Typical physically-based 2D flood models solve the Shallow Water Equations (SWEs), requiring high computational resources. Many of these models have been developed to obtain better performance, while maintaining the required accuracy, by reducing the complexity of the SWEs. This reduction is usually achieved by approximating or neglecting less significant terms of the equations (Hunter et al., 2007; Yen and Tsai, 2001). The JFLOW model (Bradbrook et al., 2004), Urban Inundation Model (UIM) (Chen et al., 2007), and the diffusive version of LISFLOOD-FP (Hunter et al., 2005) solve the 2D diffusion wave equations that neglect the inertial (local acceleration) and advection (convective acceleration) terms (Yen and Tsai, 2001). The inertial version of LISFLOOD-FP (Bates et al., 2010) solves the SWEs without the advection term. In either version of LISFLOOD-FP the flow is decoupled in the Cartesian directions. Other models use the full SWEs but focus on the use of multi resolution grids or irregular mesh, like InfoWorks ICM (Innovyze, 2012) and MIKE FLOOD (DHI Software, 2014; Hénonin et al., 2013). These last two models are commercial packages, and the code applied in the optimisation techniques is not in the public domain.
