU2011-7529-1,
EGU General Assembly 2011
Louis, I. F., Karantonis, G. A., Voulgaris, N. S., Louis, F. I. (2004). The contribution of geophysical
methods in the determination of aquifer parameters: the case of Mornos River delta, Greece.
International Journal of Electrical and Electronics Engineering
Lovatini, A.,Umbach, K., Patmore, S. (2009). 3D CSEM inversion in a frontier basin offshore West
Greenland. EAGE First brake, Vol. 27, 2009.
67
Lucius, J. E., Powers, M. H. (2002). GPR Data Processing Computer Software for the PC. U.S.
Department of the Interior, U.S. Geological Survey, 2002.
Manheim, F. T., Krantz, D. E., Bratton, J. F. (2004). Studying ground water under Delmarva coastal
bays using electrical resistivity. Ground water—Oceans Issue Vol. 42, No. 7, p. 1052–1068,
2004.
Mathers, S., Zalasiewicz, J. (1994). A guide to the sedimentology of unconsolidated sedimentary
aquifers (UNSAs). British Geological Survey, 1994.
McFarland, M.J., Sims, R.C. (1991). Thermodynamic framework for evaluating PAH degradation in
the subsurface. Ground Water, Vol. 29, No. 6, November – December 1991.
Mirus, B. B., Perkins, K. S., Nimmo, J. R., Singha, K. (2009). Hydrologic Characterization of Desert
Soils with Varying Degrees of Pedogenesis: 2. Inverse Modeling for Effective Properties.
Vadose Zone Journal, Vol.8, No2, p.496–509, 2009.
Mohanty, B.P., Bowman, R.S., Hendrickx, J.M.H., van Genuchten, M.T. (1997). New piecewise-
continuous hydraulic functions for modeling preferential flow in an intermittent-flood-irrigated
field. Water Resources Research 33, 2049-2063.
Montevechi, J.A.B., da Silva Costa, R.F., Leal, F., de Pinho, A.F., Marins, F.A.S., Marins, F.F., de
Jesus, J.T. (2008). Combined use of modeling techniques for the development of the conceptual
model in simulation projects. Proceedings of the 2008 Winter Simulation Conference (eds. S.J.
Mason, R.R. Hill, L. Monch, O. Rose, T. Jefferson, J.W. Fowler), 2008.
Morin, R. H., LeBlanc, D. R., Troutman, B. M. (2010). The influence of topology on hydraulic
conductivity in a sand-and-gravel aquifer.–Ground water Vol. 48, No. 2, p. 181–190, 2010.
Morin, R. H. (2005). Negative correlation between porosity and hydraulic conductivity in sand-and-
gravel aquifers at Cape Cod, Massachusetts, USA. US Geological Survey USGS Staff Published
Research, University of Nebraska – Lincoln, 2005.
Murphy, E.M., Ginn, T.R. (2000). Modeling microbial processes in porous media. Hydrogeol. J. 8,
2000, 142-158. (from Holden 2005)
Nachtergaele, F., van Velthuizen, H., Verelst, L. (2009). Harmonized World Soil Database Version
1.1, March 2009.
Naudet, V., Revil, A., Bottero, J.-Y. (2003). Relationship between self-potential (SP) signals and
redox conditions in contaminated groundwater. Geophysical Research Letters, Vol. 30, No. 21,
2003.
Naudet, V., Revil, A., Rizzo, E., Bottero, J. Y., Begassat, P. (2004). Ground water redox conditions
and conductivity in a contaminant plume from geoelectrical investigations. Hydrology and
Earth System Sciences 8 (1), 8-22, 2004.
Nichols, E. M., Roth, T. L. (2006). Downward solute plume migration: assessment, significance, and
implications for characterization and monitoring of “diving plumes”. Regulatory Analysis and
Scientific Affairs, API soil and groundwater technical task force Bulletin 24, 2006.
Niwas, S., de Lima, O.A.L. (2003). Aquifer parameter estimation from surface resistivity data.
Groundwater, Vol. 41, No. 1, p. 94-99, 2003.
O’Reilly, K. T., Magaw, R. I., Rixey, W. G. (2001). Predicting the effect of hydrocarbon and
hydrocarbon-impacted soil on groundwater. American Petroleum Institute, Soil and
Groundwater Technical Task Force, No. 14, 2001.
Orozco, A. F., Oberdörster, C., Zschornack, L., Leven, C., Weiss, H., Kemna, A. (2011). Delineation
of BTEX contamination plume with SIP imaging. Geophysical Research Abstracts Vol. 13,
EGU2011-10263, EGU General Assembly 2011.
