No description available.

The number of flood events and the estimated costs of floods have increased dramatically over the past few decades. To reduce the negative impacts of flooding, reliable flood forecasting is essential for early warning and decision making. Although various flood forecasting models and techniques have been developed, the assessment and reduction of uncertainties associated with the forecast remain a challenging task. Therefore, this thesis focuses on the investigation of Bayesian methods for producing probabilistic flood forecasts to accurately quantify predictive uncertainty and enhance the forecast performance and reliability. In the thesis, hydrologic uncertainty was quantified by a Bayesian post-processor - Hydrologic Uncertainty Processor (HUP), and the predictability of HUP with different hydrologic models under different flow conditions were investigated. Followed by an extension of HUP into an ensemble prediction framework, which constitutes the Bayesian Ensemble Uncertainty Processor (BEUP). Then the BEUP with bias-corrected ensemble weather inputs was tested to improve predictive performance. In addition, the effects of input and model type on BEUP were investigated through different combinations of BEUP with deterministic/ensemble weather predictions and lumped/semi-distributed hydrologic models. Results indicate that Bayesian method is robust for probabilistic flood forecasting with uncertainty assessment. HUP is able to improve the deterministic forecast from the hydrologic model and produces more accurate probabilistic forecast. Under high flow condition, a better performing hydrologic model yields better probabilistic forecast after applying HUP. BEUP can significantly improve the accuracy and reliability of short-range flood forecasts, but the improvement becomes less obvious as lead time increases. The best results for short-range forecasts are obtained by applying both bias correction and BEUP. Results also show that bias correcting each ensemble member of weather inputs generates better flood forecast than only bias correcting the ensemble mean. The improvement on BEUP brought by the hydrologic model type is more significant than the input data type. BEUP with semi-distributed model is recommended for short-range flood forecasts. / Dissertation / Doctor of Philosophy (PhD) / Flood is one of the top weather related hazards and causes serious property damage and loss of lives every year worldwide. If the timing and magnitude of the flood event could be accurately predicted in advance, it will allow time to get well prepared, and thus reduce its negative impacts. This research focuses on improving flood forecasts through advanced Bayesian techniques. The main objectives are: (1) enhancing reliability and accuracy of flood forecasting system; and (2) improving the assessment of predictive uncertainty associated with the flood forecasts. The key contributions include: (1) application of Bayesian forecasting methods in a semi-urban watershed to advance the predictive uncertainty quantification; and (2) investigation of the Bayesian forecasting methods with different inputs and models and combining bias correction technique to further improve the forecast performance. It is expected that the findings from this research will benefit flood impact mitigation, watershed management and water resources planning.

O monitoramento, a previsão e o controle de eventos extremos, como as inundações, é imprescindível, principalmente em áreas urbanas, devido à maior densidade populacional, bens materiais, saneamento e infraestruturas envolvidos no processo. O objetivo deste estudo, consistiu em classificar de forma automática o uso e cobertura da terra em área urbana, em um ortofotomosaico com altíssima resolução espacial (16 cm), cobrindo a área do bairro centro de Lajeado (Estado do Rio Grande do Sul – Brasil) e sua posterior aplicação, com dados dos arruamentos e de um Modelo Digital de Elevação (MDE), para a estruturação de um framework automatizado, baseado em plataformas livres, com capacidade de monitorar níveis de inundações sobre essa área urbana, em escala espacial e temporal. Para a classificação do