Coastal wetlands in California are critically positioned at the interface between increasingly developed watersheds and the coastal ocean. These wetlands provide habitat for fish and wildlife, provide nutrients to surrounding coastal waters, and create recreational opportunities (Mitsch and Gosselink 1986). This report describes a circulation and transport model that is designed for tidal wetland circulation and mixing studies. Given the importance of wetland restoration projects to offset the impact of coastal development, wetland circulation and mixing models are of great utility for evaluating the advantages and disadvantages of various wetland restoration alternatives relative to issues such as bathing water quality, eutrophication, and sedimentation. An application of the model to a restored wetland and guidelines for its use are presented.

The United States experienced record-breaking losses from natural disasters in 2017. Damages from floods were particularly costly, largely because of the high exposure and value of capital at risk. Without targeted mitigation strategies, flood losses are expected to escalate due to urbanization and the projected changes to the frequency of extreme weather events. This dissertation provides and evaluates engineering methods for flood risk mitigation in an era of global change. The body of work contains three research articles: two provide and critique methodology for incorporating changing environmental conditions in the design of flood control infrastructure along rivers and coasts, and the third contributes guidance for improving the clarity and utility of commonly produced flood hazard maps. The results demonstrate that attempting to forecast the impact of changing environmental conditions on extreme floods substantially increases predictive uncertainty relative to traditional methods. Thus, so-called ``non-stationary" methods have limited utility for decision making and are difficult to use as the basis for infrastructure design or mitigation planning. Our inability to meaningfully forecast changes in extremes over long time periods has important implications for engineering, but the first step is to acknowledge the unknown. Additional factors of safety, adaptable decision making, and better communication of what we actually do know are the keys to successfully reducing losses.

Fecal indicator bacteria (FIB) impairment of coastal water bodies is a wide-spread problem that impacts recreational water contact and shellfish harvesting. Storm water runoff is a major contributor to FIB during wet weather periods leading to beach closures or warnings to stay out of the water for several days after a storm event. However, FIB impairments also occur during the dry season when the levels of water-contact recreation are highest, especially at enclosed beaches. The causes and remedies of these problems can be difficult to identify.

This dissertation addresses the impacts of dry-weather runoff on FIB impairment in Newport Bay, California. The vast majority of runoff (95%) enters the bay at two creeks located several kilometers from recreational waters, while only about 1% of the runoff enters the bay through over a hundred small drains located within the recreational waters. This dissertation examines the relative impact of these large and small runoff sources based on bay-wide mixing and transport processes. Dry weather flow rates are reported as a function of drainage area based on limited field sampling and a regression analysis, and loads of FIB into the bay area estimated based on measured FIB concentrations in runoff and the volumetric flow rate. A 2D flow and scalar transport model is developed and calibrated for prediction of bay-wide salinity, and subsequently applied to simulate the relative impact of large and small drains on FIB impairment. Results show that small drains with minimal fresh water discharges contribute disproportionally to FIB impairment due to their proximity to recreational waters and minimal mixing therein. Further, a "trap and release" mechanism whereby runoff accumulates in drain pipes during rising tides and is released at low tide is modeled and shown to also contribute to the number of FIB exceedances.

This dissertation also presents analysis of the erosion thresholds and erosion rates for sediments in Newport Bay which indicate that sediment resuspension occurs only in the main channel and only during energetic spring tide conditions, during the dry weather months. This result suggests that resuspension of contaminated sediments is most likely not a significant contributor to FIB impairment during dry-weather periods.

