ABSTRACT
Despite the increasing development and use of three-dimensional hydrodynamic models, two-dimensional models are still one of the main constituents in many engineering applications and specially in storm surge models. Bode and Hardy [4] review the topic of storm surge models, with a description of the most important components of the model, as well as defining the state of the art within this topic. They consider 2D models as the backbone of state of the art operational storm surge prediction for which coastal water levels are of major concern. The result of these models do not always match the observations, due to all the simplifications introduced. Real-time assimilation of data into storm surge models has been used during the last years to improve the model results. Applications of the Steady state Kalman filter (Heemink [31]) on linear 2D shallow water models for storm surge forecasting can be found in Bolding [5] Heemink [32], De Vries [66], Vested et al. [65] and Heemink et al. [34]. The last reference also includes an application of the RRSQRT filter on a twin test for storm surge forecasting in the North Sea. The RRSQRT filter permits the use of a set of observations that are variable in time and space, which is very convenient for data assimilation in storm surge modelling. The RRSQRT filter has been tested using a real case, the storm occurred on the North Sea during February 1993, and preliminary results can be found in Cañizares et al. [10].
