A linked simulation-optimization model for simultaneously estimating the Manning's surface roughness values and their parameter structures in shallow water flows

Abstract

A linked simulation-optimization model is proposed for simultaneously estimating the Manning's surface roughness parameters and their associated parameter structures for one-dimensional shallow water flows. In the simulation part of the model, hydrodynamic flow process is simulated by modeling the given flow reach on HEC-RAS. The association of unknown parameter structure with the roughness values is accomplished by partitioning the given flow reach into sub-regions using the one dimensional Voronoi Diagram (VD). The developed simulation model is then linked to an optimization model where heuristic Particle Swarm Optimization algorithm is used. The main objective of the PSO based optimization model is to determine the roughness parameters and their associated parameter structures along the flow reach by minimizing the discrepancies between simulated and measured water elevations at several observation locations. The applicability of the model is evaluated on a flow reach of East Fork River, WY, USA under three scenarios by considering the known/unknown roughness parameter structures and unsteady flow conditions. Furthermore, a sensitivity analysis is conducted to determine the influence of different PSO parameters on solution accuracy. Identified results indicated that the model provides better results than those obtained by different optimization methods in literature and significantly improves the calibration statistics by considering the variability of roughness parameters along the flow reach. © 2013 Elsevier B.V.