O uso de imagens de satélite tem aprimorado o estudo da dinâmica do uso do solo. Os mapeamentos da cobertura da terra se tornaram instrumentos fundamentais na avaliação das alterações na paisagem provocadas pela ação antrópica e sua influência sobre o planejamento regional e urbano. A utilização de ferramentas computacionais como os Sistemas de Informação Geográfica e os modelos hidrológicos têm auxiliado essas avaliações. O modelo hidrológico SWAT (Soil and Water Assessment Tool), desenvolvido pelo Departamento de Agricultura norte-americano, é um modelo de base física e parâmetros distribuídos, auxiliado por uma interface gráfica de software GIS. Por seu valor ambiental e econômico, a caracterização ambiental da área de drenagem da Represa de Itupararanga, com 936,54 \'KM POT.2\', localizada no estado de São Paulo, tem fundamental importância e a utilização de dados atuais permite que o estudo da área seja mais preciso no que diz respeito aos seus resultados. A partir de imagens do sensor Liss 3 do satélite ResourceSat-1, foi produzido o mapa atual de uso e ocupação do solo através da classificação supervisionada de máxima verossimilhança, validada por coletas de campo de ponto de controle com índice Kappa de 0,64 e índice de exatidão global de 71%. Os parâmetros morfológicos demonstraram que a área de estudo é bem drenada e possui baixa probabilidade de enchentes. A modelagem hidrológica foi calibrada para o período de janeiro de 2005 a dezembro de 2008 e resultou em coeficiente de eficiência de 0,41 e tendência percentual de 0%. A validação alcançou resultado de 0,301 e 5,5% para estes avaliadores, valores de ajuste considerados aceitáveis. Foram encontrados os valores de 9,66 e 1,5 para nitrogênio e fósforo, respectivamente, em kg/ha.ano, para os valores de poluição difusa, resultados elevados na comparação entre pesquisas com este foco. A comparação entre cenários identificou a necessidade de adotar práticas conservacionistas através do planejamento da ocupação para tornar sustentáveis as atividades dentro bacia e atenuar as pressões sobre os recursos naturais. / The use of satellite images has improved the study of the land use dynamics. Land cover mapping have become fundamental tools in the assessment of landscape changes caused by human action and its influence on regional and urban planning. Computation tools such as Geographic Information Systems and hydrological models have supported these findings. The hydrological model SWAT (Soil and Water Assessment Tool), developed by U. S. Agriculture Department, a model physically based and semi-distributed parameters with a GIS graphical interface. Due your environmental and economic value, the morphologic characterization of Itupararanga watershed, with 936,54 \'KM POT.2\', located in São Paulo state, have fundamental relevance and use of current data alows the study be more accurate. From images of Liss 3 sensor of ResourceSat-1 satelite, the current land use map was produced through supervised classification by maximum likelihood method and validated by control points on field collection, resulting in Kappa coefficient of 0.64 and overall accuracy rate of 71%. The morphological parameters showed that watershed is well drained and has low probability of flooding. The hydrological modeling was calibrated from January of 2005 to December of 2008 and had efficient value of 0.41 and trend rate of zero. Validation reached results of de 0.301 and 5.5% for these coefficients, considered acceptable for adjust values. Were found values of 9.66 and 1.5 for nitrogen and phosphorus, in kg/ha.year, respectively, for the values of diffuse pollution, high results in comparison between studies with this focus. The comparison between scenarios indentified the need to adopt conservation pratices through occupation planning to make the activities inside the basin and mitigate the pressure on natural resources.

