Petascale Hydrologic Modeling - Needs and Challenges

In recent decades, computational schemes have been developed to model point-scale processes using physics-based or conceptual approaches. The integration of these processes across space-time has been limited by computational power to either high-resolution over small spatial domains, or coarse resolution over large spatial domains. These modeling approaches have lead to improved understanding at both small and large scales, but have required parameterization of important phenomenon, and the corresponding lack of model sensitivity to changes and uncertainties in parameter values. The CI-WATER project aims to develop a peta-scale modeling approach to simultaneously allow simulation of high-resolution processes such as snow melt, vegetation succession and runoff generation at the scale of an individual hillslope, while integrating over large areas using an unstructured grid. The CI-WATER High Performance Computing (HPC) model will include needed physics-based process modules and parameters to simulate future change with a minimum of conceptualization. Significant challenges exist, however. The ultimate objective of the model development is to simulate large watersheds such as the Upper Colorado River above Lake Powell (287,000 km2). This will require generation of an unstructured mesh with over 107 nodes, coupling with remote sensing, modeled, and/or space-time interpolated forcing data, linkage of multi-physics flow processes, and parameter assignment and estimation using a reasoning engine. Optimization of the model mesh, visualization of inputs and and model outputs as well as debugging in the HPC environment represent further challenges. The CI-WATER HPC model architecture and initial tools are discussed, as well as development activities for the next project year. 1Dept. of Civil & Architectural Engineering, Univ. of Wyoming 2Dept. of Mathematics, Univ. of Wyoming 3Dept. of Ecosystem Science and Management, Univ. of Wyomin