2.5. Inland Inundation Modeling
A well-established 2D hydraulic model (FloodMap-Inertial, Yu and Lane 2011) is utilized to derive coastal flood inundation for each scenario. The model, as well as its earlier diffusion-based version (FloodMap, Yu and Lane 2006a,b) have been tested and verified in a number of different environments (e.g., Tayefi et al., 2007; Lane et al., 2008; Casas et al., 2010; Yin et al., 2013, 2016, 2019; Yu and Coulthard, 2015). It adopts a simplified 2D solution to numerically solve the inertial form of the 2D shallow water equations by neglecting the convective acceleration term in a raster-based domain. Inland flood routing takes the same structure as the inertial algorithm of Bates et al. (2010), but with a slightly different method to calculate the time step. Rather than using a global Courant–Freidrich–Levy Condition where the time step for the next iteration is calculated based on the maximum water depth and velocity found at the last time step (e.g. Bates and De Roo, 2000), the Forward Courant–Freidrich–Levy Condition (FCFL) approach described in Yu and Lane (2011) for the diffusion-based version of FloodMap is used in the inertial model to maintain simulation stability and minimize numerical diffusion.
To apply the model for inland flood simulation, two types of data are required: floodplain topography and boundary conditions. In terms of the floodplain topography, a city-wide Digital Elevation Model (DEM) constructed from 0.5 m topographic contours is available (Yin et al., 2013), with a grid cell resolution of 50 m and a vertical accuracy of 0.1-0.2 m. The locations and heights of sea dikes along the coast, provided by Shanghai Water Authority (SWA), are then overlaid onto the original ‘bare earth’ DEM. An empirically-based uniform manning’s n roughness coefficient of 0.06 is used in the simulation to represent the effect of urban features (e.g. buildings) on flood routing. For each synthetic storm, water level time series (every 30 minutes) at locations spaced at 1000 m intervals along the coast, derived from the ADCIRC model run, were used as flow boundary conditions to drive the flood modeling. The performance of the coupled ADCIRC-FloodMap method was investigated by Yin et al., (2016), where the ADCIRC-FloodMap method demonstrated higher efficiency and accuracy in inundation prediction at the city scale than using the ADCIRC model alone, which is particularly important for hazard modeling involving thousands of simulations.