La modélisation hydrologique permet de quantifier la transformation pluie-débit au sein d’un bassin versant. Bien que les modèles parviennent généralement à représenter de manière acceptable le fonctionnement des bassins versants, cette représentation, nécessairement simplifiée, reste imparfaite, et une quantification des incertitudes est souhaitable. Cette thèse s’intéresse à la quantification de l’incertitude prédictive en modélisation hydrologique. Le principal objectif de nos travaux est d’explorer différentes méthodes qui permettent d’associer à des simulations ou des prévisions de débits déterministes des distributions probabilistes. Nous distinguons le contexte de simulation du contexte de prévision et adoptons dans ces deux cas une démarche comparative et pragmatique qui permet d’évaluer différentes approches sur un large échantillon de bassins versants français, à l’aide de critères d’évaluation adaptés. En simulation, nos travaux ont porté sur deux méthodes liées à l'estimation des paramètres des modèles hydrologiques, la méthode GLUE et le calage bayésien, ainsi que sur deux approches plus pragmatiques, l'approche multi-modèles, et le post-traitement statistique. Nos résultats suggèrent que les approches telles que GLUE qui ne s'appuient que sur un ensemble de différents jeux de paramètres ne parviennent pas, en général, à représenter de manière adéquate l'incertitude prédictive totale. L'utilisation d'un modèle d'erreur extérieur au fonctionnement interne du modèle hydrologique est nécessaire. Les méthodes de post-traitement suffisamment flexibles pour caractériser les erreurs résiduelles obtenues en calage parviennent à refléter de manière plus satisfaisante les marges d'erreurs du modèle hydrologique utilisé. Nous proposons également une méthode qui permet d'obtenir une estimation de l'incertitude prédictive pour les bassins non jaugés, au moyen d'un transfert des marges d'erreurs constatées sur les bassins jaugés. Les résultats indiquent que la méthode est prometteuse et fournit dans la plupart des cas des intervalles de confiance fiables et fins sur les sites non-jaugés. En prévision, nos travaux ont porté d'une part sur la comparaison de différentes méthodes de post-traitement statistique, et d'autre part sur l'interaction entre l'assimilation de données et le post-traitement au sein d'une chaîne de prévision hydrologique d'ensemble. Les résultats obtenus montrent l'importance de la prise en compte de l'évolution de l'incertitude prédictive en fonction de l'échéance de prévision et mettent en évidence les gains de performance qui peuvent être obtenus quand la quantification de l'incertitude s'appuie sur une meilleure caractérisation de la situation de la situation de prévision. Enfin, nos travaux indiquent que l'utilisation conjointe de l'assimilation de données et d'une méthode de post-traitement permet d'améliorer les performances d'une chaîne de prévision hydrologique d'ensemble. / Hydrological modelling aims to quantify the rainfall-runoff relationship at the catchment scale. Although hydrological models are generally able to provide an acceptable representation of the catchment behaviour, this representation is necessarily simplified and imperfect, as a consequence an evaluation of uncertainties is desirable. This thesis focuses on the quantification of predictive uncertainty in hydrological modelling. Our main objective was to explore several methods that can be used to complete the deterministic predictions of a rainfall-runoff model with probabilistic distributions. Two prediction cases were distinguished, namely simulation and forecast. We set up a comparative framework to evaluate different uncertainty quantification methods on a large set of French catchments. In simulation mode, we focused on two methods related to the parameter estimation issue, the GLUE uncertainty method and the Bayesian framework, as well as two more pragmatic approaches, a multi-model approach and the post-processing method. Our results suggest that the GLUE-like methods based on ensembles of various parameter sets do not provide an adequate representation of the total predictive uncertainty. An external model error is necessary. Post-processing methods that are flexible enough to adequately describe the residual errors of the hydrological model obtained during calibration give more satisfactory estimates of the modelling uncertainty. We also present a method able to transfer model uncertainty estimates from gauged to ungauged catchments. Our results demonstrate that the method holds interesting perspectives, providing in most cases reliable and sharp uncertainty bounds at ungauged locations. In a forecasting context, we compared several post-processing methods and evaluated the interactions between data assimilation and post-processing in an ensemble forecasting modelling chain. Results show the crucial role of the lead time on the estimates of predictive uncertainty. They also suggest that some improvement can be achieved when the evolution of flows is better taken into account by a post-processing method. Last, we investigated the interactions between data assimilation and post-processing in hydrological ensemble forecasting and showed the benefits of using both in an ensemble forecast system.

