Section: New Results
Trajectory reconstruction for Lagrangian data assimilation
This work is performed in cooperation with M. Honnorat and J. Monnier (INRIA team MOISE) in the framework of the ASSIMAGE ACI: an hydrological model based on Saint-Venant equations (state space constituted of water elevation and 2D velocity) is used in a data assimilation framework for flood forecasting.
The practical difficulty lies in access to observation data, especially 2D velocity. We thus applied tracking methodologies on video acquisitions of water flow. This has been performed on the experimental installation of the Fluid Mechanic and Acoustic Laboratory of University of Lyon (LMFA) of river simulation. Trajectories are computed by tracking targets with a maximum correlation matching method. The trajectories are then assimilated in a Lagrangian framework in order to retrieve initial condition and/or parameters in the hydrological model.