Analysis of three-dimensional versus two-dimensional bedform-induced hyporheic exchange

Abstract:

The hyporheic zone is a critical ecotone for maintaining the health of river systems due to the exchange of water and nutrients between streams and groundwater. Within riverbeds, hyporheic exchange is generally forced by variation in riverbed topography such as due to bedforms. In the past, a vast majority of research on bedform-driven hyporheic flow has focused on two-dimensional (2D) scenarios, while little has been done on more realistic three-dimensional (3D) situations. We investigated hyporheic exchanged using a sinuous-crested 3D dune which is superimposed with successive crest lines of 2D dunes, and compared it to a 2D dune with similar wavelength and height. These 2D and 3D dunes are depicted in detail in McLean (1997) and Maddux (2003), respectively. A series of modeling studies are conducted both in 2D and 3D with similar open channel Reynolds numbers (Re). Turbulent flow in the water column is simulated by solving the Reynolds-averaged Navier-Stokes equations with the k-ω turbulence closure model, and a Darcy flow model is applied for the underlying porous media. These two sets of equations are coupled via the pressure distribution on the sediment-water interface (SWI). Results show that the pressure gradient along the SWI is highly controlled by the spatial structure of bedforms, which consequently determines flow dynamics in the porous media. Hyporheic flux is a function of Re for both 2D and 3D via a power-law trend; however, the hyporheic flux in the 3D dunes is generally higher and much more sensitive to Re. The depth and volume of the interfacial exchange zone of the 3D-bedform driven flow are only slightly different from the 2D situation, showing that the dimensionality of bedform has less impact on the exchanged zone. The mean fluid residence times for both 3D/2D dunes are related to Re by an inverse-power law relationship, they are different at low Re and become similar at higher Re. A 2D idealization seems a reasonable approximation for the more complex 3D situation.