De nombreuses études portant sur les impacts du changement climatique sur la ressource en eau ont été réalisées à l'échelle des grands bassins versants continentaux, celle des petits bassins versants demeurant peu étudiée. La région Bourgogne se situe à la croisée de plusieurs bassins hydrographiques (Seine, Loire, Rhône), et offre une grande diversité de contextes hydrogéologiques. Les impacts du changement climatique sont susceptibles de varier localement et leur détermination requiert l'implémentation d'une chaîne de modélisation hydroclimatique à l'échelle locale. L'objectif de cette thèse a consisté en la mise en place d'un tel outil de modélisation à l'échelle des petits bassins versants de Bourgogne. Un échantillon de bassins versants et deux modèles hydrologiques de conceptualisation différente ont été choisis : un modèle pluie-débit global (GR4J) et un modèle semi-distribué à base physique (SWAT). Ces modèles ont permis d'approcher les différents aspects des impacts sur la ressource en eau (débit à l'exutoire, ressource spatialisée, qualité de l'eau). Les modèles ont été calibrés, et leur robustesse a été testée en contexte de changement climatique, validant ainsi leur utilisation sur une période future. Une désagrégation spatiale des données climatiques a été nécessaire afin d'alimenter les modèles hydrologiques sur la période de projection. L'analyse de données observées a permis de mettre en évidence une rupture dans les températures de surface en 1987/88 (en Bourgogne comme en France métropolitaine), et une modification des régimes hydrologiques de part et d'autre de cette rupture. Les projections réalisées sur la période 1980-2100 ont permis de montrer : i) peu d'évolution des débits moyens annuels mais une accentuation de la saisonnalité, via une augmentation des débits hivernaux et une diminution des débits estivaux, en lien avec le régime des précipitations et avec l'augmentation de l'évapotranspiration. ii) des indicateurs d'étiages montrant une aggravation de leur sévérité et une emprise temporelle plus grande durant l'année hydrologique, iii) une qualité de l'eau dégradée. Afin d'ouvrir de nouvelles perspectives grâce à cette chaîne de modélisation (en termes aide à la décision), l'aptitude de la chaîne de modélisation à évaluer les impacts de divers scénarios de changement de pratiques agricoles a été testée. / Numerous studies dealing with climate change impacts on water ressources have been done at global scale, but the moderated size watersheds scale is less studied. The Burgundy region is located over several hydrographical basins (Seine, Loire and Rhône river), and provides a great diversity of hydrogeological and climate contexts. For these reasons, impacts of climate change on water resources are expected to be very different in space, and require an entire hydroclimatic modelling chain at local scale to be assessed. The main objective of this thesis consisted of the implementation of a modelling tool at the moderated-size watersheds scale of Burgundy. In this way, several watersheds and two hydrological models have been chosen : a global and empirical rainfall-runoff model (GR4J) and a physical based and semi-spatialised model (SWAT). These two models allow us to assess every aspects of climate change impacts : quantitative (streamflow and spatialised ressource) and qualitative impacts.Models have been calibrated, including a cross calibration/validation test in climate contrasted periods, to confirm their robustness. A disaggregation of global climate data has been done in order to feed models during projection.Observations analysis highlighted a climate shift (in Burgundy as well as over the entire french territory), resulting in a modification of hydrological regime. Models projection over the XXIth century showed i) not much evolution of annual streamflow, but an accented seasonality, with an increase of winter streamflow, and a decrease of summer streamflow, linked with the rainfall regime and the evapotranspiration increase, ii) drought flow indices showing an increase of severity of the drought flow period, iii) a decrease in water quality. In order to explore the capacities of the modelling chain, several sensibility tests have been done, based on management practices scenarios.