uso e cobertura da terra foram testados os classificadores por árvore de decisão Boosted C5.0, Random Forests e Classification and Regression Trees (CART). Primeiramente, foram identificadas as seguintes classes: vegetação arbórea; vegetação herbácea (gramíneas); solo exposto; sistema viário (calçamento); telhados metálicos e telhados cerâmicos claros; telhados de concreto e fibrocimento; telhados metálicos e cerâmicos escuros; e sombra. Por meio do programa eCognition foram aplicados sete níveis de segmentação do ortofotomosaico, coletadas as amostras e definidos os atributos para cada classe O treinamento para os classificadores foi realizado no programa R. Para a análise da exatidão de classificação, foram gerados pontos de checagem aleatórios, que foram comparados com as classes das três imagens classificadas, para o cálculo da matriz de erros e do índice Kappa. A imagem classificada pelo algoritmo Random Forests apresentou a maior Exatidão Global (EG = 82,20%) e Índice Kappa (K = 0,79), seguido pela imagem classificada pelo algoritmo Boosted C5.0 (EG = 80,4%; K = 0,77) e pelo CART (EG = 64,90%; K = 0,57). Já o framework foi baseado na equação de regressão fluviométrica Encantado/Lajeado. Os resultados dessa equação podem ser visualizados como mapas em uma interface WEBSIG, onde estão simuladas as áreas e infraestruturas inundadas no bairro centro de Lajeado. Foram projetados diferentes níveis históricos de inundações e esse modelo foi validado a partir da comparação dos dados simulados com os medidos de uma inundação ocorrida em 10 de outubro de 2015. O erro altimétrico obtido foi inferior a 1 m. O framework deste estudo realiza o monitoramento do nível de inundação para a área urbana de Lajeado com até 6 horas de antecedência, demonstrando a eficácia desta simulação. / The monitoring, forecasting and control of extreme events, such as floods, is essential, especially in urban settlements, due to the greater population density, material assets, sanitation and infrastructures on these areas. This work aims to classify automatically the urban land use and land cover in an orthophoto mosaic with very high spatial resolution (16 cm) covering the central district area of Lajeado (Rio Grande do Sul State - Brazil) and its subsequent application, with data of the streets and a Digital Elevation Model (MDE), for the structuring of an automated framework, based on free platforms, with capacity to monitor flood levels in this urban area, on a spatial and temporal scale. We tested the decision tree classifiers Boosted C5.0, Random Forests and Classification and Regression Trees (CART). First, the following classes of land use and land cover were identified: forest land; herbaceous land (grasses); bare soil; road system (pavement); metal roofs and clear ceramic roofs; concrete and fiber cement roofs; dark metallic and ceramic roofs; and shade. The eCognition software were used to processes seven levels of segmentation of the orthophoto mosaic, and for collecting samples and attributes from each one these classes The decision tree methods were performed in R software. For the classification accuracy assessment, we generated random check points, which were compared with the classes of the three classified images, in order to calculate the error matrix and Kappa index. The image classified by the algorithm Random Forests presented the highest Global Accuracy (GA = 82.20%) and Kappa Index ( = 0.79), followed by the image classified by the Boosted C5.0 (GA = 80.4%; = 0.77) and the CART algorithm (GA = 64.90%,  = 0.57). The Framework was based on the fluviometric regression equation Encantado/Lajeado. The results of this equation can be visualized as maps in a WEBGIS interface, where the flooded areas in the downtown neighborhood of Lajeado are simulated. Different historical flood levels were projected and this model was validated by comparing the simulated data with those measured from a flood occurred on October 10th, 2015. The altimetric error obtained was less than 1 m. The framework of this study carries out the monitoring of the flood level for the Lajeado urban area with 6 hours in advance, demonstrating the effectiveness of this simulation.

