Aquitards are common hydrogeological features in the subsurface. Typically, pumping tests are used to parameterize the hydraulic conductivity of heterogeneous aquitards. However, they do not take spatial variability and uncertainty into account. Alternatively, core-scale measurements of hydraulic conductivity are used in geostatistical upscaling methods, for which their correlation lengths are needed, but this information is extremely difficult to obtain. This study investigates whether a pumping test can be used to obtain the correlation lengths needed for geostatistical upscaling and account for the uncertainty about heterogeneous aquitard conductivity. Random realizations are generated from core-scale data with varying correlation lengths and inserted into a groundwater flow model which simulates the outcome of an actual pumping test. The realizations yielded a better fit to the pumping test data than the traditional pumping test result, assuming homogeneous layers are selected. Ranges of horizontal and vertical correlation lengths that fit the pumping-test well are found. However, considerable uncertainty regarding the correlation lengths remains, which should be considered when parameterizing a regional groundwater flow model.

Aquitards are common hydrogeological features and their hydraulic conductivity is an important property for various groundwater management issues. Predicting their hydraulic conductivity proves challenging, given its dependence on numerous variables. In this study, the dominant factors for predicting aquitard hydraulic conductivity are identified. To this end, a random forest model is trained on a dataset consisting of more than 1000 hydraulic conductivity measurements of core-scale sediment samples from a wide range of stratigraphic units and depths in the Netherlands. The dataset contains textural properties, such as the grain size distribution and porosity, as well as structural data, such as location, sampling depth, stratigraphical unit, lithofacies, organic carbon content, carbonate content and groundwater chloride concentration. Results show that clay fraction, stratigraphic unit, depth, lithofacies and x-coordinate are the most important features for predicting the hydraulic conductivity. Here, x-coordinate is presumably a proxy for distance from marine influence. Using a more detailed grain size distribution or using derived parameters such as the grain size percentiles does not improve the model any further. Our findings indicate that structural properties play a significant role in predicting aquitard conductivity, as they serve as indicators of processes such as compaction and soft-sediment deformation. The model is furthermore an effective method to estimate hydraulic conductivity for sediment samples without conducting costly and time-consuming hydraulic conductivity measurements.