The Blackmouth Shiner (Notropis melanostomus) is a small North American cyprinid that is listed as a Species of Concern due its relatively small range, occurring at only a few sites in Mississippi, Alabama, and Florida. Due to limited data and the small number of actual samples of N. melanostomus, the true characteristics of viable habitat and distribution remain unknown. The objective of my research was to utilize remote sensing data to gain a better understanding of the habitat characteristics where the N. melanostomus has been collected and use this information to identify other areas were populations are likely to occur during future sampling efforts. In particular, Landsat data were used to map the spatial and temporal extent of water inundation over a 20-year time-series within floodplain water bodies surrounding the Pascagoula River to determine the effects on the presence or absence of Blackmouth Shiners at historic collection sites. These characteristics could be used to inform future site selections within the Pascagoula River drainage as well as identify other river systems that have similar inundation patterns and morphology within and proximal to the known range.

Philosophiae Doctor - PhD / The Okavango Delta is recognized as one of the famous inland wetlands and its sustainable use is important for socio-economic development of Botswana. The Okavango delta comprises permanent swamps, seasonal swamps, and drylands on islands within the delta and the surrounding areas, sustained by Okavango river inflows from upstream and local rainfall. TheOkavango River splits into several distributary channels within the delta. Areas which are flooded annually vary in response to varying inflows into the delta. Peak inflows into the delta occur during the February to May period. Due to the low gradient over the delta, these inflows move slowly resulting in peak outflows from the delta occurring during the June to August period. The inundated area over the entire delta increases from May until it reaches maximum inAugust and starts to decrease from September, reaching minimum inundated area in the months of December and January. The incoming flood wave into the delta and maximum inundation is out of phase with the local rainfall season.Communities living within and around the delta derive their livelihoods from tourism, hunting, fishing, livestock rearing, and crop production. Crop production is carried out on drylands and within floodplains. Some of the households take advantage of the increase in soil moisture arising from this inundation along floodplains to cultivate their crops as the floods recede. This practice is locally referred to as molapo farming which highly depends on inundation of floodplains. The availability of floodplain inundation highly depends on the magnitude of inflows into the delta and the local rainfall which are highly variable resulting in uncertainty regarding successful crop production, availability of livestock grazing areas, and uncertainty in reliance on the wetlands resources such as fishing. The uncertainty experienced in timing of extreme events which cause flooding of resulting in water reaching areas or floodplains where it is not wanted, and also uncertainity in timing of low flows, therefore water not reaching some parts of the delta.Several hydrological studies have been carried out with the aim of improving the understanding of the spatial and temporal dynamics of flows throughout the delta including predicting areas that are likely to be inundated each year. The significant gap addressed by this research is to improve the understanding of the spatial and temporal influence of magnitude and timing of flows on floodplain inundation. Local rainfall on the delta is highly variable over time and space due to its convective nature. This research also addresses the rainfall temporal and spatial variations and its implications on floodplain inundation. The knowledge about spatial extent and duration of floodplain inundation should assist in predicting each year the viability of molapo farming. Three research site, Shorobe, Tubu and Xobe are selected as case studies to understand the dynamics of floodplain inundation induced either by inflows or local rainfall. Local rainfall during the December to March period enables the crops to reach maturity. The onset of the rainy season is very important in supporting sowing of crop seeds. Local rainfall on the delta varies considerably. Aerial rainfall interpolation shows a change in rainfall magnitudes over space in different rainfall months, i.e different parts of the delta receive different rainfall magnitudes in different months of the rainy season. The spatial variation is mainly associated with the migration of the ITCZ southwards first through East Africa during October andNovember and down over Southern Africa in December to February. The movement of the ITCZ brings rainfall concentration on the northern and eastern parts of the Okavango Delta during December to January and bringing rainfall concentration to the northwestern part of the delta around February. However, rainfall spatial correlation between stations can be poor even within the first 150 km therefore implying neighboring places do not experience floodplain inundation by rainfall at the same time. The poor spatial correlation of rainfall between neighboring stations reflects the erratic nature of rainfall in the Okavango Delta characterised by localized thunderstorms. Change detection shows change points in rainfall which can be associated with ENSO episodes. A change point is identified in 1976 and 1977 which can be associated with the El Nino episodes during those years and two change points identified in 1999 and 2004 which can be associated with the La Nina episodes, therefore rainfall induced floodplain inundation can also be associated with wet and dry ENSO episodes. Rainfall does not show any significant trends except for an increasing trend on 10th percentile of Shakawe rainfall. Rainfall also does not show any cyclic behavior. Rainfall over the Okavango Delta can be divided into three unique homogenious sub-regions; sub-region 1: the northern part following the GEV probability distribution and being the region with highest rainfall amounts; sub-region 2: the lower northern and the outlet parts of the Okavango Delta following the GPA distribution with moderate rainfall; and sub-region 3: the middle part of the delta extending to the western and the eastern fringes of the delta, following the P3 distribution and having the lowest rainfall.The main characteristic that defines the Okavango Delta flows at Mohembo is its cyclic behavior. Three significant cycles are identified, close to 10, 20 and 40 years. No significant trends are identified, only a decreasing trend in minimum flows. Change points are identified in 1979 and 1988 and these can be explained by the existing cyclicity since no major land use changes have taken place in the Okavango River Basin upstream before 1989. The existence of cyclicity in Okavango River flows at Mohembo also explains the periodic wetting and drying of different floodplains in the delta. A long period of low flows was experienced from 1983 until 2003 and floodplain inundation extent was greatly reduced, more especially during the 1993-2003. During the 1993-2003 period, flows could no longer reach Maun Bridge along Thamalakne River, therefore leaving molapo floodplains around Boteti River, Gomoti River and Thaoge River to dry out. The 10 and 40 year return floods are important as they indicate the probability of a flood magnitude which has potential to result in major inundation in the Okavango Delta. Therefore, flood magnitudes with recurrence interval 10 and 40 years have high probability of occurring and can cause major floodplain inundation as they can be above the 2009 flood of 969 m3/s, which was the return of major inundation of Okavango Delta floodplains after a long period of dryness. The Ngoqa-Maunachira distributary channel of the Okavango River receives 32% of flow volumes entering the Okavango Delta at Mohembo. 12 % of the Mohembo flow volumes reach the Jao-Boro distributary whilst 1% is received by the Thaoge distributary. Therefore more inundation is experienced along the Ngoqa-Maunachira system compared to the other two. Only about 2% of the Mohembo flow volumes leave the Okavango Delta through Boteti River. Long term shifting of flow direction amongst reaches along the Okavango Delta distributaries is evident more especially along the Ngoqa-Maunachira River system. This results in shifting of inundation. Sub-surface water respond significantly to local rainfall and inflows with high soil moisture conditions retained at 60 cm and 100 cm below the ground.

