PosterPDF Available

Chalk Aquifer Management (CHARM): Groundwater modelling of a complex aquifer for the sustainable management of strategic drinking water reserves in Flanders (Belgium)

Authors:

Abstract

Groundwater is an important source of drinking water in Belgium. In some regions, locally produced groundwater is the source of 100% of the drinking water supply. One of the most important aquifers in the eastern part of Belgium is the Chalk Aquifer. This aquifer is phreatic in the northern part of Wallonia, but dips down towards the north, in Flanders, where it quickly reaches large depths. This largely confined aquifer is of strategic societal importance because it is well protected against negative influences from the surface (nitrates, pesticides) on the water quality. However, geological and hydrogeological information is scarce leading to important uncertainties regarding sustainable yields. Due to the large depth of the aquifer in its confined part, relatively little borehole information is available. Furthermore, the Chalk Aquifer is characterized by a double porosity system which results in a strong heterogeneity and spatial variability of the hydrogeological properties. The goal of the CHARM project is to analyse the capacity of the Chalk Aquifer on a regional scale, and to deliver a management instrument that can be used for decision-making with regards to the quantitative use of this strategic aquifer for drinking water purposes. Special attention will be given to the characterization of all sources of uncertainty and its incorporation in a groundwater flow model. In a first step, the geology and hydrogeological parameters of the Chalk aquifer are characterized in detail. Based on gamma-ray logs, flow measurements, pumping tests and literature data, the horizontal and vertical variability of the hydrogeological parameters are identified. Exploitation results are linked to geological and hydrogeological data providing insights why some exploitations have a higher yield than others. Next, a regional groundwater model (MODFLOW) is set-up. The integrated Bayesian multi-model approach of Mustafa et al. (2018) is adapted, so that input, parameter and conceptual model uncertainty can be quantified. This is done by coupling the MODFLOW model with the DiffeRential Evolution Adaptive Metropolis (DREAM) algorithm (Vrugt, 2016) and by applying Bayesian Model Averaging (BMA). Based on the results of this approach, well-founded decisions can be made regarding the quantitative use of this aquifer considering all different sources of uncertainty, which is of strategic importance for long-term drinking water purposes.
Chalk Aquifer Management (CHARM):
Groundwater modelling of a complex aquifer for the sustainable management
of strategic drinking water reserves in Flanders (Belgium)
Gert Ghysels1,*, Syed Mustafa1, Simon Six2, Alexander Vandenbohede2, Tom Diez2, Gijsbert Cirkel3 & Marijke Huysmans1
1Department of Hydrology and Hydraulic Engineering, Vrije Universiteit Brussel, Brussels, Belgium; 2De Watergroep, Brussels, Belgium;
3KWR Watercycle Research Institute, Nieuwegein, The Netherlands
1. Introduction & Study Area 2. Problem Statement & Objectives
Due to several reasons, the uncertainty regarding the Chalk
Aquifer is large:
3. Link between well yield and geology
Two main groups of flow measurements are observed:
Left: only flow in 2-3m interval correlated with hardground
(phosphatic gravel), rest of filter does not contribute
Right: flow more evenly distributed over filter due to presence of
fractures
5. Uncertainty Analysis
4. Spatial variability of hydraulic conductivity
A correlation between hydraulic conductivity and depth of the
Cretaceous sediments is observed: high K when shallow, low K when
buried deep
6. Outlook
0 2.5 5 7.5 10 km
Paleozoic
Cretaceous
Fm. of Hannut
Fm. of Kortrijk
Fm. of Diest
Fm. of Brussels
Fm. of Heers
SN
0
50
100
150
-50
mTAW
0
50
100
150
-50
-100
-150
-200
-250
-300
-350
-400
-450
-500
Modified from Department LNE, ALBON (2008)
0 25 50 75 100 km
Lux.
The Netherlands
Germany
France
CHARM
Study Area
Cretaceous
Due to the relatively large depth of the aquifer in the northern
part, borehole data is limited
Strong spatial variability of hydraulic properties due to double
porosity system: primary matrix vs. secondary fracture porosity
Slow response of the aquifer on (historical) changes: information
on past extractions needed to explain trends in hydraulic head
Regional effect of extraction: need for larger-scale groundwater
models
*Corresponding author. Email: gert.ghysels@vub.be
The main goals of the CHARM project are:
To analyse the capacity of the Chalk Aquifer on a regional
scale
To deliver a management instrument that can be used for
decision-making regarding the use of this aquifer for drinking
water purposes
Special attention will be given to the characterization of all
sources of uncertainty and its incorporation in groundwater
flow models
5 0 5 10 15 20 km
<1 m/d
1-10m/d
10-100 m/d
>100 m/d
0 5 10k
m
K (m/d)
0.01 0.1 110 100
Primary permeability of Cretaceous deposits is low, but in the South
increased secondary permeability is observed due to:
Groundwater models of the Chalk Aquifer are set up on different scales
Application of the integrated Bayesian multi-model approach to
quantify input, parameter and conceptual model uncertainty of
Mustafa et al. (2018) in which MODFLOW is coupled with DREAM (Vrugt,
2016) and Bayesian Model Averaging is applied to assess total prediction
uncertainty
Uniform prior distribution Inferred posterior distribution
Recharge
multipliers
Abstraction
multipliers
Model
parameters
DREAM
MODFLOW
Simulation
Observation
MODFLOW
Output
Prediction uncertainty
Heteroscedastic
error-model
parameters
Fracturing of deposits due to decompression following erosion
of the overburden
Increased fluid circulation and chemical weathering in river
valleys
Groundwater is an important source of drinking water in Belgium.
In some regions, locally produced groundwater is the source of 100% of
the drinking water supply.
The Chalk Aquifer is one of the most important aquifers in the
Flanders Region. It is of strategic societal importance because it is
well protected against negative influences from the surface
(nitrates, pesticides) on the water quality.
A full Bayesian approach is computationally expensive. Alternative
methods will be explored and compared, e.g.:
Approximate Bayesian Computation (ABC) (Vrugt et al., 2013)
Bayesian Evidential Learning (BEL) (Scheidt et al., 2018; Hermans
et al., 2018)
Different scenarios will be compared by using a decision model to
convert the results of the uncertainty analysis to concrete strategies.
Based on this approach, well-founded decisions can be made regarding
the quantitative use of the Chalk aquifer for drinking water
purposes.
References:
Hermans et al. 2018. Uncertainty Quantification of Medium-Term Heat Storage From Short-Term Geophysical Experiments using Bayesian Evidential Learning. Water Resources Research.
Mustafa et al. 2018. Estimation and Impact Assessment of Input and Parameter Uncertainty in Predicting Groundwater Flow With a Fully Distributed Model. Water Resources Research.
Scheidt et al. 2018. Quantifying Uncertainty in Subsurface Systems. American Geophysical Union.
Vrugt et al. 2013. Toward diagnostic model calibration and evaluation: Approximate Bayesian computation. Water Resources Research.
Vrugt, 2016. Markov chain Monte Carlo simulation using the DREAM software package: Theory, concepts, and MATLAB implementation. Environmental Modelling and Software.
Yield per meter (m³/h/m)
Depth (m-ground level)
Yield
GR
0 5 10 150 20 40
25
30
35
40
45
65
70
75
80
85
90
95
100
Yield per meter (m³/h/m)
Depth (m-ground level)
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.