68
Pasteris, G., Werner, D., Kaufmann, K., Hohener, P. (2002). Vapor phase transport and biodegradation
of volatile fuel compounds in the unsaturated zone: a large scale lysemeter experiment. Environ.
Sci. Technol., 36, 2002, 30-39. (from Holden 2005)
Pathak, B.K., Iida, T., Kazama, F. (2007). Denitrification as a component of nitrogen budget in a
tropical paddy field. Global NEST Journal, Vol 9, No 2, pp 159-165, 2007.
Pedersen, T. S. (1994). Væsketransport i umettet sone. Stratigrafisk beskrivelse av toppsedimentene på
forskningsfeltet, Moreppen og bestemmelse av tilhørende hydrauliske parametre del 1. Master
Thesis. Oslo: University of Oslo. 122 pp.
Pidlisecky, A., Haber, E. Knight, R. (2007). RESINVM3D: A 3D resistivity inversion package.
Geophysics,Vol. 72, No. 2, 2007.
Prommer, H., Davis, G. B., Barry, D. A. (2000). Biogeochemical transport modelling of natural and
enhanced remediation processes in aquifers. Land Contamination & Reclamation, 8 (3), 2000.
Rathfelder, K.M., Lang, J.R., Abriola, L.M. (2000). A numerical model (MISER) for the simulation of
coupled physical, chemical and biological processes in soil vapor extraction and bioventing
systems. J. Contam. Hydrol. 43, 2000, 239-270. (from Holden 2005)
Regberg, A., Singha, K., Tien, M., Picardal, F., Zheng, Q., Schieber, J., Roden, E., Brantley, S.L.
(2011). Electrical conductivity as an indicator of iron reduction rates in abiotic and biotic
systems. Water Resources Research, Vol. 47, 2011.
Remy, N. (2004). Geostatistical Earth Modeling Software: User’s Manual, May 2004
Revil, A. (2013). Effective conductivity and permittivity of unsaturated porous materials in the
frequency range 1 mHz - 1GHz. Water Resources Research submitted.
Rinaudo, J.-D., Göerlach, B., Loubier, S., Interwies, E. (2003). Economic assessment of groundwater
protection, Executive summary. European Commission, 2003.
Robinson, S. (2008). Conceptual modelling for simulation, Part I: definition and requirements. Journal
of the Operational Research Society, 59, 2008, p. 278-290.
Robinson, S. (2008). Conceptual modelling for simulation, Part II: a framework for conceptual
modelling. Journal of the Operational Research Society, 59, 2008, p. 291-304.
Rolle, M., Prabhakar Clement, T., Sethi, R., Di Molfetta, A. (2008). A kinetic approach for simulating
redox-controlled fringe and core biodegradation processes in groundwater: model development
and application to a landfill site in Piedmont, Italy. Hydrological Processes, 22, 4905-4921,
2008.
Rose, S., Long, A. (1988). Monitoring dissolved oxygen in ground water: some basic consideration.
GWMR, 1988.
Ross, G. J., Wang, C. (1982). Lepidocrocite in a calcareous, well-drained soil. Clays and Clay
Minerals, Vol. 30, No. 5, 394-396, 1982.
Samouelian, A., Cousin, I., Tabbagh, A., Bruand, A., Richard, G. (2005). Electrical resistivity survey
in soil science: a review. Soil and Tillage Research. Volume 83, Issue 2, September 2005, Pages
173–193.
Schaap, M.G., Leij, F.J., van Genuchten, M.T. (1998). Neural network analysis for hierarchical
prediction of soil hydraulic properties. Soil Science Society of America Journal 62, 847-855.
Schaap, M.G., Leij, F.J., van Genuchten, M.T. (2001). ROSETTA: a computer program for estimating
soil hydraulic parameters with hierarchical pedotransfer functions. Journal of Hydrology 251,
163-176.
69
Schotanus, D., van der Zee, S. E.A.T.M., Bloem, E., Eggen, G. (2011). Measurement of contaminant
fluxes in the soil during snowmelt, at high spatial resolution. Geophysical Research Abstracts
Vol. 13, EGU2011-3503, EGU General Assembly 2011.
Schulte, E.E., Kelling, K. A. (1999). Soil and applied manganese. Cooperative Extension Publications,
University of Wisconsin, 1999.
Schurig, C., Miltner, A., Zschornack, L., Kaestner, M. (2011). Extending the in situ microcosm
approach (BACTRAP®) to field sites without groundwater wells – a new Direct-Push probe.
