A new data-based modelling method for identifying parsimonious nonlinear rainfall/flow models

Vincent Laurain (1) , Marion Gilson (1) , Sylvain Payraudeau (2) , Caroline Grégoire (2) , Hugues Garnier (1)

1 CRAN - Centre de Recherche en Automatique de Nancy
2 LHyGeS - Laboratoire d'Hydrologie et de Géochimie de Strasbourg

The identification of rainfall/runoff relationship is a challenging issue, mainly because of the complexity to find a suitable model for a whole given catchment. Conceptual hydrological models fail to describe correctly the dynamic changes of the system for different rainfall events (e.g. intensity or duration). However, the need for such relationship grows with the water pollution increase in agricultural regions. Lately, a well-known type of model in the control field appears to be a suitable candidate for water processes identification: the Linear Parameter Varying (LPV) models. This paper depicts a novel refined instrumental variable based method for the identification of Input/Output LPV models and this algorithm is applied to identify a parsimonious nonlinear rainfall/flow model of a 42 ha vineyard catchment located in Alsace, France.

Domaines: Automatique / Robotique

Vincent Laurain, Marion Gilson, Sylvain Payraudeau, Caroline Grégoire, Hugues Garnier. A new data-based modelling method for identifying parsimonious nonlinear rainfall/flow models. International Congress on Environmental Modelling and Software, IEMSS 2010, Jul 2010, Ottawa, Canada. pp.CDROM. ⟨hal-00508680⟩