Добірка наукової літератури з теми "Rainfall events scale"

La région méditerranéenne est considérée une des zones les plus vulnérables au changement climatique.Plusieurs études élaborées sur l'évolution temporelle de différents paramètres météorologiques ont trouvé une variabilité climatiques significative à la fin du XXe siècle.En plus, d'après les modèles de projection climatique, il est attendu que les changements climatiques s'intensifient sur toute la région vers la fin du XXIe siècle.Par conséquence,leurs impacts deviennent de plus en plus dangereux et couteux. La pluie est considérée comme la signature la plus sensible du climat pour l'Homme.Ainsi,son analyse et la caractérisation des régime pluviométriques dans la région permet d'appréhender son évolution future. Cette recherche s'appuie surtout sur une base de données journalières des observations sur 70 stations pluviométriques sur une période de 50 ans (1960-2009) et sur une échelle régionale concernant le nord de la Tunisie. D'autre part, les pluies journalières calculées par des modèles de réanalyse ERA-Interim sont également considérées. Ces données ont l'avantage d'être complètes dans le temps et dans l'espace. Elles peuvent jouer un rôle important dans la compréhension de la variabilité climatique d'où il est primordial d'évaluer leur qualité par rapport aux observations. Le traitement des données pluviométriques est également inédit puisqu'il s'agit d'analyser la variabilité spatiale et temporelle à l'échelle de structures pluvieuses organisées en événements pluvieux. Bien que les précipitations, en Tunisie, ont été analysées par plusieurs hydrologues et géographes à différentes échelles allant de quelques minutes à des années, le découpage en épisodes pluvieux et épisodes secs proposé dans ce travail est original.Cette approche vise à prendre en considération le caractère intermittent de la pluie, qui est une des propriétés fondamentales des précipitations.L'agrégation des jours de pluie a conduit à envisager six descripteurs des événements pluvieux pour chaque station sur une période de 50 ans.L'espace multidimensionnel ainsi créé est analysé en première partie par une méthode factorielle classique (l'analyse en composantes principales ACP). Puis par la méthode de classification non-linéaire SOM (Self-Organizing Map) associée à une classification hiérarchique ascendante (CHA). Les deux approches ont permis de comprendre les structures des données et d'en définir une typologie. L'ACP a conduit à résumer les descripteurs de pluie adoptés en trois composantes principales : La première représente un indicateur de quantité de pluie, la seconde l'intermittence des pluies et la troisième est un indicateur de structure. L'interprétation spatiale a divisé la zone d'étude en trois régions d'orientation NE-SO, avec une opposition entre la façade Nord-Ouest et son arrière-pays et la façade Sud-Est et son arrière-pays avec une zone intermédiaire située entre ces deux régions. Par ailleurs, la thèse investigue les corrélations entre les composantes principales et des indices climatiques dérivant les modes de variabilité climatiques. Des corrélations significatives ont été mises en évidence pour les indices Oscillation nord Atlantique et Oscillation Arctique. Aussi, des liens entre les anomalies de pression à la surface de la mer et les composantes principales ont été montré par des une analyse composite. Les méthodes combinées (SOM et CHA) sont appliquées aux descripteurs de pluie associés au réseau de mesure et met en évidence 4 classes ayant des typologies différentes pour les structures d'épisodes pluvieux. Leur variabilité spatiale et temporelle a ensuite été analysée. Ces classes sont utilisées comme une référence pour l'analyse des données estimées par le modèle de réanalyse. La comparaison avec les observations a montré que les distributions statistiques n'ont pas la même asymétrie.Par contre, le modèle montre une bonne cohérence des structures temporelle des classes de pluie avec les observations à une échelle régionale<br>The Mediterranean region is considered one of the most vulnerable areas to climate change because of its socio-cultural wealth and its biodiversity. Several studies about the evolution of different parameters found a significant climate variability at the end of the 20th century. Moreover,according to climate projection models, an intensification of this climate change is expected, over the region, to the end of the 21th century.Consequently, their impacts become more dangerous and expensive. Rain is considered to be the most sensitive signature