Novel Computational and Stochastic Methods for Characterizing Joint Flood Hazards under Extreme Hurricane Events
Torres, Jacob M.
Doctor of Philosophy
In the United States, coastal flood risk management is traditionally predicated on protecting against extreme hurricane-induced storm surge (expressed in annualized return periods). However, (1) hurricane storm surge and (2) hurricane rainfall-runoff are not mutually exclusive coastal flood hazards. Little research has emphasized the need for characterizing the joint physical surge-rainfall processes for enhancing our understanding of the coastal hydrologic landscape as it relates to resiliency. This investigation involves the collection of interconnected studies that provide a set of original research contributions to the body-of-knowledge. A key research component entails the coupled development and calibration of advanced numerical models for in-depth hydrodynamic analyses of complex discrete-event hurricane simulations. These models include an improved distributed hydrologic model for simulating large-scale watershed response from hurricane rainfall events; the utilization of a revalidated (2) finite element hydrodynamic model that couples Simulating WAves Nearshore with ADvanced CIRCulation (SWAN+ADCIRC) for quantifying storm surge and wave dynamics; and an (3) improved unsteady riverine hydraulic model for associating storm surge and rainfall-runoff momentum interactions at the Houston Ship Channel (HSC). The coupling of these models is driven by the need for a centralized numerical testbed for which complex hydrological processes at the coastal-riverine interface can be studied. Modeled storm types include historical, pseudo-synthetic, and fully-synthetic hurricanes, some of which involve a novel spatial and temporal translation of hurricane windfields and rainfall to the model domain. Derived insights on coupled surge-rainfall processes are applied to storm surge mitigation for the Houston Ship Channel (HSC) and Galveston Bay regions. The modeled concepts of storm surge barrier systems are also analyzed under 1-D (unsteady HEC-RAS) and 2-D modeling (SWAN+ADCIRC) frameworks. Overall, findings reveal how hurricanes producing relatively little storm surge but high rainfall can be comparable in flood potential to the inverse scenario of hurricanes producing high storm surge and little rainfall for a given location. This is owed to heterogeneous topologies of coastal watershed boundaries, and varies with hurricane landfall location. This relationship affects our ability to protect coastal communities from threats against downstream hurricane storm surge and upstream hurricane rainfall-runoff. The storm surge barriers modeled at the HSC and Galveston Bay are shown be hydraulically feasible for the scenarios analyzed. However, results show that structural barrier placement across relatively incised channels that drain coastal watersheds are more prone to complexities involving timing separations of peak storm surge and rainfall-runoff. This complexity is exacerbated when considering dynamic operations of surge barrier systems and highlights the importance for accurate hurricane flood forecasting. Equally vital towards a practical achievement of coastal resiliency are the continuous functionalities of critical and interconnected infrastructure systems during hazard events. This study investigates the feasibility of coupling physics-based models with graph theory for developing proxies on engineered performance for water distribution systems (WDS). Results show strong correlations can exist for certain graph theory metrics and WDS performance, thus improving its favorability as a practice-ready approach with broader insights on pipe network topologies that are more robust to coastal natural hazards and other disasters.
Hurricane Rainfall, Storm Surge, Surge Barrier, Surge Protection, Galveston Bay, Houston Ship Channel, SWAN+ADCIRC, Water Distribution Systems, EPANET, Graph Theory, Statistical Inference