O monitoramento, a previsão e o controle de eventos extremos, como as inundações, é imprescindível, principalmente em áreas urbanas, devido à maior densidade populacional, bens materiais, saneamento e infraestruturas envolvidos no processo. O objetivo deste estudo, consistiu em classificar de forma automática o uso e cobertura da terra em área urbana, em um ortofotomosaico com altíssima resolução espacial (16 cm), cobrindo a área do bairro centro de Lajeado (Estado do Rio Grande do Sul – Brasil) e sua posterior aplicação, com dados dos arruamentos e de um Modelo Digital de Elevação (MDE), para a estruturação de um framework automatizado, baseado em plataformas livres, com capacidade de monitorar níveis de inundações sobre essa área urbana, em escala espacial e temporal. Para a classificação do uso e cobertura da terra foram testados os classificadores por árvore de decisão Boosted C5.0, Random Forests e Classification and Regression Trees (CART). Primeiramente, foram identificadas as seguintes classes: vegetação arbórea; vegetação herbácea (gramíneas); solo exposto; sistema viário (calçamento); telhados metálicos e telhados cerâmicos claros; telhados de concreto e fibrocimento; telhados metálicos e cerâmicos escuros; e sombra. Por meio do programa eCognition foram aplicados sete níveis de segmentação do ortofotomosaico, coletadas as amostras e definidos os atributos para cada classe O treinamento para os classificadores foi realizado no programa R. Para a análise da exatidão de classificação, foram gerados pontos de checagem aleatórios, que foram comparados com as classes das três imagens classificadas, para o cálculo da matriz de erros e do índice Kappa. A imagem classificada pelo algoritmo Random Forests apresentou a maior Exatidão Global (EG = 82,20%) e Índice Kappa (K = 0,79), seguido pela imagem classificada pelo algoritmo Boosted C5.0 (EG = 80,4%; K = 0,77) e pelo CART (EG = 64,90%; K = 0,57). Já o framework foi baseado na equação de regressão fluviométrica Encantado/Lajeado. Os resultados dessa equação podem ser visualizados como mapas em uma interface WEBSIG, onde estão simuladas as áreas e infraestruturas inundadas no bairro centro de Lajeado. Foram projetados diferentes níveis históricos de inundações e esse modelo foi validado a partir da comparação dos dados simulados com os medidos de uma inundação ocorrida em 10 de outubro de 2015. O erro altimétrico obtido foi inferior a 1 m. O framework deste estudo realiza o monitoramento do nível de inundação para a área urbana de Lajeado com até 6 horas de antecedência, demonstrando a eficácia desta simulação. / The monitoring, forecasting and control of extreme events, such as floods, is essential, especially in urban settlements, due to the greater population density, material assets, sanitation and infrastructures on these areas. This work aims to classify automatically the urban land use and land cover in an orthophoto mosaic with very high spatial resolution (16 cm) covering the central district area of Lajeado (Rio Grande do Sul State - Brazil) and its subsequent application, with data of the streets and a Digital Elevation Model (MDE), for the structuring of an automated framework, based on free platforms, with capacity to monitor flood levels in this urban area, on a spatial and temporal scale. We tested the decision tree classifiers Boosted C5.0, Random Forests and Classification and Regression Trees (CART). First, the following classes of land use and land cover were identified: forest land; herbaceous land (grasses); bare soil; road system (pavement); metal roofs and clear ceramic roofs; concrete and fiber cement roofs; dark metallic and ceramic roofs; and shade. The eCognition software were used to processes seven levels of segmentation of the orthophoto mosaic, and for collecting samples and attributes from each one these classes The decision tree methods were performed in R software. For the classification accuracy assessment, we generated random check points, which were compared with the classes of the three classified images, in order to calculate the error matrix and Kappa index. The image classified by the algorithm Random Forests presented the highest Global Accuracy (GA = 82.20%) and Kappa Index ( = 0.79), followed by the image classified by the Boosted C5.0 (GA = 80.4%; = 0.77) and the CART algorithm (GA = 64.90%,  = 0.57). The Framework was based on the fluviometric regression equation Encantado/Lajeado. The results of this equation can be visualized as maps in a WEBGIS interface, where the flooded areas in the downtown neighborhood of Lajeado are simulated. Different historical flood levels were projected and this model was validated by comparing the simulated data with those measured from a flood occurred on October 10th, 2015. The altimetric error obtained was less than 1 m. The framework of this study carries out the monitoring of the flood level for the Lajeado urban area with 6 hours in advance, demonstrating the effectiveness of this simulation.

