Uso de técnicas de mineração de dados na identificação de áreas hidrologicamente homogêneas

• Ano de defesa: 2009
• Autor(a) principal: Diniz, Roberta Brito Nunes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraí­ba; BR; Informática; Programa de Pós Graduação em Informática; UFPB
• Palavras-chave em Português: Mineração de dados; Clusterização; Regionalização hidrológica; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/tede/6080

The lack of fluviometric data and the bad quality of existing data related to watercourses of Brazilian northeast have obligated hydrology specialists to search for new solutions, improving the knowledge and methodologies to build the region up according to its environment limitations. Through the hydrologic regionalization techniques it's possible to transfer data and information among similar watersheds. In this context, the purpose of this work is to identify hydrologically similar regions in the State of Paraiba using Clustering - a kind of data mining technique - to find patterns that allow data transposition from one region to other. Algorithms were used with methods based on partition, hierarchical methods, and methods based on neural networks. It is used data of 41 watersheds located in the State of Paraíba. For all the watersheds, 32 physical and climatological characteristic were determined, being six characteristics of the linear measures of the watersheds, three of shape, nine of the drainage net, seven of the relief, two of the drainage capacity, three of the classes of soils and two of the climatology. Four sceneries were composed for the execution of the algorithms of which three were obtained through the technique of selection of attributes. To evaluate the groups obtained by the algorithms three validation indexes were applied, namely index of Silhouette, index Davies-Bouldin and index Dunn. The results of the statistical validation showed that the algorithm Ward stood out in the clustering of the 41 watersheds, identifying 6 homogeneous regions in the State of Paraíba. The obtained results showed the viability of the use of techniques of data mining in studies of hydrologic regionalization and practical applications in engineering of water resources in the ambit of the State of Paraíba.