This study investigate the performance of simulation realizations between two hydrodynamic models, including Sobek and 3Di. Of which, 3Di takes the advantages of quadtree technique to reduce computational grids and subgrid method to compute roughness to the level of pixel grid. Several DEMs (digital elevation models) of different resolutions are produced from the 1 x 1 m DEM and DSM ( digital surface model).
The framework of this project includes: (1) preparation of model input data, such as DEM adjustment for hydrodynamic models, roughness map, infiltration map, rainfall data in netCDF format (a [x,y,z,t] format commonly used in metrology), etc.; (2) comparison between sobek and 3Di models with several DEMs mentioned previously; (3) comparison of DEMs considering the blockage effect of buildings; (4) initial evaluation of 2D inundation simulation resulted projected on a 3D visualization model.
Results of the first year project show that 3DI runs 30-300 times faster than Sobek in pure 2D flood simulation. 3Di is capable of dealing with higher resolution DEMs with ditched channels of 1x1m, while Sobek may only deal with 10x10m DEMs (in this case study) with 10x10m ditched channels. Thus, the Sobek simulations are compromised in terms of resolution, area evaluations and flood extents.
Furthermore, the blockage of buildings also demonstrates an interesting case of the LiDAR DEM and DSM in 2D flood simulations.