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.

In the last few centuries, many streams in the eastern United States have been severely disturbed by land use change and are now disconnected from their original floodplain due to the aggradation of legacy sediment. Currently, stream-floodplain reconnection is advocated as a stream restoration practice to take advantage of ecosystem services. The objective of this study is to compare two current stream restoration approaches for their nutrient flushing ability: 1) a conventional approach leaves legacy sediment on the floodplain; and 2) an ecological approach that involves removing the accumulated legacy sediment in order to restore the original floodplain surface wetland, revealing a buried A soil horizon. Soil cores were taken from the surficial legacy sediment layer and the buried A soil horizon in the floodplain of a 550-meter reach of Stroubles Creek in the Valley and Ridge province near Blacksburg, VA, to evaluate potential for flushable DOC, TDN, NO3-, NH4+, and SRP content. In addition, an inundation model was developed to evaluate the extent of flooding under the two restoration scenarios. The inundation model results and nutrient flushability levels were then used to simulate the release of nutrients as a function of stream restoration approach. Results indicate that the buried A horizon contained less flushable nutrients, but the ecological restoration would have a higher frequency of inundation that allows for more flushable nutrient release at the annual scale. Understanding the nutrient release potential from the floodplain will provide the ability to estimate net nutrient retention in different stream-floodplain reconnection strategies. / Master of Science

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第17867号 / 工博第3776号 / 新制||工||1577(附属図書館) / 30687 / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 中川 一, 教授 戸田 圭一, 准教授 川池 健司 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第20679号 / 工博第4376号 / 新制||工||1680(附属図書館) / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 寶 馨, 教授 立川 康人, 准教授 佐山 敬洋 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Modeling the movement of flood waters can be accomplished using many different methods with varying degrees of physical detail. Numerical models utilizing simple routing methods or simplified versions of the Navier-Stokes equations can be used to improve the public's flood preparedness. Three numerical models are used in this thesis to investigate flood preparedness: (1) an existing HEC-ResSim model of Coralville Reservoir, (2) an existing one-dimensional HEC-RAS model of the Iowa River through Coralville and Iowa City, and (3) a coupled one/two-dimensional hydraulic MIKE Flood model of the Cedar River through Cedar Falls/Waterloo. The HEC-ResSim model of Coralville Reservoir, provided by the United States Army Corps of Engineers, requires reservoir elevation-storage curves, inflow hydrographs and user-defined operation rules. This model utilizes level pool routing to determine changes in reservoir water levels and attenuation of hydrographs. The Muskingum routing method is used to route controlled releases downstream and determine satisfaction of constraints. The model is used to determine the impact of operational changes and sedimentation effects on historic flood events. Simulations indicate sedimentation has no effect on peak discharges of extreme events, but more aggressive operations plans may provide additional storage prior to extreme events. The existing HEC-RAS of the Iowa River through Iowa City is used to develop a library of inundation maps to be hosted on the National Weather Service Advanced Hydrologic Prediction Service's river forecast website. The modeling method assumes steady gradually varied flow. Post-processing and visualization of simulation results are completed using a digital elevation map of the study area developed using topography, bathymetry, and structural elevations. A coupled one/two-dimensional MIKE Flood model is developed for the Cedar River through Cedar Falls/Waterloo using topography, bathymetry, land use, and structural data. The river channel is modeled using MIKE 11, a one-dimensional unsteady hydraulic model, while the flood plain is modeled using MIKE 21, a two-dimensional hydraulic model utilizing depth-averaged Navier-Stokes equations. The model is used to develop a sequential levee closure plan for downtown Waterloo and will also be used to develop a library of inundation maps.

