Laboratory and field studies of preferential flow dynamics in unsaturated fractured porous media

Abstract

<p>Recharge dynamics within the vadose zone (variable saturation conditions) of consolidated fractured rock formations are an ongoing challenge when it comes to process understanding and predictive modeling. The proper delineation of fast (macropores, fractures, conduits) and slow (matrix) flow components in these systems and their interaction with each other remains a complex puzzle and holds a key to enhance process-based infiltration models.</p><p>We conducted laboratory and field experiments to study infiltration dynamics through porous-fractured systems. Laboratory experiments were carried out with analogue fracture networks on meter scale. Orthogonal networks were created by placing equally sized blocks with a constant gap between to glass plates, which were mount by metal clamps. Vertical flow through different network configurations (apertures, intersection types, topology, flow rates) was studied for (1) porous media (sandstone) and (2) non-porous media (glass) to delineate the control of network features on flow dynamics, as well as the effect of fracture-matrix interaction. Matrix imbibition was found to strongly control the preferential flow velocity during flow path evolution. Higher infiltration rates lead to more by-pass at fracture intersections, whereas low infiltration rates favor flow partitioning into horizontal fractures. Vertical flow progression within the non-porous network is significantly faster due to the lack of imbibition. Semi-analytical tools, such as transfer functions, and source-responsive dual-domain models are tested to reproduce the experimental data and to incorporate key features of fracture networks in future modeling approaches. We additionally obtained experimental data from infiltration dynamics at porous-fractured field sites on meter scale to compare them to the well-controlled laboratory experiments and to evaluate the applicability of the results to actual field processes.</p>