Water resources management in the Sahel region of West Africa is extremely difficult because of high inter-annual rainfall variability as well as a general reduction of water availability in the region. Observed changes in streamflow directly disturb key socioeconomic activities such as the agriculture sector, which constitutes one of the main survival pillars of the West African population. Seasonal rainfall forecasting is considered as one possible way to increase resilience to climate variability by providing information in advance about the amount of rainfall expected in each upcoming rainy season. Moreover, the availability of reliable information about streamflow magnitude a few months before a rainy season will immensely benefit water users who want to plan their activities. However, since the 90s, several studies have attempted to evaluate the predictability of Sahelian weather characteristics and develop seasonal rainfall and streamflow forecast models to help stakeholders take better decisions. Unfortunately, two decades later, forecasting is still difficult, and forecasts have a limited value for decision-making. It is believed that the low performance in seasonal forecasting is due to the limits of commonly used predictors and forecast approaches for this region. In this study, new seasonal forecasting approaches are developed and new predictors tested in an attempt to predict the seasonal rainfall over the Sirba watershed located in between Niger and Burkina Faso, in West Africa. Using combined statistical methods, a pool of 84 predictors with physical links with the West African monsoon and its dynamics were selected, with their optimal lag times. They were first reduced through screening using linear correlation with satellite rainfall over West Africa. Correlation analysis and principal component analysis were used to keep the most predictive principal components. Linear regression was used to get synthetic forecasts, and the model was assessed to rank the tested predictors. The three best predictors, air temperature (from Pacific Tropical North), sea level pressure (from Atlantic Tropical South) and relative humidity (from Mediterranean East) were retained and tested as inputs for seasonal rainfall forecasting models. In this thesis it has been chosen to depart from the stationarity and linearity assumptions used in most seasonal forecasting methods: 1. Two probabilistic non-stationary methods based on change point detection were developed and tested. Each method uses one of the three best predictors. Model M1 allows for changes in model parameters according to annual rainfall magnitude, while M2 allows for changes in model parameters with time. M1 and M2 were compared to the classical linear model with constant parameters (M3) and to the linear model with climatology (M4). The model allowing changes in the predictand-predictor relationship according to rainfall amplitude (M1) and using AirTemp as a predictor was the best model for seasonal rainfall forecasting in the study area. 2. Non-linear models including regression trees, feed-forward neural networks and non-linear principal component analysis were implemented and tested to forecast seasonal rainfall using the same predictors. Forecast performances were compared using coefficients of determination, Nash-Sutcliffe coefficients and hit rate scores. Non-linear principal component analysis was the best non-linear model (R2: 0.46; Nash: 0.45; HIT: 60.7), while the feed-forward neural networks and regression tree models performed poorly. All the developed rainfall forecasting methods were subsequently used to forecast seasonal annual mean streamflow and maximum monthly streamflow by introducing the rainfall forecasted in a SWAT model of the Sirba watershed, and the results are summarized as follows: 1. Non-stationary models: Models M1 and M2 were compared to models M3 and M4, and the results revealed that model M3 using RHUM as a predictor at a lag time of 8 months was the best method for seasonal annual mean streamflow forecasting, whereas model M1 using air temperature as a predictor at a lag time of 4 months was the best model to predict maximum monthly streamflow in the Sirba watershed. Moreover, the calibrated SWAT model achieved a NASH value of 0.83. 2. Non-linear models: The seasonal rainfall obtained from the non-linear principal component analysis model was disaggregated into daily rainfall using the method of fragment, and then fed into the SWAT hydrological model to produce streamflow. This forecast was fairly acceptable, with a Nash value of 0.58. The evaluation of the level of risk associated with each seasonal forecast was carried out using a simple risk measure: the probability of overtopping of the flood protection dykes in Niamey, Niger. A HEC-RAS hydrodynamic model of the Niger River around Niamey was developed for the 1980-2014 period, and a copula analysis was used to model the dependence structure of streamflows and predict the distribution of streamflow in Niamey given the predicted streamflow on the Sirba watershed. Finally, the probabilities of overtopping of the flood protection dykes were estimated for each year in the 1980-2014 period. The findings of this study can be used as a guideline to improve the performance of seasonal forecasting in the Sahel. This research clearly confirmed the possibility of rainfall and streamflow forecasting in the Sirba watershed at a seasonal time scale using potential predictors other than sea surface temperature.