of climate for humans. Thus, its analysis and the characterization of rainfall regimes over the region allow to apprehend its future evolution. This study is mainly based on daily rainfall observations collected from 70 rain gauge stations over 50-year period (1960-2009) on a regional scale covering all of northern Tunisia. On the other hand, daily precipitations produced by ERA-Interim reanalysis model, equally at regional scale, are also considered in this study. These estimated data have the advantage of being complete over the time and the space. They can have an important role in understanding the climate variability, hence it is essential to assess their quality in relation to observations. The rainfall data processing is novel. Indeed, the analysis of spatial and temporal variability analysis was carried out on the rain event scale. Although rainfall in Tunisia has been analysed by several hydrologists and geographers at different scales ranging from a few minutes to years, the division into rainy episodes and dry episodes proposed in this analysis is original. This approach aims to take into account the intermittent nature of the rain which is one of the fundamental properties of the precipitations. The aggregation of rainy days led to consider six descriptors of rainy events for each measurement point over a period of 50 years. The multidimensional space, thus, created was analysed in the first part by a classic factorial method PCA (Principal Component Analysis), then by the non-linear classification method SOM (Self-Organizing Map) combined with Hierarchical agglomerative clustering (HAC). The two approaches allowed to understand the rainfall data structures and to define a typology. The principal component analysis summarized the six rainfall descriptors adopted into three main components: the first one is an indicator of the rainfall quantity, the second one represents the intermittent character of rain over the season and the third one is a structure indicator. The spatial interpretation divided the study area into three regions of NE-SW orientation, with an opposition between the North-West facade and its hinterland and the South-East facade and its hinterland with an intermediate zone located between these two regions. Moreover, the thesis investigated the correlations between the principal components of PCAs and climate patterns indices. Significant correlations were found for the North Atlantic Oscillation and Arctic Oscillation indices. In addition, relationships between sea surface pressure anomalies and principal components were shown by a composite analysis. The combined methods (SOM and HAC) were applied to rainfall descriptors produced by the rain gauge stations network and highlighted 4 classes with different typologies of wet spells structure. Their spatial and temporal variability was, then, analysed. These classes were used as a reference for the analysis of the reanalysis data.The univariate and multivariate analysis of the model data and the comparison with the observations showed that the number of rainy days and the duration of the events are significantly overestimated in the reanalyses. Moreover, the statistical distributions didn't have the same asymmetry. On the other hand, the model showed a good coherence of the temporal structures of the rainfall classes with the observations on a regional scale

Submitted by Maike Costa (maiksebas@gmail.com) on 2016-01-05T12:43:51Z No. of bitstreams: 1 arquivototal.pdf: 7469211 bytes, checksum: 99e23a5e3e827ec0fdf765d03f0f8601 (MD5)<br>Made available in DSpace on 2016-01-05T12:43:51Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 7469211 bytes, checksum: 99e23a5e3e827ec0fdf765d03f0f8601 (MD5) Previous issue date: 2015-03-27<br>Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq<br>Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES<br>The assessment of rainfall-runoff transformation mechanisms with high temporal resolution data has proven crucial to the improving of hydrological studies, by providing detailed knowledge and information about the properties and the characteristics of their main variables. This study aimed at assessing the influence of rainfall event definition criteria in determining the characteristics of hyetographs, as well as in investigating the hydrological relations of rainfall events and antecedent soil moisture with those of the runoff hydrograph, by means of statistical techniques, in experimental basins in Northeast Brazil. To this end, a portion of the data was acquired, in sub-hourly scale, from the databases