O modelo Soil and Water Assessment Tool (SWAT) tem sido utilizado para avaliar os impactos do uso e manejo da terra nos recursos hídricos, sedimentos e agroquímicos em diversas escalas e condições ambientais em todo o mundo. Entretanto, pelo fato de ter sido desenvolvido em centros de pesquisa norte-americanos, alguns parâmetros e rotinas de simulação não refletem adequadamente determinados processos de bacias localizadas em regiões tropicais. Nesse sentido, o presente trabalho visou aprimorar a modelagem hidrológica do SWAT através da revisão e modificação de processos relacionados à simulação da evapotranspiração. Os procedimentos propostos incluíram a alteração das rotinas de dormência vegetal no código fonte do modelo e a modificação dos cronogramas de operações de manejo e parâmetros do banco de dados de crescimento das plantas, visando reproduzir com maior precisão o ciclo das culturas em bacias tropicais. As modificações foram testadas em cinco bacias localizadas no Estado de São Paulo, com áreas entre 42 e 5.959 km², sendo comparados os resultados obtidos antes e depois da implementação das mesmas. Com as alterações, a análise do balanço hídrico anual evidenciou um aumento nos valores de evapotranspiração de cerca de 61% nas bacias, aproximando-se dos totais anuais de evapotranspiração calculados através de métodos empíricos, bem como houve redução significativa do escoamento superficial. Verificou-se uma melhoria da simulação de vazão em todas as bacias, sendo obtidos valores superiores para o Coeficiente de Eficiência de Nash-Sutcliffe (NSE) quando comparados à simulação sem as alterações. A calibração e validação foram realizadas com base na simulação modificada, sendo obtidos valores de NSE mensais entre 0,71 e 0,93 na calibração e 0,53 e 0,88 na validação, enquanto os valores diários de NSE situaram-se entre 0,51 e 0,82 na calibração e 0,38 e 0,83 na validação. A calibração a partir de uma simulação na qual as distorções dos processos hidrológicos da bacia estivessem previamente minimizadas resultou em bons resultados sem alteração excessiva dos parâmetros, indicando uma simulação hidrológica de melhor consistência. / The Soil and Water Assessment Tool (SWAT) has been used to predict the impact of land management practices on water, sediment, and agricultural chemical yields in a wide range of scales and environmental conditions across the globe. However, originally developed in the United States, some parameters and routines are unrealistic for simulating in tropical watersheds. In this sense, this work aims to improve the hydrologic modeling of SWAT model by reviewing and modifying parameters and routines related to evapotranspiration process. In order to adequately represent crop growth in tropical basins, the proposed procedures included changes in dormancy routines of SWAT source code and modifications of scheduled management operations and plant growth database parameters. These modifications were tested in five different basins at São Paulo State, Brazil, with areas ranging from 42 to 5959 km², by comparing the results before and after their implementation. Annual water balance analysis showed an increase in evapotranspiration about 61% for basins, approaching the total annual evapotranspiration estimated by empirical methods. Hence, it was observed that surface runoff and base flow components showed a decrease. The modifications resulted in improved flow simulation for all basins, showing better Nash-Sutcliffe Efficiency Coefficient (NSE) values compared to the unchanged simulation. Calibration and validation processes used the modified simulation database, being achieved monthly NSE between 0.71 – 0.73 at calibration and 0.53 – 0.88 at validation, while daily NSE were 0.51 – 0.82 at calibration and 0.38 – 0.83 at validation. Overall, minimizing distortions in hydrological processes at pre-calibration step resulted in good estimations without excessive modification of parameters at calibration, attesting a consistent hydrological modeling for the basins analyzed.

El conocimiento de los procesos hidrológicos es esencial para la gestión de los recursos hídricos tanto desde el punto de vista cuantitativo (crecidas o sequías) como desde el punto de vista cualitativo (contaminación). El funcionamiento hidrológico de las cuencas mediterráneas es aún bastante desconocido a pesar de los diferentes estudios realizados desde hace una veintena de años. Los progresos realizados en la identificación y modelización de los procesos hidrológicos corresponden casi en la totalidad a investigaciones realizadas en clima templado-húmedo (Bonell y Balek, 1993; Buttle, 1994). Esta falta de información, fuerza según Bonell (1993) a la "transferencia de resultados", a pesar de la necesidad evidente de desarrollar aproximaciones diferentes, principalmente en el ámbito de la modelización (Pilgrim et al. 1988). Por lo que se refiere a la modelación hidrológica, los estudios disponibles (Durand et al., 1992; Parkin et al., 1996; Piñol et al., 1997 entre otros) muestran serias dificultades para reproducir las primeras crecidas de otoño, después del periodo estival seco. Para estas cuencas parece difícil modelizar correctamente uno o más años hidrológicos completos con un solo juego de parámetros (Piñol et al., 1997, Bernal et al., 2004). El clima mediterráneo está caracterizado por una dinámica estacional muy marcada del