Parameter estimation of a real hydrological system using an adjoint method

van Tri Nguyen (1) , Didier Georges (1) , Gildas Besancon (1, 2)

1 GIPSA-SYSCO - GIPSA - Systèmes non linéaires et complexité
2 IUF - Institut universitaire de France

This paper deals with the problem of parameter estimation in a real hydrological catchment of Niger, including Manning roughness and Horton infiltration coefficients. Firstly, the dynamics of water flow on this catchment are modeled by using the continuity equation of Saint-Venant and the Horton infiltration model whose the parameters are supposed to be unknown. The estimation approach is then based on the calculus of variations which is applied on a Lagrangian objective functional, in order to get the gradients of this functional with respect to parameters needing to be estimated. These gradients are used as inputs of a quasi-Newton optimization algorithm to solve the estimation problem. The presented approach is implemented with real data of the Niger catchment called Tondi Kiboro to get the desired values of parameters.

Mots clés: Horton model; Parameter estimation; Infinite dimensional systems; Inverse problem; Saint-Venant model; Adjoint method; Tondi Kiboro catchment

Domaines: Automatique / Robotique

van Tri Nguyen, Didier Georges, Gildas Besancon. Parameter estimation of a real hydrological system using an adjoint method. ALCOSP 2016 - 12th IFAC Workshop Adaptation and Learning in Control and Signal Processing, IFAC, Jun 2016, Eindhoven, Netherlands. ⟨hal-01333740⟩