Coupling surface flow and subsurface flow in complex soil structures using mimetic finite differences

• Published in: Advances in water resources 2020-10, Vol.144 (C), p.103701, Article 103701
• Main Authors: Coon, Ethan T., Moulton, J. David, Kikinzon, Evgeny, Berndt, Markus, Manzini, Gianmarco, Garimella, Rao, Lipnikov, Konstantin, Painter, Scott L.
• Format: Article
• Language: English
• Subjects: Complex stratigraphy; ENVIRONMENTAL SCIENCES; Integrated hydrology; Mimetic finite differences
• ISSN: 0309-1708; 1872-9657
• DOI: 10.1016/j.advwatres.2020.103701

•Model surface and subsurface hydrology using mimetic finite differences.•The discretization allows for accurate solutions on complex soil structures.•A new scheme uses a discrete approach to couple surface and subsurface flow.•The method is demonstrated on benchmarks and patchy ground cover. We explore the coupling of surface and subsurface flows on fully unstructured meshes that conform to complex soil structures. To accommodate the distorted meshes that inevitably result from explicit representation of complex soil structures, we leverage the structure of the Mimetic Finite Difference (MFD) spatial discretization scheme to couple surface and subsurface flows. The MFD method achieves second-order accuracy and maintains local mass conservation on distorted meshes. We couple the diffusion wave approximation for surface flows to the Richards equation for subsurface flow, ensuring continuity of both pressure and flux between the surface and subsurface. The MFD method is particularly convenient for this coupling because it uses face-based constraints in the subsurface system that can be expressed as face-pressure unknowns. Those unknowns are coincident with surface cell-based unknowns, thus allowing the discrete surface system to be directly substituted into the subsurface system and solved implicitly as a global system. Robust representation of the transition between wet and dry surface conditions requires upwinding of the relative permeability and is facilitated by globalization in the nonlinear solver. The approach and its implementation in the Advanced Terrestrial Simulator (ATS) are evaluated by comparison to previously published benchmarks. Using runoff from soils with patchy groundcover (duff) as an example, we show that the new method converges significantly faster in mesh convergence tests than the commonly used two-point flux approximation.