régimen de precipitaciones y de la evapotranspiración, que favorece la alternancia durante el año de periodos secos y húmedos. Esto modifica fuertemente el estado hidrológico de la cuenca, de lo que deriva un comportamiento hidrológico complejo y no-lineal (Piñol et al. 1999). La necesidad de comprender el funcionamiento hidrológico de un sistema responde a dos cuestiones importantes: por un lado es el procedimiento más indicado para proporcionar elementos útiles a la gestión integrada de los recursos hídricos y por otro lado es fundamental para la modelación del comportamiento de nutrientes por ejemplo como el nitrato. / Medici ., C. (2010). Progressive development of a hydrologic and inorganic nitrogen conceptual model to improve the understanding of small Mediterranean catchments behaviour [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/8428 / Palancia

The natural movement of water is known as the hydrological cycle. Many different factors affect the hydrological cycle and the availability of freshwater. The availability of freshwater is essential for many human activities. Three of the UN SDG’s (Sustainable Development Goals), number 2, 6 and 7, directly or indirectly relate to freshwater supplies. SDG number 2, zero hunger; a part of fulfilling this goal is to have enough available freshwater to irrigate crops. SDG number 6 is about access to clean water and sanitation. It is possible to desalinate saline water, treat unclean water or transport water long distances, but these are often costly and energy demanding processes. Therefore, it is important to manage the freshwater supplies that are available wisely. Access to clean and affordable energy is SDG number 7. One source of clean and affordable energy is hydropower, in order to produce electricity in a hydropower plant, plenty of water and a change in potential energy is needed . To be able to fulfil all these goals, it is important to include all of the perspectives when making policies regarding agriculture, building hydropower plants or making big infrastructure changes. The complex relationships between the different areas of interest can make it complicated to analyse the effects of a change made in any area. The concept of connecting different perspectives is called Nexus. This aim of this thesis was to evaluate different computer-based hydrological modelling tools, and how they can be used for Nexus assessments. This was done by performing a market research, choosing two tools for further assessment, creating evaluation criteria and perform a case study on a watershed in Uganda. The case study results and the evaluation criteria were then compared to an existing model in WEAP. The two selected hydrological modelling tools were MIKE SHE and SWAT. These two modelling tools were evaluated based on specified evaluation criteria. In the case study they modelled the same area in Kamwenge, Uganda. The output of the models was compared and calibrated against observed flow in the river at the outlet of the watershed. The results of the case study were incomplete since the MIKE SHE model was not calibrated successfully. The different models have different strengths. The format of output data and flexibility of the program is superior in MIKE SHE, while SWAT is more intuitive and demands less computer power. Over all SWAT is easier to use and model in for a novice user, while MIKE SHE requires more expertise in order to run a successful model. SWAT is able to model more of the Nexus perspectives and are therefore the preferred model. / Många faktorer påverkar tillgängligheten av sötvatten som är väsentligt för många mänskliga aktiviteter. Tre av FN:s globala hållbarhets mål, nummer 2, 6 och 7 relaterar direkt eller indirekt till tillgång till sötvatten. Mål nummer 2, ingen hunger; en del av att uppfylla detta mål är att det måste finnas vatten för att bevattna grödor. Mål nummer 6, tillgång till rent vatten och sanitet; det är möjligt att rena vatten eller att transportera det längre distanser, men det är ofta energikrävande och kostsamt. Därför är det viktigt att förvalta tillgängliga sötvatten resurser klokt. Hållbar energi för alla är mål nummer 7. En källa till hållbar och ren energi är vattenkraft, för att producera vattenkraft behövs en skillnad i potentiell energi och mycket vatten. För att kunna uppfylla alla dessa mål är det viktigt att inkludera alla perspektiv när man bestämmer policys för jordbruk, bygger vattenkraftverk eller gör andra stora ändringar i infrastrukturen. De komplicerade förhållandena mellan dessa olika intresseområden för vatten kan göra det svårt att förutse hur en förändring i en av dem påverkar de andra. Konceptet att koppla ihop olika perspektiv kallas Nexus. Målet med denna uppsats var att utvärdera olika hydrologiska modelleringsverktyg och hur de kan användas för Nexus bedömning. Det gjordes genom att en marknadsundersökning genomfördes och två modelleringsverktyg valdes ut. De två utvalda modelleringsverktygen var MIKE SHE och SWAT. Dessa två verktyg utvärderades baserat på specificerade kriterier. I fallstudien modellerade båda modelleringsverktygen samma område i Kamwenge, Uganda. De båda modellerna kalibrerades mot observerade utflöden från avrinningsområdet. Resultaten jämfördes med en existerande modell i WEAP. Resultaten från fallstudien är ofullständiga då kalibreringen för MIKE SHE modellen inte lyckades.