Two kinds of tsunami models are used in this thesis to simulate tsunami propagation in the ocean. One is the linear dispersion tsunami model developed by Port and Airport Research Institute (PARI), Japan. The other is COrnell Multigrid COupled Tsunami model (COMCOT) developed by the School of Civil and Environmental Engineering, Cornell University, that carries on the tsunami run-up computation to the nearshore region. Two kinds of tsunami models have the same mechanism of initial wave profile, which is the vertical seabed displacement as the initial tsunami profile proposed by Mansinha and Smylie (1971). Both models describe the tsunami by the same shallow water equations. At first, the feasibility of the PARI model is established by comparing with the record in Maldives during the South Asia tsunami in December 2004. Then, the COMCOT model in applied to the Pingtong earthquake in December 2006 and is validated by comparing with the tidal station records. Possible submarine fault activities around Taiwan and the Western Pacific ring is simulated by the The PARI model based on moment magnitude scale (¢Ûw) close to the South Asia tsunami. Seven sources are chosen: the Hokkaido, the East Japan, the Ryukyu Islands, the Guishan Island in Taiwan, the Fukien of mainland China, the Luzon Island and the New Guinea. The results suggest the northeast and southwest part of Taiwan have potential tsunami risk. Finally, we simulate the fault activity between Taiwan and Luzon islands by the COMCOT model. The inundation area extends northward to the Tso-Ying and San-Min districts, eastward to the Siao-Gang district and Fengshan city. The Kaohsiung harbor can resist tsunami hazard for moment magnitude scale (¢Ûw) up to 7.58 with maximum wave height of 5.5 meters.

Using hydraulic modeling and Geographic Information System (GIS) software, the 100-year flood was delineated for the municipality of Avon Park located in Highlands County, Florida. A detailed and rigorous approach was undertaken to first collect and develop an extensive spatial database to store the data collected that is pertinent to the model. This analysis combined ArcGIS version 8.3 and the Interconnected Channel and Pond Routing (ICPR) model version 3.02 to develop hydraulic models that assigned regulatory flood elevation within the watershed. The model results were post processed and brought into GIS to delineate the 100-year flood.The steps outlined in this thesis with respect to the use of GIS as a tool in model pre and post-processing are applicable to many models. Hence, the methodology outlined in this thesis adds to the existing pool of knowledge about the use of GIS in hydraulic modeling. By documenting all the steps related to data acquisition, data processing and manipulation, the model interface and GIS, and the post processing of the model results, this thesis can serve as resources for future studies that utilize both GIS and hydraulic modeling.

Resumo: Este trabalho pretende contribuir para o entendimento do comportamento temporal dos habitats do sistema rio-planície do alto curso do Rio Paraná, último trecho livre de represamento desse rio em território brasileiro. Buscou-se avaliar os processos de inundação e as alterações na cobertura vegetal. A avaliação dos processos de inundação dos ambientes da planície de inundação, bem como os fluxos das águas e identificação das áreas alagadas, foi realizada através da análise de uma modelo que represente as variações topográficas e análise de imagens multiespectrais. Para tanto, foram utilizados dados do modelo digital de elevação do SRTM, aos quais foram aplicados diversos processamentos como interpolação baseada em geoestatística e modelagem da superfície de tendência para eliminar ruídos e melhorar a resolução espacial. Sobre o MDE realizou-se as simulações dos níveis de inundação baseadas no fatiamento do produto numérico de terreno. Essa informação também serviu para entrada na classificação da cobertura vegetal. Outro tipo de representação do terreno utilizado no trabalho foi gerado a partir de procedimentos fotogramétricos sobre imagens digitais aéreas adquiridas em um aerolevantamento, realizado... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: This work attempts to contribute to understanding the temporal habitats behavior of Upper Paraná's floodplain. For that, it was evaluated the flood processes and alterations in the floodplain vegetation. The evaluation of the flood processes and the flows of waters and identification of flooded areas, it was accomplished through SRTM topographical surface. The geostatistic and trend analysis were applied in the digital elevation model (DEM) to noise elimination and improve spatial resolution. The surface was sliced to simulate the floods levels. The digital elevation model was also used in the vegetation classification. Another type of digital elevation model was produced based on photogrammetry processing on aerial digital images, acquired in September of 2008. The temporal comparison was adopted to understand the relations among hydrodynamics and antropic use on the spatial patterns of the vegetable covering. For that, the vegetable covering mapping was accomplished with TM and MSS multispectral images and altimetry data. For the temporal mapping, geometric and radiometric transformations on images data were necessary, like georeferencing, atmospheric correction and radiometric normalization. / Orientador: Nilton Nobuhiro Imai / Coorientador: Paulo Cesar Rocha / Banca: João Osvaldo Rodrigues Nunes / Banca: Edvard Elias de Souza Filho / Mestre