Compared with other types of floods, timely and accurately predicting flash floods is particularly challenging due to the small spatiotemporal scales in which the hydrologic and hydraulic processes tend to develop, and to the short lead time between the causative event and the inundation scenario. With continuous increased availability of data and computational power, the interest in applying techniques based on machine learning for hydrologic purposes in the context of operational forecasting has also been increasing. The primary goal of the research activities developed in the context of this thesis is to explore the use of emerging machine learning techniques for enhancing flash flood forecasting. The studies presented start with a review on the state-of-the-art of documented forecasting systems suitable for flash floods, followed by an assessment of the potential of using multiple concurrent precipitation estimates for early prediction of high-discharge scenarios in a flashy catchment. Then, the problem of rapidly producing realistic highresolution flood inundation maps is explored through the use of hybrid machine learning models based on Non-linear AutoRegressive with eXogenous inputs (NARX) and SelfOrganizing Maps (SOM) structures as surrogates of a 2D hydraulic model. In this context, the use of k-fold ensemble is proposed and evaluated as an approach for estimating uncertainties related to the surrogating of a physics-based model. The results indicate that, in a small and flashy catchment, the abstract nature of data processing in machine learning models benefits from the presentation of multiple concurrent precipitation products to perform rainfall-runoff simulations when compared to the business-as-usual single-precipitation approach. Also, it was found that the hybrid NARX-SOM models, previously explored for slowly developing flood scenarios, present acceptable performances for surrogating high-resolution models in rapidly evolving inundation events for the production of both deterministic and probabilistic inundation maps in which uncertainties are adequately estimated. / Thesis / Doctor of Science (PhD) / Flash floods are among the most hazardous and impactful environmental disasters faced by different societies across the globe. The timely adoption of mitigation actions by decision makers and response teams is particularly challenging due to the rapid development of such events after (or even during) the occurrence of an intense rainfall. The short time interval available for response teams imposes a constraint for the direct use of computationally demanding components in real-time forecasting chains. Examples of such are high-resolution 2D hydraulic models based on physics laws, which are capable to produce valuable flood inundation maps dynamically. This research explores the potential of using machine learning models to reproduce the behavior of hydraulic models designed to simulate the evolution of flood inundation maps in a configuration suitable for operational flash flood forecasting application. Contributions of this thesis include (1) a comprehensive literature review on the recent advances and approaches adopted in operational flash flood forecasting systems with the identification and the highlighting of the main research gaps on this topic, (2) the identification of evidences that machine learning models have the potential to identify patterns in multiple quantitative precipitation estimates from different sources for enhancing the performance of rainfall-runoff estimation in urban catchments prone to flash floods, (3) the assessment that hybrid data driven structures based on self-organizing maps (SOM) and nonlinear autoregressive with exogenous inputs (NARX), originally proposed for large scale and slow-developing flood scenarios, can be successfully applied on flashy catchments, and (4) the proposal of using k-folding ensemble as a technique to produce probabilistic flood inundation forecasts in which the uncertainty inherent to the surrogating step is represented.

This study deals with the investigation of characteristics of a flash flood and development of design of flood mitigation facilities occurred in G&uuml / zelyurt in North Cyprus on 18th of January, 2010 and development of design of flood mitigation facilities. Hydrologic and hydraulic modeling of this flood event has been utilized to develop solutions for preventing the region from the flood. Topographical maps and soil properties are used in hydrological modeling. The data are inserted into a geographical information system program (ARC-GIS) where basin properties are obtained. Since there is no any stream flow gauging station along the creeks in the study area, the synthetic unit hydrograph is developed by using Soil Conversation Service Method to obtain design flood hydrographs. In hydraulic modeling, the cross-section data of Fabrika Creek and Bostanci Creek are taken by using global navigation satellite system (GNSS) device and total station. These data are entered into the HEC-RAS program. Flood inundation maps are generated for both creeks. After hydrological and hydraulic modeling, two solutions are proposed. The first one is to build a detention basin for storing water and a lateral channel. for diverting extra flow from Bostanci Creek to Fabrika Creek. The second solution is to build a lateral channel from Bostanci Creek to G&uuml / zelyurt Dam for diverting all water during a flood event. Based on hydrologic, hydraulic, and cost analysis, the first solution is accepted to be the feasible solution. In addition, flow carrying capacities of the creeks are improved.

This master thesis gives an account for the numerous uncertainties that prevail one-dimensional hydraulic models and flood inundation maps, as well as suitable assessment methods for different types of uncertainties. A conducted uncertainty assessment on the river Voxnan in Sweden has been performed. The case study included the calibra-tion uncertainty in the spatially varying roughness coefficient and the boundary condi-tion uncertainty in the magnitude of a 100-year flood, in present and future climate conditions. By combining a scenario analysis, GLUE calibration method and Monte Carlo analysis, the included uncertainties with different natures could be assessed. Significant uncer-tainties regarding the magnitude of a 100-year flood from frequency analysis was found. The largest contribution to the overall uncertainty was given by the variance between the nine global climate models, emphasizing the importance of including projections from an ensemble of models in climate change studies. Furthermore, the study gives a methodological example on how to present uncertainty estimates visually in probabilistic flood inundation maps. The conducted method of how the climate change uncertainties, scenarios and models, were handled in frequency analysis is also suggested to be a relevant result of the study.

Tsunamis are one of the most catastrophic natural events impacting coastal regions often generated by undersea earthquakes. Nevertheless, in enclosed basins, i.e., fjords, reservoirs and lakes, subaerial or submarine landslides can initiate devastating tsunamis with similar consequences. Although a subaerial or submarine landslide that impinges into a large water body can generate a tsunami, subaerial landslides are much more efficient tsunami generators than its counterpart. In this study we aim to integrate laboratory scale experiments of tsunami generation by subaerial landslide with numerical models. The work focuses on the numerical validation of two three-dimensional Navier-Stokes (3D-NS) models, FLOW-3D and our developed model TSUNAMI3D. The models are validated based on previous large scale laboratory experiments performed by a tsunami research team lead by Dr. Hermann Fritz, Georgia Institute of Technology. Three large scale landslide scenarios were selected from the set of laboratory experiments, namely, fjord like, headland and far field coastline. These scenarios showed that complex wave fields can be generated by subaerial landslides. The correct definition and evolution of the wave field are key to accurate modeling the ensuing tsunami and its effect in coastal regions. In this study, comparisons are performed between numerical results and laboratory experiments. Methodology and key parameters for soil rheology are defined for model validations. Results of the models are expected to be under the allowable errors indicated by the National Tsunami Hazard Mitigation Program (NTHMP), National Oceanic and Atmospheric Administration (NOAA) guidelines for validation of tsunami numerical models. The ultimate goal of this research is to obtain better tsunami calculation tools for real-world application of 3-D models for landslide tsunamis, which are necessary for the construction of inundation maps in the Gulf of Mexico and the Caribbean regions.