Water is ubiquitous on our planet and constitutes a vital part of ecosystems. It supports the life of all beings on the earth while simultaneously evokes water-related issues such as water shortage, water contamination. As UN advocates, a globally shared blueprint for available clean water is depicted in Sustainable Development Goals (SDGs). However, there still exists a gap between current water management situations and our sustainable goals Modelling based on Hydro-Meteorological Data provides a way to understand regional hydrological processes and monitor environmental chemistry changes, especially for anthropogenic pollution. Furthermore, hydrological models make it possible to predict changes in water quantity and quality, under the context of climate change. The study area of this project is located in the Kringlan catchment, Norrström basins, occupying an area of 54.5 km2. The local discharges merge into Rastälven river and flow to the east, eventually discharging into the Baltic Sea. This project builds up a water balance model based on the meteorological data in the time frame from 2011 to 2012. The water balance model is calibrated to accurately simulate realistic hydrological components interactions, during each process, various parameters have been tested and adjusted to improve model robustness. Meanwhile, the project tries to strike a balance between the complexity of the model and amount of time it takes to run the model. The calibrated model is also validated to ensure model performance using statistical analysis. Additionally, a particle tracking model for the saturated zone is developed on the basis of the water balance model. Chloride is chosen as the trace element due to its feature of unreactive in ecological systems. The model results could also provide a value to groundwater age estimation. Suggested by previous researches targeting the area, leakage from vegetation and forest soil in this catchment have contributed to imbalances in local Cl- budgets. An internal source of chloride from soil leaching is specified in the model at the same time with an external source from stream discharge. The coupled modelling through the application of MIKE SHE software and calibration process help us to understand dynamic processes of hydrological modelling and chloride particle transport in the Kringlan catchment. A future improvement to consider is extending the current model boundary to a larger area and introducing more reference data. It is also possible to establish a fully integrated solute transport model to investigate Chloride transport in the catchment. / Vatten ersätter en viktig del av ekosystemet men det framkallar vattenrelaterade problem som vattenbrist och vattenförorening samtidigt. Emellertid finns det fortfarande ett gap mellan nuvarande vattenhanteringssituationer och våra hållbara mål. Modellering baserad på meteorologiska data erbjuder en möjlighet att förstå regionala hydrologiska processer och övervaka förändringar av miljömässiga kemikalier, särskilt för antropogena föroreningar. Dessutom finns det en hög potential för att förutse förändringar i vattenmängd och kvalitet med hydrologiska modeller, i samband med klimatförändringar. Studieområdet ligger i Kringlans upptagningsområde som ett av Norrström basins, med en yta på 54,5 km2. De lokala utsläppen sammanfogas i Rastälven och strömmar österut, så småningom mynnar i Östersjön. Detta projekt bygger upp en vattenbalansmodell baserad på meteorologiska data inom tidsramen från 2011 till 2012. Vattenbalansmodellen är kalibrerad för att exakt simulera realistiska hydrologiska komponentinteraktioner. För att förbättra modellens robusthet har olika parametrar testats och anpassats under varje process. Samtidigt försöker projektet att hitta en balans mellan modellens komplexitet och hur lång tid det tar att driva modellen. En partikelspårningsmodell för den mättade zonen har utvecklats med utgångspunkt i vattenbalansmodellen. Klorid används som spårämne eftersom det är inert i ekologiska system. Modellsresultaten kan också ge ett värde för grundvattenberäkningen. Tidigare undersökningar inriktade på området föreslår att läckage från vegetation och skogsmark i detta avrinningsområde har bidragit till obalanser i lokala Cl- budgetar. Med hjälp av MIKE SHE modellen har vi undersökt dynamisk process för hydrologisk modellering och kloridpartikelspårning i Kringlan avrinningsområde. Vad som kan gör i framtiden är att förlänga den nuvarande modellgränsen till ett större område med mer referensdata. Det är också möjligt att upprätta en fullständigt integrerad lösningsmodell för att undersöka kloridtransporter i ett avrinningsområde.