Geophysical Research Abstracts Vol. 13, EGU2011-434, EGU General Assembly 2011.
Sen, P.N., Goode, P.A., & Sibbit, A. (1988). Electrical-Conduction in Clay Bearing Sandstones at
Low and High Salinities. Journal of Applied Physics 63, 4832–4840.
Serrano, S. (1997). The Theis solution in heterogeneous aquifers. Ground water, Vol. 35, No. 3, 1997.
Shtivelman, V.,Gendler, M., Goldberg, I. (2011). 3D geological model of potential CO2 reservoir for
the Heletz test site. Geophysical Institute of Israel, POB 182, LOD 71100. Geophysical
Research Abstracts. Vol. 13, EGU2011-1307-1, 2011
Simunek, J., Suarez, D. L. (1994). Two-dimensional transport model for variably saturated porous
media with major ion chemistry. Water Resources Research, Vol. 30, No. 4, p. 1115-1133,
1994.
Simunek, J., van Genuchten, M. Th., Sejna, M. (2011). The HYDRUS software package for
simulating the two- and three-dimensional movement of water heat, and multiple solutes in
variably-saturated media, Technical manual Version 2.0. PC Progress, Prague, Czech Republic,
258 pp., 2011.
Slater, L., Binley, A., Daily, W., Johnson, R. (2000). Cross-hole electrical imaging of a controlled
saline tracer injection. Journal of Applied Geophysics 44, 85–102.
Slater, L.D., Day-Lewis, F. D., Ntarlagiannis, D., O’Brien, M., Yee, N. (2009). Geoelectrical
measurement and modeling of biogeochemical breakthrough behavior during microbial activity,
Geophys. Res. Lett., 36, 2009.
Smiciklas, K. D., Moore, A. S., Adams, J. C. (2008). Fertilizer nitrogen practices and nitrate levels in
surface water within an Illinois watershed. Journal Of Natural Resources & Life Sciences
Education, Vol. 37, 2008.
Søiland, A. (2011). Geophysical monitoring of degradable de-icing chemicals in the unsaturated zone
during snowmelt. Norwegian University of Life Sciences, Department of Plant and
Environmental Sciences, Master Thesis, 2011.
Somerfield, P. J. (2008). Identification of the Bray-Curtis similarity index: Comment on Yoshioka
(2008). Marine Ecology Progress Series, Vol. 372, p. 303–306, 2008.
Stumpp, C., Engelhardt, S., Hofmann, M., Huwe, B. (2009). Evaluation of pedotransfer functions for
estimating soil hydraulic properties of prevalent soils in a catchment of the Bavarian Alps.
European Journal of Forest Research 128, 609-620.
Stuut, R.-J., Groenewoud, P. (2011). Integration of innovative soil characterisation technologies High
Resolution Seismic imaging (HRS) and Rapid Optical Screening Tool (ROST-CPT / Membrane
Interface Probe (MIP-CPT). Geophysical Research Abstracts Vol. 13, EGU2011-3309-3, EGU
General Assembly 2011.
Susset, B,. Grathwohl, P. Numerical and analytical modelling of organic leaching in column tests: "a
priori" prediction of release rates based on material properties. University of Tübingen,
Department of Applied Geology
70
Travis, B.J., Rosenberg, N.D. (1997). Modeling in situ bioremediation of TCE at Savannah River:
effects of product toxicity and microbial interactions on TCE degradation. Environ. Sci.
Technol. 31, 1997, 3093-3102. (from Holden 2005)
Tuller, M., Nearing, G. S., Jones, S. B., Heinse, R. (2011). Geophysical characterization of inactive
mine tailings - a first step for revegetation. Geophysical Research Abstracts Vol. 13, EGU2011-
5751-1, EGU General Assembly 2011.
Vance, D. B. (1994). Iron – the environmental impact of a universal element. National Environmental
Journal, Vol. 4, No. 3, May/June 1994.
van Genuchten, M.T. (1980). A Closed-Form Equation for Predicting the Hydraulic Conductivity of
Unsaturated Soils. Soil Science Society of America Journal 44, 892-898.
van de Putte, W., Van Herreweghe, S., Vansina, W., Van Straaten, M. (2011). Enhanced In Situ Soil
Analysis (EnISSA) of volatiles and semi-volatile components. Geophysical Research Abstracts
Vol. 13, EGU2011-7230, EGU General Assembly 2011.
van Epps, A. (2006). Phytoremediation of petroleum hydrocarbons. U.S. Environmental Protection
Agency Office of Solid Waste and Emergency Response Office of Superfund Remediation and
Technology Innovation Washington, DC, 2006.
van Houten, M., Eddies, R., Koomans, R. (2011). Development of a geophysical method for
quantitative risk assessment. Innovative methods for the quantification of industrial by-products
in roads. Geophysical Research Abstracts Vol. 13, EGU2011-3732-2, EGU General Assembly
2011.
van Vliet, M., Kok, K., Veldkamp, T. (2009). Linking stakeholders and modellers in scenario studies:
The use of Fuzzy Cognitive Maps as a communication and learning tool. JFTR, 2009.
