| Surface water, plant water, soil and groundwater, and the atmosphere are all linked components of the hydrologic continuum. Understanding the interaction between these components and the ability to predict the availability, variability and quality on large scale, requires an accurate and efficient solution strategy. Here we present the underpinnings of a framework referred to as Hydrologic Modeling System (HMS), to couple physics, numerics, data and computation, with the goal to simulate coupled hydrologic process interactions at multiple spatio-temporal scales. The primary component of the framework is a physics-based, spatially distributed, fully coupled, constrained unstructured mesh based Finite-Volume model that simultaneously solves integrated hydrologic processes in heterogeneous, anisotropic domains. The holistic approach developed here, emphasizes the need for efficient simulations through spatially adaptive domain decomposition strategies, use of multi-processor clusters, and seamless and dynamic flow of data between data-management systems and hydrologic models. The modeling framework has been applied from hillslope (10-100m) to catchment (100-1000m) to synoptic scales (>100km) by using different number and approximation of process equations depending on model purpose and computational constraint. Examples will demonstrate how this model provides insight into the influence of drainage from unsaturated zone on delayed water table drawdown, the role of water table position on infiltration and surface runoff, and the interaction of overland flow-groundwater exchanges in relation to the dynamics of infiltrating/exfiltrating surfaces on the hillslopes. Large scale implementation of the model in Little-Juniata Watershed (845 km2) unfolds a range of multiscale/multiprocess interactions including the influence of local upland topography and stream morphology on spatially distributed, asymmetric right-left bank river-aquifer interactions, and, the role of macropore and topography on ground water recharge magnitude, time scale and spatial distribution. Finally, the computational challenges posed by using such complex model will be addressed, along with an outlook for future efforts along these lines. |