The water-energy nexus holds paramount relevance in the context of the transition to a carbon free energy system, being water the only renewable energy source with reliable storage capacity. Modelling hydropower production in a large domain over a long time window represents an open challenge due to a variety of reasons: firstly, high-resolution, large-scale hydrological modelling in a context of uncertainty needs calibration, thus representing a computationally intensive task due to the large domain and time window over which calibration is needed; secondly, as stated by many works in literature, hydropower production modelling and in particular reservoir modelling is a very information-demanding procedure, and excessive simplifications adopted to face the lack of information might lead to consistent bias in the predictions. This thesis can be subdivided into three main parts: firstly, the model that was used to perform every analysis, HYPERstreamHS, will be presented. The model is a continuous, large-scale hydrological model embedding a dual-layer MPI framework (i.e. Message Passing Interface, a common standard in parallel computing) that ensures optimal scalability of the model, greatly reducing the computation time needed. Explicit simulation of water diversions due to hydropower production is also included in the model, and adopts only publicly available information, making the model widely applicable. Secondly, a first validation of the model will be presented, and the adopted approach will be compared with some other approaches commonly found in literature, showing that the inclusion of a high level of detail is crucial to ensure a reliable performance of the model; this first application was performed on the Adige catchment, where extensive information on human systems was available, and allowed to effectively assess which information were indispensable and which, in turn, could be simplified to some extent while preserving model performance. Finally, the model setup has been applied on a relevant portion of the Western Italian Alps; in this case, two different meteorological input forcing data sets were adopted, in order to assess the differences in their performance in terms of hydropower production modelling. This latter study indeed represents a preliminary analysis and will provide stepping stone to extend the modelling framework to the Italian Alpine Region.

No description available.

L'évolution constante des villes passe par l'urbanisation des zones encore disponibles induisant des effets sur les bilans hydriques des celles-ci. De plus, le changement climatique susceptible d'exacerber les extrêmes (dont les inondations) influence lui aussi ces bilans. La ville est donc un objet hydrologique spécifique qu'il faut replacer dans son contexte évolutif, ce qui élargit considérablement la gamme d'échelles spatio-temporelles à prendre en compte pour son analyse et sa simulation. L'Union Européenne considère que la gestion du risque d'inondation doit remplacer la défense classique contre celle-ci. Cette nouvelle approche est plus holistique : elle prend en compte toutes les composantes du risque et cherche a réduire la vulnérabilité des récepteurs (habitants, bâtiments et infrastructures). Elle débouche sur la question de la résilience des systèmes urbains où les technologies correspondantes doivent être intégrées en des systèmes résilients aux inondations. Il est donc indispensable de développer des outils permettant l'évaluation de la performance de ces derniers, et ce à différentes échelles. Ces préoccupations ont défini l'axe de développement de Multi-Hydro : faire interagir des modèles déjà éprouvés représentant une composante du cycle de l'eau, permettre d'effectuer ainsi des progrès substantiels dans la modélisation de l'eau en ville avec une facilité d'utilisation. Multi-Hydro est ainsi basé sur des équations physiques supportées par des modèles distribués couplés. Grâce à un outil SIG dédié, MH-AssimTool, les informations géographiques et physiques nécessaires à la modélisation sont facilement assimilés pour chaque zone et résolution. En effet, une attention particulière été portée sur les observables ayant le moins de dépendance en échelle. L'emploi d'outils d'analyse multi-échelles permet de représenter leur variabilité et de définir des paramétrisations robustes du fonctionnement hydrologique à différentes échelles. L'ensemble de ces développements a été utilisé pour aborder la question de la résilience face aux inondations à différentes échelles d'un système urbain, dans le cadre de différents projets européens (SMARTeST, RainGain, BlueGreenDream) ou nationaux (Ville Numérique), à l'aide d'une approche systémique sur des scénarios pour plusieurs cas d'étude :- un petit bassin versant de Villecresnes (Val-de-Marne) qui a servi a l'évaluation des impacts de chaque modification apportée au modèle au cours de son développement.