O monitoramento, a previsão e o controle de eventos extremos, como as inundações, é imprescindível, principalmente em áreas urbanas, devido à maior densidade populacional, bens materiais, saneamento e infraestruturas envolvidos no processo. O objetivo deste estudo, consistiu em classificar de forma automática o uso e cobertura da terra em área urbana, em um ortofotomosaico com altíssima resolução espacial (16 cm), cobrindo a área do bairro centro de Lajeado (Estado do Rio Grande do Sul – Brasil) e sua posterior aplicação, com dados dos arruamentos e de um Modelo Digital de Elevação (MDE), para a estruturação de um framework automatizado, baseado em plataformas livres, com capacidade de monitorar níveis de inundações sobre essa área urbana, em escala espacial e temporal. Para a classificação do uso e cobertura da terra foram testados os classificadores por árvore de decisão Boosted C5.0, Random Forests e Classification and Regression Trees (CART). Primeiramente, foram identificadas as seguintes classes: vegetação arbórea; vegetação herbácea (gramíneas); solo exposto; sistema viário (calçamento); telhados metálicos e telhados cerâmicos claros; telhados de concreto e fibrocimento; telhados metálicos e cerâmicos escuros; e sombra. Por meio do programa eCognition foram aplicados sete níveis de segmentação do ortofotomosaico, coletadas as amostras e definidos os atributos para cada classe O treinamento para os classificadores foi realizado no programa R. Para a análise da exatidão de classificação, foram gerados pontos de checagem aleatórios, que foram comparados com as classes das três imagens classificadas, para o cálculo da matriz de erros e do índice Kappa. A imagem classificada pelo algoritmo Random Forests apresentou a maior Exatidão Global (EG = 82,20%) e Índice Kappa (K = 0,79), seguido pela imagem classificada pelo algoritmo Boosted C5.0 (EG = 80,4%; K = 0,77) e pelo CART (EG = 64,90%; K = 0,57). Já o framework foi baseado na equação de regressão fluviométrica Encantado/Lajeado. Os resultados dessa equação podem ser visualizados como mapas em uma interface WEBSIG, onde estão simuladas as áreas e infraestruturas inundadas no bairro centro de Lajeado. Foram projetados diferentes níveis históricos de inundações e esse modelo foi validado a partir da comparação dos dados simulados com os medidos de uma inundação ocorrida em 10 de outubro de 2015. O erro altimétrico obtido foi inferior a 1 m. O framework deste estudo realiza o monitoramento do nível de inundação para a área urbana de Lajeado com até 6 horas de antecedência, demonstrando a eficácia desta simulação. / The monitoring, forecasting and control of extreme events, such as floods, is essential, especially in urban settlements, due to the greater population density, material assets, sanitation and infrastructures on these areas. This work aims to classify automatically the urban land use and land cover in an orthophoto mosaic with very high spatial resolution (16 cm) covering the central district area of Lajeado (Rio Grande do Sul State - Brazil) and its subsequent application, with data of the streets and a Digital Elevation Model (MDE), for the structuring of an automated framework, based on free platforms, with capacity to monitor flood levels in this urban area, on a spatial and temporal scale. We tested the decision tree classifiers Boosted C5.0, Random Forests and Classification and Regression Trees (CART). First, the following classes of land use and land cover were identified: forest land; herbaceous land (grasses); bare soil; road system (pavement); metal roofs and clear ceramic roofs; concrete and fiber cement roofs; dark metallic and ceramic roofs; and shade. The eCognition software were used to processes seven levels of segmentation of the orthophoto mosaic, and for collecting samples and attributes from each one these classes The decision tree methods were performed in R software. For the classification accuracy assessment, we generated random check points, which were compared with the classes of the three classified images, in order to calculate the error matrix and Kappa index. The image classified by the algorithm Random Forests presented the highest Global Accuracy (GA = 82.20%) and Kappa Index ( = 0.79), followed by the image classified by the Boosted C5.0 (GA = 80.4%; = 0.77) and the CART algorithm (GA = 64.90%,  = 0.57). The Framework was based on the fluviometric regression equation Encantado/Lajeado. The results of this equation can be visualized as maps in a WEBGIS interface, where the flooded areas in the downtown neighborhood of Lajeado are simulated. Different historical flood levels were projected and this model was validated by comparing the simulated data with those measured from a flood occurred on October 10th, 2015. The altimetric error obtained was less than 1 m. The framework of this study carries out the monitoring of the flood level for the Lajeado urban area with 6 hours in advance, demonstrating the effectiveness of this simulation.

