Shallow aquifer characterization using direct push technologies
There is great potential in the use of direct push technologies for aquifer characterization. Geotechnical direct push techniques exist already for a long time, but the use of e.g. cone penetration tests to estimate aquifer hydraulic conductivity and condition groundwater flow models is very limited. With proper calibration data, they however represent an important source of information, and allow for efficient regional aquifer characterization campaigns (an example is provided in Fig. 1). Recent developments in hydraulic direct push technologies allow to characterize the aquifer properties hydraulically, and form the ideal calibration data for regional geotechnical direct push campaigns, omitting the need for borehole drilling and coring. Fig. 2 shows an example dataset from hydraulic direct push characterization campaign.
|Figure 1: Hydraulic conductivity estimates (red = high K; blue = low K) for a ~60 km² x 40 m deep geotechnical direct push (CPT) campaign.|
- Rogiers B, Gedeon M, Mallants D, Batelaan O, Huysmans M, Dassargues A. 2013. Building confidence in contaminant transport modelling through the integration of multiple data sources and explicit representation of geological heterogeneity. 15th Annual Conference of the International Association for Mathematical Geosciences, IAMG 2013, Madrid, Spain, 2-6 September, 2013. Conference
- Rogiers B, Vienken T, Batelaan O, Gedeon M, Mallants D, Huysmans M, Dassargues A. 2013. Multi-scale aquifer characterization and groundwater flow model parameterization using direct push technologies. Novel Methods for Subsurface Characterization and Monitoring: From Theory to Practice. NovCare 2013, Leipzig, Germany, 13-16 May, 2013. Presentation - Conference
- Vienken T, Tinter M, Rogiers B, Leven C, Dietrich P. 2012. Evaluation of field methods for vertical high resolution aquifer characterization. Abstract ID 1494473, AGU Fall Meeting, San Francisco, USA, 3-7 December 2012. Abstract - Poster - Conference
Optimization and inversion of random fields and global parameters in groundwater flow modelling
As numerical groundwater flow models are often highly parameterized, the optimization or inversion of such models is a challenging task. Moreover, accounting for complex subsurface heterogeneity makes parameter reduction difficult and the optimization or inversion procedures even more complex. Efficient algorithms and methodologies should be developed to tackle this problem. Fig. 3 shows a simple example where McMC sampling is used to obtain the posterior parameter distributions from a groundwater flow model. The evolution of the 2D marginal distribution of hydraulic conductivity of a certain aquifer layer and the groundwater recharge, from a Markov chain, is shown.
|Figure 3: Evolution of a Markov chain and the corresponding posterior density in two-dimensional space. The example shows clearly the correlation between hydraulic conductivity (K) of a shallow aquifer and recharge in a groundwater flow model.|
- Laloy E, Rogiers B, Vrugt JA, Mallants D, Jacques D. 2013. Efficient posterior exploration of a high-dimensional groundwater model from two-stage MCMC simulation and polynomial chaos expansion. Water Resources Research 49(5): 2664–2682. Paper
- Rogiers B, Mallants D, Batelaan O, Gedeon M, Huysmans M, Dassargues A. 2012. The usefulness of CPTs for deterministic spatially heterogeneous, large-scale aquitard parameterisation. In: Oswald, S.E., Kolditz, O., Attinger, S. (Eds.), Models - Repositories of Knowledge, IAHS Publ. 355: 41-47. Proceedings ModelCare 2011, Leipzig, Germany, 18-22 September 2011. - ISBN 978-1-907161-34-6. Proceedings - Abstract - Conference - Book of abstracts
Machine learning in geosciences
When data types related to the hydrogeological parameters of interest are available, machine learning techniques should be used to quantify the relationship between the variables as accurately as possible if a training dataset is available. An example of such techniques are neural networks (Fig. 4). Without the available parameters of interest, unsupervised learning might be another option to enhance the interpretation of certain data or make data classifications. Fig. 5 shows the result of clustering a large CPT dataset, revealing different lithostratigraphical units in the subsurface.
