Lessons learned from simulations of hillslope hydrologic response with a physics-based model (Invited)

Friday, 26 September 2014: 1:15 PM
Luisa Hopp1, Peter A A Troch2, Luke A Pangle2, Stephen DeLong3, Travis E Huxman4 and April Lynda James5, (1)University of Bayreuth, Bayreuth, Germany, (2)University of Arizona, Tucson, AZ, United States, (3)US Geological Survey, Menlo Park, CA, United States, (4)University of California, Irvine, Irvine, CA, United States, (5)Nipissing University, Department of Geography, North Bay, ON, Canada
Abstract:
The appeal of physics-based models lies in their promise to describe distributed hydrologic response at different scales in an uncalibrated manner if material properties, geometries and boundary conditions are known. Their successful application is often severely hampered, however, as detailed spatial information required to represent the 3D flow domain of interest is hardly ever available beyond the plot scale. Thus, also physics-based models are subject to simplifying assumptions and the use of effective parameters. Designing rigorous test scenarios is difficult because physics-based models possess many ways to be adjusted, e.g. material and process parameters, implemented boundary conditions, numerical solution strategies, and we are faced with the problem of equifinality, even within the bounds of plausibility.

Here we present our modeling experiences with three hillslopes that differ in their complexity, the extent of known properties and boundary conditions and the availability of data describing hydrologic response. The studied hillslopes are the Panola hillslope (Georgia, USA), a constructed hillslope as part of a cover system over waste rock (SE Alaska, USA) and the experimental hillslopes built within the Biosphere 2 Landscape Evolution Observatory (LEO) project. The hillslopes range from natural under ambient atmospheric conditions to constructed under ambient atmospheric conditions, to constructed with fully-controlled boundary conditions. The hydrologic response of the three hillslopes to various natural and artificial water inputs was simulated with a physics-based model. We discuss the encountered challenges in defining and parameterizing the modeling domains, along with some shortcomings and workarounds we applied and the options we had for testing if the model "was working for the right reasons". Due to the experimental control and high density of geometric, material and state variable data, the LEO hillslopes in particular provide a unique test bed for addressing questions that relate to uncalibrated modeling, equifinality and rigorous testing of the model structure. As such hydrologic models are more frequently being used as a foundation of understanding diverse landscape processes and making predictions under changing conditions, their rigorous evaluation with common data sets is essential and key to building effective interdisciplinary knowledge.

Previous Abstract | Next Abstract >>