Evaluation of dispersion parameters for River São Pedro, Brazil, by the simulated annealing method

Abstract

The dispersion coefficient parameterizes different and complex phenomena that control substance transport in water bodies. This article presents an inverse problem approach solved by the simulated annealing (SA) algorithm to estimate dispersion coefficient and other parameters that can model dispersion in rivers with non-uniform geometry. The method is compared with two other classical methods, the peak concentration and Chatwin. The SA algorithm, peak and Chatwin methods were applied to the advection–dispersion model, whereas the SA was tested in the dispersion dead zone model (DDZM) as well. The methods are assessed in the study of a tracer transport achieved in a small mountain river. Comparing with observations, the best performance was obtained conjugating SA and the DDZM. The classical methods showed high sensitivity to the river discharge determination method, being more precise when the sections method was adopted. Considering this restriction, the inverse problem formulation solved by the SA approach showed to be a promising alternative, given that it can be built in a way that the river discharge itself can be estimated.

Keywords:

• inverse problem
• advection–dispersion equation
• dispersion dead zone model
• dispersion in mountain rivers
• simulated annealing algorithm

Acknowledgements

The authors acknowledge partial financial support from FAPERJ (Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro) and CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior). One of the authors (FDMN) acknowledges the partial financial support from CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico, Brazil).