Data assimilation in a flood modelling system using the ensemble Kalman filter

Abstract

Data assimilation in a combined 1D-2D numerical flood modelling system is considered. The model is based on a dynamic linking between existing and well- established 1D and a 2D numerical modelling systems enhanced with new features which are targeted specifically towards modelling of floods. This combination ensures a maximum of flexibility by allowing modelling some areas in 2D detail (floodplain), while other areas can be modelled in 1D (river network). For this combined modelling system data assimilation facilities have been implemen¬ted for assimilation of water level measurements. The data assimilation system is based on the ensemble Kalman filter (EnKF) methodology. In the EnKF the probability density of the model state is represented by an ensemble of model states. In a model forecast each ensemble member is propagated according to the dynamical system subjected to model errors, and the resulting ensemble then provides estimates of the forecast state vector and the corresponding covariance matrix. When measurements are available, the ensemble is updated using the standard Kalman filter updating scheme to provide an updated probability density of the model state. In this paper the implementation of the EnKF in the combined 1D-2D model is discussed and is demonstrated on a flood model application in Bangladesh.