Numerical Innovations for Improved Computational Efficiency in the Simulation of Basin-Wide Integrated Surface-Subsurface Flow: Testing of Sub-gridding and Sub-timing Methods

*Don DeMarco1, George Matanga2, Dua Guvanasen3 and Mary Kang1,4
1HydroGeoLogic, Inc., Kitchener, Ontario, Canada, 2U.S. Bureau of Reclamation, Sacramento, California, U.S.A., 3HydroGeoLogic, Inc., Reston, Virginia, U.S.A., 4now at Princeton University, Princeton, New Jersey, U.S.A.

Abstract

Sub-gridding and sub-timing techniques are a recent innovation in numerical modeling.  The benefit of these advanced methods lies in their ability to provide computational advantage for simulation of challenging hydrologic problems.  These methods are particularly adept for problems that require simulation across a wide range of spatial and/or temporal scales; therefore regional integrated surface-subsurface flow problems, among others, are well suited for sub-gridding and sub-timing application. 

Integrated surface-subsurface flow and transport simulations typically can be fraught with temporal over-discretization problems because the characteristic response times for stream flow, overland runoff and subsurface flow can differ by orders of magnitude.  Moreover, the local and regional scale aspects of integrated problems present spatial discretization challenges.  The sub-time stepping approach can apply different time step sizes to the sub-domains that have different response times.  The sub-gridding approach permits high resolution detailing in localized sub-regions of the model without the computational burden of extending the refinement to the model boundaries, whether vertical or horizontal. 

Sub-gridding and sub-timing methods were tested using the HydroGeoSphere model and three cases: two local-scale models adopted from the Smith-Wolhiser (1971) and Abdul (1985) experiments; and a regional-scale, simplified integrated surface-subsurface model of the San Joaquin Valley and Tulare Basin, California.

Results indicate the sub-timing approach can significantly improve the solution efficiency for solving integrated surface-subsurface problems. Additional efficiency can be achieved by dynamically adjusting the sub-timed nodes and the number of sub-time steps over the simulation period for a given problem.  With respect to sub-gridding, the method can achieve localized high resolution results at a computational savings, compared to cases of more extensive mesh refinement.  Sub-gridding is most suitable when the regions to be sub-gridded are localized portions of the overall domain.
*Point of Contact: Don DeMarco, Senior Hydrogeologist, HydroGeoLogic, 301-684 Belmont Avenue West, Kitchener, Ontario, Canada.  Phone: (519) 884-9868 FAX: (519) 747-3959 email: ddemarco@hgl.com

Biographical Sketch: 

Mr. DeMarco has more than 14 years of project management, environmental investigation and engineering experience, and has served as the Regional Office Manager of HGL’s Waterloo office since 2001.  Mr. DeMarco’s technical expertise includes the application of advanced numerical and analytical models of hydrological processes, including surface and subsurface flow and contaminant transport, the transport and fate of organic compounds in groundwater, multiphase flow, geostatistics, stochastic analyses of subsurface flow and transport, and field characterization techniques.  He has developed conceptual, stochastic, and deterministic numerical models of porous and fractured aquifer systems; and evaluated transport behavior and fate and bioremediation of radionuclides, metals, chlorinated solvents and BTEX contaminants in soil and groundwater systems.  He has been registered as a Professional Geoscientist in Ontario since 2002.