Estimation of hydraulic conductivity and its uncertainty from grain-size data using GLUE and artificial neural networks

in Mathematical Geosciences (2012), 44(6), 739-763

Various approaches exist to relate saturated hydraulic conductivity (Ks) to grain-size data. Most methods use a single grain-size parameter and hence omit the information encompassed by the entire grain ... [more ▼]

Various approaches exist to relate saturated hydraulic conductivity (Ks) to grain-size data. Most methods use a single grain-size parameter and hence omit the information encompassed by the entire grain-size distribution. This study compares two data-driven modelling methods, i.e.multiple linear regression and artificial neural networks, that use the entire grain-size distribution data as input for Ks prediction. Besides the predictive capacity of the methods, the uncertainty associated with the model predictions is also evaluated, since such information is important for stochastic groundwater flow and contaminant transport modelling. Artificial neural networks (ANNs) are combined with a generalized likelihood uncertainty estimation (GLUE) approach to predict Ks from grain-size data. The resulting GLUE-ANN hydraulic conductivity predictions and associated uncertainty estimates are compared with those obtained from the multiple linear regression models by a leave-one-out cross-validation. The GLUE-ANN ensemble prediction proved to be slightly better than multiple linear regression. The prediction uncertainty, however, was reduced by half an order of magnitude on average, and decreased at most by an order of magnitude. This demonstrates that the proposed method outperforms classical data-driven modelling techniques. Moreover, a comparison with methods from literature demonstrates the importance of site specific calibration. The dataset used for this purpose originates mainly from unconsolidated sandy sediments of the Neogene aquifer, northern Belgium. The proposed predictive models are developed for 173 grain-size -Ks pairs. Finally, an application with the optimized models is presented for a borehole lacking Ks data. [less ▲]

Geostatistical analysis of primary and secondary data in a sandy aquifer at Mol/Dessel (Belgium)

in Cokx, L.; Van Meirvenne, M.; Bogaert, P. (Eds.) et al 8th International Conference on Geostatistics for Environmental Applications (GeoENV2010) (2010, September)

In the framework of the disposal of short-lived low- and intermediate-level radioactive waste in a near-surface disposal facility in Dessel, Belgium, additional extensive site characterization has been ... [more ▼]

In the framework of the disposal of short-lived low- and intermediate-level radioactive waste in a near-surface disposal facility in Dessel, Belgium, additional extensive site characterization has been performed in 2008. The gathered data now enclose 388 hydraulic conductivity measurements on samples of 8 cored boreholes. Secondary information as grain size analysis, porosity, and borehole geophysical parameters was also gathered. In addition, the geology of the study area has also been thoroughly characterized by a set of 178 cone penetration tests (CPTs) to approximate 50 m depth. This dataset allowed to refine the hydrostratigraphical model of the region. The existing groundwater model, based on large-scale effective hydraulic properties, was updated accordingly. The next step is a small-scale probabilistic approach 1) to validate the current existing deterministic groundwater models and 2) to support design for a monitoring network. In preparation for stochastic realizations of the subsurface, a geostatistical analysis of the available primary and secondary data is performed. [less ▲]