Vereecken, H., Weynants, M., Javaux, M., Pachepsky, Y., Schaap, M.G., Genuchten, M.T.v. (2010).
Using Pedotransfer Functions to Estimate the van Genuchten-Mualem Soil Hydraulic
Properties: A Review. Vadose Zone Journal 9, 795-820.
Wahyudi, A., Bartzke, M., Bogaert, P., Küster, E., Altenburger, R. (2011). Implementation of
probability table for qualitative reconstruction of pollutants plume using biological assay data.
Geophysical Research Abstracts Vol. 13, EGU2011-2175-1, EGU General Assembly 2011.
Watts, R. J. (2006). Improved understanding of Fenton-like reactions for the in situ remediation of
contaminated groundwater including treatment of sorbed contaminants and destruction of
DNAPLs. Washington State University, 2006.
Waxman, M.H., Smits, L.J.M. (1968). Electrical conductivities in oil-bearing shaly sands.
Transactions, Vol. 243, 1968.
Wehrer, M. (2012). Communication on How to integrate holistically techniques on organically
contaminanted sites – the hydrogeochemical point of view. E-mail, 2012.
Weidemeier, T., Swanson, M., Wilson, J., Kampbell, D., Miller, R. , Hansen, J. (1996).
Approximation of biodegradation rate constants for monoaromatic hydrocarbons (BTEX) in
ground water. University of Nebraska – Lincoln, U.S. Environmental Protection Agency Papers,
1996.
Werkema Jr., D. D., Atekwana, E. A., Endres, A., Sauck, W. A. (2004). Temporal and spatial
variability of high resolution in situ vertical apparent resistivity measurements at a LNAPL
impacted site US EPA: How to evaluate alternative cleanup technologies for underground
storage tank sites – A guide for corrective action plan reviewers. May 2004.
Wierenga, P. J. (1977). Solute distibution profiles coupled with steady-state and transient water
movement models. Soil Science Society of America Proceedings, 41: 1050-1054.
71
Winship, P., Binley, A. & Gomez, D. (2006). Flow and transport in the unsaturated Sherwood
Sandstone: characterization using cross-borehole geophysical methods. In Barker, R. D. &
Tellam, J. H. (eds) vol. 263 Fluid flow and solute movement in sandstones: the onshore UK
Permo-Triassic red bed sequence, pp. 219-231. London: Geological Society of London.
Yeh, T. C. J., Liu, S., Glass, R. J., Baker, K., Brainard, J. R., Alumbaugh, D. & Labrecque, D. J.
(2002). A geostatistically based inverse model for electrical resistivity survey and its
applications to vadose hydrology. Water Resources Research, 38 (12): 1278-1291.
Zappa, G., Bersezio, R., Felleti, F., Giudici, M. (2006). Modeling heterogeneity of gravel-sand,
braided stream, alluvial aquifers at the facies scale. Journal of hydrology Vol. 325, no. 1-4,
2006.
Zavattaro, L., Romani, M., Sacco, D., Bassanino, M., Grignani, C. (2006). Fertilization management
of paddy fields in Piedmont (NW Italy) and its effects on the soil and water quality. Paddy
Water Environ., 4: 61–66, 2006.
van der Zee, D.-J., Tolk, A., Pidd, M., Kotiadis, K., Tako, A. A. (2011). Education on conceptual
modeling for simulation – Beyond the craft: a summary of a recent expert panel discussion. SCS
M&S Magazine, 2, 2011.
US EPA (2011): Conductivity, Monitoring and assessment, September 2011.
US EPA (2009): Petroleum brownfields: selecting a reuse option, October 2009.
US EPA (1997): Chapter III Surface Geophysical Methods.www-epa-gov-oust-pubs-esa-ch3, March
1997.
US EPA (2009): Guidance on the development, evaluation, and application of environmental models.
Council for Regulatory Environmental Modeling (CREM), March 2009.