|Figure 5: Example results of site-specific soil behaviour type classification obtained by model-based clustering.|
- Rogiers B, Mallants D, Batelaan O, Gedeon M, Huysmans M, Dassargues A. 2012. Estimation of hydraulic conductivity and its uncertainty from grain-size data using GLUE and artificial neural networks. Mathematical Geosciences 44(6): 739–763. Paper - Author's version - R script - Blog post
- Rogiers B, Mallants D, Batelaan O, Gedeon M, Huysmans M, Dassargues A. 2011. Site-specific soil classification from cone penetration tests and borehole data: a multivariate statistical analysis. NovCare 2011, Cape Cod, 9-11 May 2011. Presentation - Conference
Estimating aquifer properties from outcrop analogues
In the framework of risk assessment studies for waste disposal or contaminated sites, the small-scale heterogeneity within the sediments adjacent to such disposal systems, or subject to the migration of contaminant plumes is of high importance. Detailed quantification of such small-scale spatial variabilty is however not straightforward. Core drilling for sample collection and analysis remains time-consuming and expensive, and for detailed characterization of horizontal spatial variability, many such boreholes are required. For shallow aquifers, the recent developments in direct-push technologies present a more efficient alternative, but characterization of small-scale heterogeneity remains labour-intensive.
An alternative for the in situ or laboratory characterization of subsurface aquifer sediments is the use of outcrops as easily accessible analogues for studying subsurface sediments. This is especially useful for the characterization of small-scale spatial variability, as this can be observed directly in small outcrops. The availability of cost-effective in situ measurement techniques for important hydrogeological parameters makes the study of outcrop sediments even more attractive.
|Figure 6: Temporary Poederlee Sands outcrop during tunnel excavation (http://goo.gl/Mo747G).|
- Rogiers B, Beerten K, Smeekens T, Mallants D, Gedeon M, Huysmans M, Batelaan O, Dassargues A. 2013. The usefulness of outcrop analogue air permeameter measurements for analysing aquifer heterogeneity: testing outcrop hydrogeological parameters with independent borehole data. Hydrol. Earth Syst. Sci. 17: 5155-5166. Paper (OA) - Discussion paper
- Rogiers B, Beerten K, Smeekens T, Mallants D, Gedeon M, Huysmans M, Batelaan O, Dassargues A. 2013. Derivation of flow and transport parameters from outcropping sediments of the Neogene aquifer, Belgium. Geologica Belgica 16(3): 129-147. Paper (OA)
- Rogiers B, Beerten K, Smeekens T, Mallants D, Gedeon M, Huysmans M, Batelaan O, Dassargues A. 2013. The usefulness of outcrop analogue air permeameter measurements for analysing aquifer heterogeneity: Quantifying outcrop hydraulic conductivity and its spatial variability. Hydrological processes. Paper
Heat as a groundwater tracer
Heat is more and more being used as a groundwater tracer. Mainly in the surface- and ground-water interaction research it is currently heavily used, but also in shallow aquifers or large-scale flow systems it can yield useful results.
|Figure 7: Example of hydraulic head profiles (equated to topography) resulting in large-scale groundwater flow and cooling of the subsurface (decrease in measured heat-flow density) down to 1.5 km deep.|
Use of geostatistics in radiological characterization
For optimizing radiological characterization and decontamination of building structures in the framework of nuclear facility decommissioning, the use of geostatistics can help in designing sampling plans, integrating different types of measurements or performing risk assessment. Properly accounting for spatial variability using geostatistical algorithms has the potential to reduce both workloads and waste volumes associated with radiological characterization and decontamination considerably, and therefore forms an important line of research.
|Figure 8: Change of support from circular (measurements) to the point scale for stochastic simulation, to a square at a practical resolution.|