Rapid flood damage prediction in short lead-time scenarios
Research on short lead-time flood inundation and damage assessment traditionally focuses on developing tools for long term planning. Studies that investigate the ability to provide short lead-time analyses are uncommon because of sparcity and spatial inconsistency of short-term hydrologic forecasts. Recent advances in continental scale hydrology make possible the ability to predict discharge at nearly all locations in the conterminous United States (CONUS). Theoretically, these discharge outputs enable investigation into the plausibility of short lead-time hydraulic, and damage analyses during flood events. A system predicting the hydraulics and damage potential of floods is only feasible after addressing a number of limitations. For instance, hydraulic models require a characterization of the stream channel, either through site surveys or through approximations. Further, the damage estimation methodology requires building inventories, which are developable by similar mechanisms. This dissertation advances knowledge on how to possibly address these limitations. The first study in this dissertation investigates the use of hydraulic geometries. Hydraulic geometries relate river channel geometry to bankfull discharge. However, the research presented here indicates that hydraulic geometries may estimate both the channel geometry and multiple depths of flow, under certain geomorphic and anthropogenic constraints. Accurate channel geometries are necessary for hydraulic modeling. Depth of flow estimates are useful in developing stage-discharge rating curves and possibly as a standalone means of estimating inundation grids. The second and third studies in this research look to investigate a framework for rapid flood damage assessment using public domain cadastral or parcel geospatial data. The second study discusses a fuzzy logic framework for meshing emergency response Address Points with parcel data and appropriate depth-damage relationships to determine both percent and fiscal impact of flood damage. Results of the second study highlight the effectiveness of determining flood damage with cadastral data using fuzzy text matching. The third study investigates what inputs from the cadastral data are necessary for the fuzzy logic framework to approximate a detailed flood damage investigation. Results indicate that fuzzy logic can approximate a detailed study when provided discrete use descriptions, market value, and square footage.