Even though flood damage cannot be fully controlled, its effect can be minimized to some extent by careful planning, flood mitigation measures, and an effective flood warning system. Therefore, flood warning systems with flood travel time and inundation area information, derived from accurate model prediction, can be very effective to reduce potential flood damage. While a one-dimensional (1D) model was developed in the former research for the flood warning system, there has not been many comparative assessment of model performance among 1D, two-dimensional (2D), and coupled one-dimensional and two-dimensional (coupled 1D/2D) models particularly in HEC-RAS. Therefore, this research is an extension of the prior research and was especially conducted to calculate and compare the predictive capability of 1D, 2D, and coupled 1D/2D HEC-RAS models for the computation of travel time of flood and extent of flooded area needed for a flood warning system. The research was carried out in the Grand River in Lake County, Ohio. The model performance of 1D, 2D and coupled 1D/2D models were evaluated and sensitivity analysis was conducted using the same set of flow conditions and geometric conditions. The analysis suggested that 2D model could incredibly improve the model performance compared to 1D and coupled 1D/2D models, which were evaluated through the model evaluation indicators for the observed and simulated model outputs. Additionally, sensitivity analysis of input parameters, including discharge and Manning’s roughness, revealed that the 2D model was comparatively less sensitive to the changes in model inflow and Manning's roughness compared to the coupled 1D/2D and 1D models. Furthermore, the flood travel time computed using 1D model was more predicted than that of the 2D model, indicating that the 2D model would be most appropriate to provide a safe evacuation time for the community before flood events. The 1D model consistently over predicted than that of the 2D model, which was also true for the estimation of the inundation flood zone (4.1% higher).
In addition, the appropriate assessment of flood damage in the aftermath of major flooding is crucial for flood management agencies, emergency responders, and insurance companies. Therefore, damage assessment is an important step in the evaluation of the flood mitigation measures, vulnerability analysis and flood risk mapping. This is particularly true in a context that the damage assessment so far has been primarily relying on either the coarse resolution, 30m digital elevation model (DEM), or 1D hydraulic model. As this researcher is not aware of any explicit incorporation of 2D HEC-RAS model for the damage assessment among the scientific communities, another major objective of this analysis is to outline the effects of some of the key factors including the mode of hydraulic simulation (1D vs 2D), the effect of inventory data, and the effect of topography on the flood loss estimation. This was accomplished using the 1D and 2D HEC-RAS models to produce the flood depth grids from the varying degree of topographic resolutions including 30m, 10m and LiDAR-derived DEM with and without incorporating actual field survey of the river in each case. The flood loss was estimated using Hazards United States Multi-Hazards (HAZUS-MH) loss estimation software developed by Federal Emergency Management Agency (FEMA) software, for each building within study area for flood events of various recurrence intervals from 10 to 500-year return periods. This was accomplished by updating the default-building inventory within Lake County to represent the actual building information in the model. The analysis indicated that 1D model consistently overestimated the loss in general by 61.48% for the default database and 86.12% for updated inventory. The estimation of the 1D model was consistently larger compared to the 2D model for different set of topographic resolutions and recurrence intervals. These loss estimations significantly increased when analyzed using a coarse resolution terrain, which was true regardless of selecting 1D or 2D models. Furthermore, the 2D model showed a lesser percentage increase (10.45% in 10m DEM, and 25.49% in 30m DEM), whereas the 1D model exhibited a larger increment (23.17% in 10m DEM and 76.81% in 30m DEM). This analysis suggested that the loss estimation would decrease in general by 76.21% after incorporating additional local building data into the HAZUS-MH database. More specifically, this analysis concludes that 2D model with high-resolution topographic data, including the additional incorporation of local data, in HAZUS-MH database are tremendously essential for appropriate flood damage assessment.