Les modèles hydrologiques sont des outils indispensables pour calculer les débits a l’exutoire des bassins versants, la gestion des aménagements hydrauliques ou encore la prévision et la prévention des inondations. Les précipitations représentent la variable climatique principale à l’origine des débits des cours d’eau qui s’écoulent au sein d’un bassin versant. De ce fait, la réponse hydrologique du bassin est fortement dépendante de la représentativité des données d’entrée de précipitation.Les radars météorologiques, qui permettent aujourd’hui d’accéder a des mesures a haute résolution spatiale et temporelle des champs de précipitation, sont de plus en plus utilises dans le domaine de la prévision, pour le suivi des situations hydrométéorologiques. Cependant, la mesure des précipitations par radar est entachée d’erreurs qui peuvent affecter gravement la qualité des simulations de débit. De ce fait, l’utilisation des données de précipitations a haute résolution spatiale pour la modélisation hydrologique est souvent limitée par rapport a l’utilisation des données pluviométriques.Récemment, Météo-France a développe une réanalyse des lames d’eau au pas de temps horaire, sur une durée de 10 ans, en combinant l’ensemble des données de précipitation radar et pluviométriques : les mesures radars ont été corrigées et étalonnées avec le réseau de mesure au sol horaire et journalier. Dans cette thèse, nous proposons d’étudier l’intérêt de cette nouvelle base de données à haute résolution spatiale pour la modélisation pluie-débit.Dans un premier temps, nous avons développe et valide un modèle hydrologique semi-distribue qui a la capacité de fonctionner pour différentes résolutions spatiales, de la représentation globale jusqu’a une discrétisation spatiale très fine des bassins. Dans un deuxième temps, l’impact de la résolution spatiale des données d’entrée de précipitation sur la simulation des débits a été analysé. L’apport de l’information radar pour l’estimation des précipitations a été évalue par rapport a une utilisation exclusive des pluviomètres, par le biais de la modélisation pluie-débit en termes de précision des débits a l’exutoire des bassins. Enfin, le modèle semi-distribue TGR a été comparé avec le modèle global GRP actuellement opérationnel dans les Services de Prévision des Crues. L’originalité de notre travail réside sur l’utilisation de données d’observation sur un large échantillon de 181 bassins versants français représentant une grande diversité de tailles et conditions climatiques, ce qui nous permet d’apporter un diagnostic robuste et des éléments de réponse sur les problématiques scientifiques traitées. / Hydrologic models are essential tools to compute the catchment rainfall-runoff response required for river management and flood forecast purposes. Precipitation dominates the high frequency hydrological response, and its simulation is thus dependent on the way rainfall is represented. In this context, the sensitivity of runoff hydrographs to the spatial variability of forcing data is a major concern of researchers. However, results from the abundant literature are contrasted and it is still difficult to reach a clear consensus.Weather radar is considered to be helpful for hydrological forecasting since it provides rainfall estimates with high temporal and spatial resolution. However, it has long been shown that quantitative errors inherent to the radar rainfall estimates greatly affect rainfall-runoff simulations. As a result, the benefit from improved spatial resolution of rainfall estimates is often limited for hydrological applications compared to the use of traditional ground networks.Recently, Météo-France developed a rainfall reanalysis over France at the hourly time step over a 10-year period combining radar data and raingauge measurements: weather radar data were corrected and adjusted with both hourly and daily raingauge data. Here we propose a framework to evaluate the improvement in streamflow simulation gained by using this new high resolution product.First, a model able to cope with different spatial resolutions, from lumped to semi-distributed, was developed and validated. Second, the impact of spatial rainfall resolution input on streamflow simulation was investigated. Then, the usefulness of spatial radar data measurements for rainfall estimates was compared with an exclusive use of ground raingauge measurements and evaluated through hydrological modelling in terms of streamflow simulation improvements. Finally, semi-distributed modelling with the TGR model was performed for flood forecasting and compared with the lumped forecasting GRP model currently in use in the French flood forecast services. The originality of our work is that it is based on actual measurements from a large set of 181 French catchments representing a variety of size and climate conditions, which allows to draw reliable conclusions.