Document Type
Presentation
Publication Date
8-1-2014
Abstract
This study focuses on the hydrodynamic modelling of the tidally influenced “Adour maritime” catchment located in South-West of France. The local SPC is in charge of operational flood-forecasting and for that purpose, the 1D hydraulic software MASCARET developed at EDF is used. Due to various sources of uncertainties on the hydraulic parameters and hydrological forcing, the free-run is imperfect and should be corrected for improved forecast of extreme events. An Ensemble Kalman Filter algorithm was implemented on top of the hydraulic model using the OpenPalm coupler to assimilate hourly water level in-situ data from observing stations. The water-level and discharge state is sequentially corrected thus providing and improved initial state for short to medium range forecast (3h-6h). The impact of the data assimilation analysis on the entire network and for the non-assimilated variables is investigated for several major flood events. It was also shown that the reliability of the forecast closely relates to the number of members used in the EnKF algorithm. In the present study, about 20 members are necessary. As these members can be run in parallel, the cost of the ensemble-based assimilation remains compatible with real-time flood-forecasting constraints. The corrected 1D simulation provides the boundary conditions for a limited-area 2D model. The TELEMAC software, developed at EDF, is used to represent the flow for the confluence between Nive and Adour rivers (in the center of the Bayonne city) as well as in the Plaine d'Ansot (located upstream of Bayonne) where the flow is no longer mono-dimensional. The 1D (with data assimilation) and the 2D model overlap, they are coupled in order to build a data-driven high-fidelity model for the Adour catchment for operational use in the framework of flood-forecasting.
Comments
Session R48, Data Processing: Data to Computations