THE ROLE OF MEASUREMENT / MODEL COMPARISON FOR IMPROVING THE UTILITY OF COUPLED SURFACE WATER / GROUNDWATER MODELS

A variety of comprehensive hydrologic models coupling surface and variably-saturated subsurface flow are currently being used for many exciting research problems in hydrology. Model testing and benchmarking different codes against one another has provided valuable insights into issues surrounding model discretization, computational efficiency, and contrasting numerical methods. A major obstacle to widespread application of all these models for informing land and water resources management decisions remains the limited availability of data to parameterize and evaluate them effectively. In particular, extensive hydrologic-response datasets including distributed measurements of soil-water content, shallow water-table fluctuations, evapotranspiration, streamflow, and exchange fluxes, are needed to fully test their capability for realistic simulations of surface water / groundwater fluxes. This work outlines the types of input data typically available for parameterization of rainfall-runoff models before presenting applications of the Integrated Hydrology Model (InHM) for several small, well-characterized experimental catchments. Measurement / model comparisons for a range of environmental conditions illustrate the degree to which the InHM simulations can and cannot capture the complexities of real world hydrologic response. Models calibrated with hydrologic observations also provide a foundation for investigating how input parameters that represent subsurface hydraulic properties, heterogeneity, and topography can influence simulated streamflow generation processes. For example, continuous simulations for a small rangeland catchment demonstrate the impact of contrasting methods for assigning soil-hydraulic properties on the long-term water balance. Overall, the results highlight the value of developing consistent methods for integrating available input data into model parameterization and employing distributed data in model performance assessments.