Testing The Computational Method (tvdlf) For Simulating A Wide Range of Surface and Sub-Surface Flow Types In Large Watersheds.

Since 1980s, government agencies have been using several software tools to assess the water supply, flood control and environmental needs in complex watersheds. These tools considered the basic water balance within the surface and sub-surface domains. The South Florida Water Management Model (SFWMM) included several physical processes such as the surface and groundwater flow, along with the rules recorded in water control manuals. The RSM (Regional Simulation Model) extended these ideas further, creating more accurate surface and sub-surface flow algorithms and a fully integrated solution scheme that allowed the integrated model to run for multiple decades. Surface water models developed during the earlier period assumed the diffusive wave approximation.

The TVDLF (Total Variation Diminishing Lax-Friedrichs) version of the model was initiated later because the earlier versions of RSM suffered from problems of oscillation when simulating steeper landscapes. A deeper look into the problem showed that practically all the physically-based hydrologic models including MIKE-SHE/MIKE-11, GSSHA, RAS-2D have the same problem of not being able to simulate kinematic flow and fully dynamic canal flow with Froude numbers reaching critical levels. The TVDLF version of RSM was developed to solve this problem and add solution methods to solve both diffusive, kinematic, and mixed flow types under steep and flat surface and subsurface conditions. These solutions were tested using many field-scale seepage experiments, hydraulic studies, and wetland experiments in south Florida and in Sri Lanka.

The USACE and the SFWMD are carrying out a number of test applications of the RSM model using TVDLF algorithms in Mendocino valley along Russian River, California, Boggy Creek, Florida and Colombo Sri Lanka. The goal of this applications is to move beyond 1980s version of legacy models based on diffusive-wave based computational methods and represent hillslope and wetland processes using better hyperbolic-parabolic partial differential equation solvers into hydrologic models. The results highlight detailed descriptions of kinematic flow trails that are not found in diffusive-wave solutions of sheet flow.