- la partie Est de la commune de Saint-Maur-des-Fossés (Val-de-Marne) qui a subi un audit de l'état du réseau d'assainissement (cartographie précise des canalisations et campagnes de mesures) et dont les résultats préliminaires ont permis de poser la problématique de la modélisation des rivières.- un quartier d'Heywood (grande banlieue de Manchester, Royaume Uni), qui a subi plusieurs inondations durant la dernière décennie et demande une modélisation assez fine pour permettre l'évaluation de l'impact de quatre scénarios de protection.- le bassin versant de la Loup, dont l'exutoire est occupé par un bassin de stockage des eaux de pluie, a été modélisé pour quatre évènements d'intensités et de durées variables et a permis de débuter la validation du modèle.- la zone de Spaanse Polder (Rotterdam, Pays Bas), pose la problématique de la modélisation des zones très planes au système de drainage complexes (pompes, exutoires multiples). Cette zone permet de guider les développements futurs de Multi-Hydro. Dans le contexte de l'amélioration de la résilience des villes face aux inondations, Multi-Hydro se place comme étant un outil qui offre la possibilité de simuler des scénarios permettant l'évaluation des impacts à l'échelle globale de modifications à plus petites échelles. Grâce a sa facilité de mise en place que lui confère MH-AssimTool, ainsi que sa structure modulaire et sa liberté de licence, Multi-Hydro est en train de devenir un outil d'aide à la décision / The constant evolution of cities can be seen as the urbanisation of the still available areas. This introduces complex effects with respect to the balance of water. In addition, the highly variable nature of the climate and weather can easily exacerbate the extremes (including floods) thus influencing the water balance. The European Union considers that the management of flood risk is an appropriate strategy to replace conventional defense strategies against floods. This new strategy is a more holistic approach: it takes into account all the components at risk and seeks to reduce the vulnerability of receptors (people, buildings and infrastructures).Thus, resilience measures not only consist of individual technical solutions but they need to be integrated to a ‘safety chain', which requires the development of resilience systems and tools. It is therefore essential to develop tools for assessing the performance of the latter, and at different scales. These concerns have help define the development of Multi-Hydro: interacting models already proven to represent different components of the water cycle to allow substantial progress in the modelling of urban water combined with ease use. Multi-Hydro is based on physical equations supported by distributed and coupled models. With a dedicated GIS MH-AssimTool, the geographical and physical information required for modelling are easily assimilated for each zone and at each resolution. Indeed, special attention was paid to the observables with the least scale dependence. Tools for multi-scale analysis are used to represent their variability at smaller scales than their own scales, thus allowing a more robust definition of hydrological parameterisations at different scales. All of these developments have been used to address the issues involved in flooding resilience at different urban system levels, within the framework of the European (Smartest, RainGain and BlueGreenDream) and national (Ville Numérique) projects, using a systemic approach on the scenarios of several case studies:- A small watershed Villecresnes (Val-de-Marne), used to assess the impacts of each change made in the model during its development.- The eastern part of the municipality of Saint-Maur-des-Fossés (Val-de-Marne ), has undergone a state audit of the drainage network (precise mapping of pipes and measurement campaigns). The preliminary results helped raise the issue of modelling rivers.- A district at Heywood (suburbs of Manchester, UK), has suffered several floods over the last decade and requires more detailed modelling in order to allow for the assessment of impact of four protection scenarios.- The catchment area of the Loup, whose outlet is connected to a runoff water storage tank, was modelled over four events of varying durations and intensities and helped start the validation of the model.- The Spaanse Polder area (Rotterdam , Netherlands), poses the problem of modelling very flat terrain with a complex drainage system (pumps and multiple outlets). This area will help to guide the future development of Multi-Hydro. In the context of improving the resilience of cities to flooding, Multi-Hydro is therefore placed as a tool that provides the ability to simulate scenarios for impact assessment at the basin scale of changes to smaller scales. Due to its ease of implementation at various scales conferred by MH-AssimTool and its modular structure and its free access property, Multi-Hydro is becoming a support decision tool