Sediment management is costly and critical for managing ecosystems and flood risks, especially in the context of sea-level rise. The field of civil and environmental engineering has numerous models to describe sediment transport and morphology based on hydrodynamics and sediment characteristics. Despite this, multi-decadal predictions of sediment transport are difficult due to stochasticity of environmental systems and uncertainty in future anthropogenic influences such as land-use changes and dredging. Accurately predicting sediment transport is important for coastal and sediment management, especially within urban regions. This dissertation seeks to address sediment management challenges through cross disciplinary research by developing a set of models and tools in response to stakeholder needs in an urban region (Southern California) experiencing numerous sediment management challenges. The second chapter of this dissertation focuses on developing a stochastic hydro-financial model to evaluate the risk of an enviro-financial instrument known as an Environmental Impact Bond (EIB). The model and EIB are applied to address excessive sedimentation resulting in negative environmental and financial outcomes in a transnational setting, the US-Mexico Border. The risk is then tied to an ``environmental interest rate'' which can then be used to reward investors for taking on additional environmental risk, as they would with traditional financial instruments. Chapter 3 of this dissertation develops a stochastic model of estuarine basin deposition using a response surface surrogate model trained using output from a high-fidelity hydromorphodynamic model (Delft3D). The advantages of the surrogate model is a reduction in computational time by multiple orders of magnitude, facilitating stochastic simulation of basin response to multiple model forcing and management scenarios, thereby improving the information available to estuarine managers. Finally, the fourth chapter of the dissertation utilizes a synthetic future forcing taken from a stochastic model to simulate future marsh surfaces utilizing a one-way coupling between a hydromorphodynamic (Delft3D) and marsh accretion model (WARMER) to better describe step changes in marsh surface elevations in ephemeral estuarine systems. This method is quantitatively compared to existing methods and applied to an urban estuary (Upper Newport Bay in Southern California) to evaluate the impacts of SLR on surface elevations and potential habitat suitability through 2100.

Coastal areas are important habitats and contain large human populations. It is estimated that 20 million people reside along coasts below normal high tide levels and over 200 million people are vulnerable to coastal flooding during storms. Many communities are currently protected from flooding by beaches that are sometimes modified with anthropogenic (artificial) berms, so understanding and characterizing beach and berm response to storm waves is critical to adapting and mitigating climate change effects. Beach dynamics are challenging to model because of the complexity of wave dynamics, sediment transport and bed profile adjustments.

This dissertation aims to advance the state of the art in coastal flood prediction by improving modeling of beach dynamics over times scales of hours, when a combination of high tides, storm surge and waves from a storm event can threaten flooding. It is envisioned that with advances in laser scanning technology, beach profiles can be rapidly assessed to provide initial conditions to swash zone models, but improvements in mechanistic modeling are needed to predict whether a beach will be eroded and overtopped, and the extent of flooding, especially when anthropogenic beach berms are used to strengthen coastal flood defenses.

This dissertation proposes a beach model based on vertically averaged, multi-phase flow equations solved by a shock-capturing finite volume scheme. The model domain corresponds to the the so-called swash zone, the region between the shoreline and the inner surf zone where a layer-averaged model based on the assumption of hydrostatic pressure has been found to be a good approximation of system dynamics. Further offshore at intermediate to deep water depths, spectral wave models are routinely applied to describe wave transformations and output can be used as a boundary condition for the beach model.

The main contributions of this dissertation include new shock-capturing numerical methods for solving layer-averaged multi-phase flow equations, original experiments characterizing anthropogenic berm erosion and overtopping at field scale, and numerical modeling of beach and berm erosion aimed at measuring the predictive skill of the proposed model, developing an improved process understanding, and assessing strengths and weaknesses of the model.

Human development situated at the foot of mountains faces sediment-laden flood hazards characterized by high-velocity, erosive flows carrying mud and debris. In the western United States, sediment-laden flood hazards are increasing due to more frequent and severe wildfires, more intense precipitation, and the expansion of development towards mountain wildlands. Considerable work has focused on developing models of post-fire peak flows and sediment yields at the outlet of mountain canyons, but the risk to communities downstream of protective flood infrastructure common throughout the southwestern U.S. is largely unknown. We present an original modeling framework that captures the interactions between wildfires, storms, and flood infrastructure to estimate sediment-laden flood hazards. Stochastic modeling with a continuous simulation approach is used to quantify uncertainty and explicitly consider antecedent conditions. This work shows that compound post-fire flood hazards may be up to 6 times larger than suggested by the marginal hazard posed by extreme precipitation, and that future increases in wildfire severity and intensity could increase flood hazards by up to a factor of 11. Furthermore, modeling of compound post-fire flood hazards across the Santa Ana Mountains in Riverside County, California reveals hot spots of risk below catchments that last burned over 45 years ago. As urban expansion into mountainous regions continues to increase around the world, an improved understanding of the hazards facing communities “protected” by infrastructure is important to better characterize the spatial distribution of risks across populations, increase risk awareness, and inform sustainable adaptation and resilient land development practices.