Keywords

hydraulic coupling; flood forecasting; domain decomposition; data assimilation.

Location

Session B1: Data Assimilation Techniques for Uncertainty Reduction

Start Date

13-7-2016 11:10 AM

End Date

13-7-2016 11:30 AM

Abstract

In the context of hydrodynamic modelling the use of 2D models is of fundamental importance in areas where the flow is not mono-dimensional. Nonetheless, the lack of bathymetric data to develop a global 2D model over an entire network or the computational cost of a global 2D model can be prohibitive for operational flood forecasting services. To overcome this limitation a solution is to use multi-dimensional coupling with the use of 1D models where the flow is mono- dimensional and the use of 2D models when needed. Multi-dimensional coupling allows for the use of a global hydrodynamic 1D/2D model over a network but for a much reduced computational cost compared to a global 2D model. It also guarantees the continuity of the hydraulic variables at the interfaces between the 1D and the 2D models with the use of a Schwarz algorithm adapted to the hydraulic case. In this study, multi-dimensional coupling in hydraulic was combined with 2D domain decomposition and 1D data assimilation of water level observations on the “Adour maritime” hydraulic network. This network is located in South West of France where a global 1D model on the network and a local 2D model on the Bayonne area have been developed and are used in operational context by French flood forecasting services. It was shown that the local 2D model improves the hydraulic modelling compared to a 1D model and that the combination of hydraulic coupling and 2D domain decomposition makes this methods compatible with operational constraints.

COinS
 
Jul 13th, 11:10 AM Jul 13th, 11:30 AM

Multi-dimensional hydraulic coupling for flood forecasting

Session B1: Data Assimilation Techniques for Uncertainty Reduction

In the context of hydrodynamic modelling the use of 2D models is of fundamental importance in areas where the flow is not mono-dimensional. Nonetheless, the lack of bathymetric data to develop a global 2D model over an entire network or the computational cost of a global 2D model can be prohibitive for operational flood forecasting services. To overcome this limitation a solution is to use multi-dimensional coupling with the use of 1D models where the flow is mono- dimensional and the use of 2D models when needed. Multi-dimensional coupling allows for the use of a global hydrodynamic 1D/2D model over a network but for a much reduced computational cost compared to a global 2D model. It also guarantees the continuity of the hydraulic variables at the interfaces between the 1D and the 2D models with the use of a Schwarz algorithm adapted to the hydraulic case. In this study, multi-dimensional coupling in hydraulic was combined with 2D domain decomposition and 1D data assimilation of water level observations on the “Adour maritime” hydraulic network. This network is located in South West of France where a global 1D model on the network and a local 2D model on the Bayonne area have been developed and are used in operational context by French flood forecasting services. It was shown that the local 2D model improves the hydraulic modelling compared to a 1D model and that the combination of hydraulic coupling and 2D domain decomposition makes this methods compatible with operational constraints.