Skip to main content

PhD Defense | Alberto Jesus Casillas | Assessment of unconventional state variables to constrain groundwater flow models of the Neogene Aquifer in the Campine Basin in Belgium

02 September '22

Name: Alberto Jesus Casillas

September 2nd, 2022
11h00 - 13h30 (TBC)

Auditorium Valère Billiet
Campus Sterre, Building S8
Faculty of Sciences
Krijgslaan 281
9000 Gent

Livestream: click here

Assessment of unconventional state variables to constrain groundwater flow models of the Neogene Aquifer in the Campine Basin in Belgium

Numerical models are generally used to represent the physical and/or chemical properties of groundwater systems, and to be implemented in the decision making process from groundwater management supply to safety and feasibility of waste disposal systems. An aquifer can be quite complex where these properties, hydrologic features and hydraulic characteristics are required to faithfully capture the aquifer state. As several of these items are difficult to measure they are generally assumed or inferred by inverse methods. In the general practice, traditional methods only use groundwater hydraulic heads for inverse conditioning. However, the information content in groundwater hydraulic heads is limited and, therefore, the complexity of the groundwater system is unsuccessfully captured. Nevertheless, other types of observations of the state of the aquifer (state variables) can nowadays be measured in more reliable and straightforward ways. As these other types of observations are generally underrepresented in the general practice of groundwater modelling, these are regarded as unconventional as their challenging application differs from the traditional methods.


In this work, the use of unconventional state variables is explored aiming to improve the current conceptual models and reduce parameters and model outcomes uncertainties for groundwater models of the Neogene aquifer, across the Nete Catchment, in the Campine Basin in Belgium. These studies are performed in the region in context of safety and feasibility of potential disposal systems for radioactive waste.


Field campaigns were performed for the gathering of these unconventional state variables across the study domain: temperature-depth (TD) profiles, hydrochemistry, stable isotopes (δ18O, δ2H, δ13C), and age tracers (14C, 3H, 3He, 4He, 4Herad). Additionally, historical data of these state variables was also gathered and included in this work.


First, a 3D groundwater model was constructed based on a previous model for the Neogene aquifer. The extension of the hydrologic features (i.e. drains, rivers) and several structural updates were performed together with the inclusion of faults using the horizontal flow barrier MODFLOW package, spatially distributed recharge input, and a larger number of hydraulic head observations. Here, we tested different fault configurations in particular for the Rauw Fault were a groundwater level step was observed on the top of this fault. In this way, we accounted for a more realistic representation of the subsurface features which can have an important effect on the groundwater head and flux distribution. This groundwater flow model was then used for the development of subsequent transport models.


Second, a 3D heat-transport model is constructed following the updated groundwater flow model. Satellite land-surfac0e temperature (LST) images were imposed as model top boundary conditions, monthly, for the last 19 years. To have a better representation of the temperature top boundary condition and more accurate initial conditions, a paleotemperature-time series was derived for the region from several Holocene temperature reconstructions for Europe and the Northern Hemisphere. This curve was used in combination with the spatially distributed LST to model the temperature variations during the Holocene. The coupled groundwater flow and heat-transport model was conditioned with hydraulic heads and TD profiles to calibrate hydraulic conductivity. The results indicate that the use of TD profiles presents opportunities as improvement is brought by the implementation of TD profiles are additional constrain on the performance of both groundwater flow and heat-transport models. This suggest that meaningful information on important processes not being accounted by the use hydraulic heads only (e.g. groundwater fluxes) can be partially characterized by the inclusion of TD profiles.


Third, the use of age tracers and hydrochemistry is here used together with the groundwater model ages derived from advective particle tracking modelling. The initial activity for 14C was corrected following a correction model previously constructed for the Neogene aquifer. Then, the 14C model ages were estimated, and interpreted in combination with 3H/3He model ages and 4He concentrations to delineate groundwater flow systems. The estimated age tracer apparent ages were then compared with the advective particle tracking model ages to support the conceptual interpretation of the groundwater flow systems distribution. The results suggest i) local, shallow and young (0-100 years) groundwater flow system, ii) an a semi-regional, deep, older (1000-35000 years) groundwater flow system, with iii) an intermediate groundwater flow system (100 up to 1000 years) being is likely a mixture between fractions of old and young groundwater.


Finally, a 3D 4Herad production and transport model is constructed following groundwater flow model structure. Several 4Herad production sources are included being in situ from the hydrogeological formations and crustal flux diffusion into the aquifer. The production rate is calculated from uranium (U) and thorium (Th) observations. The production rate, transport parameters, and groundwater flow model parameters were inferred following a Bayesian approach (i.e. DREAMzs). The coupled groundwater flow and 4Herad transport model inversion was conditioned with hydraulic head and 4Herad  observations. An inversion of only the groundwater flow model was also performed to be able to compare the worth of 4Herad  as additional model conditioning state variable in relation to the solely use of hydraulic head. The results demonstrate the uncertainty in model parameters hence, derived groundwater fluxes can be reduced by the inclusion of 4Herad  as additional unconventional observation to constrain numerical groundwater models.


To conclude, this study shows that the implementation of additional unconventional state variable observations exhibit strong potential for conditioning groundwater flow models. Depending on the application (e.g. shallow areas or deep into the aquifer) these state variables present robust solutions to reduce model parameter and derived fluxes uncertainties.




  • Kristine Walraevens (UGENT)

SCK CEN mentors:

  • Bart Rogiers (SCK CEN)

  • Koen Beerten (SCK CEN)


  • Laurent Wouters (NIRAS/ONDRAF)


Click here for a list of obtained PhD degrees.

Share this news article