maintained by the partners of the research since 2003 and the other one was monitored and taken along the hydrological year of 2013-2014 through frequent measurements, performing essays and data processing. Then, descriptive statistics, hypothesis testing, exploratory data analysis, multivariate statistical techniques and multiple linear regression were carried out either for visual inspection or for statistical data investigation, in order to establish variation patterns and evaluate the association degree among variables. The first study revealed that the variation in minimum inter-event time criteria has reduced by almost 50% the number of events, and underestimated the average intensity of events by 4 times on average. The second study showed that the tropical coastal basin has approximately 4 times more events than those of semiarid, of which 6% of all events that remain unshaped, contribute to 38% of the total precipitation, and could be reduced by up to 97%, becoming mainly on unimodal events with peak to the left. The third study showed that soil moisture has a similar variation pattern up to 3 days in advance and exerts a significant influence, along with the precipitation and duration of rainfall events on both the runoff coefficient, the peak and the discharge of hydrographs; on the other hand, the principal component responsible for the peak and intensities of hyetographs was significant only on the runoff coefficient, which was, in turn, the best feature set for the multiple linear regression analysis, as it the coefficient of determination was 0.66. Therefore, this study highlights the importance of continuous and systematic hydro-climatological data monitoring for the experimental studies that may provide input to decision-making in water resources management measures.<br>O entendimento dos mecanismos de transformação chuva-vazão com dados em alta resolução temporal mostra-se fundamental no aperfeiçoamento da maior parte dos estudos hidrológicos, ao provê conhecimento e informação detalhada a respeito das propriedades e características de suas principais variáveis. Este trabalho teve como objetivo avaliar a influência dos critérios de definição de evento chuvoso na determinação das características dos hietogramas, bem como investigar as relações hidrológicas dos eventos chuvosos e da umidade antecedente do solo com aquelas dos hidrogramas de escoamento superficial, por meio de técnicas estatísticas, em bacias experimentais do Nordeste Brasileiro. Para tanto, uma parcela dos dados foi adquirida, em escala sub-horária, dos bancos de dados mantidos pelos parceiros da Rede de Hidrologia do Semiárido (REHISA) desde 2003 e a outra foi monitorada e obtida ao longo do hidrológico 2013-2014, por meio de frequentes medições, realização de ensaios e tratamento de dados. Em seguida, técnicas de estatística descritiva, testes de aderência e de hipóteses, métodos exploratórios de dados, análises estatísticas multivariadas e técnicas de regressão linear múltipla foram utilizadas, quer na inspeção visual, quer na investigação estatística dos dados, a fim de estabelecer padrões de variação e avaliar o grau de associação entre variáveis. O primeiro estudo realizado revelou que a variação do critério de mínimo intervalo de tempo entre eventos fez reduzir em até quase 50% o número de eventos, além de subestimar a intensidade média dos eventos 4 vezes, em média. O segundo estudo mostrou que a bacia hidrográfica tropical costeira apresenta aproximadamente 4 vezes mais eventos do que as do semiárido, dos quais 6% da totalidade dos eventos que permanecem não caracterizados, contribuem para 38% do total precipitado, e puderam ser reduzidos em até 97%, transformando-se, sobretudo, em eventos do tipo unimodal com pico à esquerda. O terceiro estudo mostrou que a umidade do solo apresenta um padrão de variação similar até 3 dias de antecedência e exerce uma influência significante, juntamente com a lâmina precipitada e a duração dos eventos chuvosos, sobre o coeficiente de escoamento, o pico e a vazão do hidrograma; por outro lado, a componente responsável pelo pico e intensidades dos hietogramas foi significante apenas sobre o coeficiente de escoamento, este que, por sua vez, foi a característica melhor ajustada pela análise de regressão múltipla, dado o coeficiente de determinação de 0,66. Portanto, este estudo evidencia a importância do monitoramento contínuo e sistemático de dados hidroclimatológicos para a realização de estudos experimentais que venham a dar subsídios à tomada de decisões em medidas de gerenciamento de recursos hídricos.