Contribuições a técnicas de agrupamento e visualização de dados multivariados utilizando mapas auto-organizáveis

Self-organizing maps (SOM) are artificial neural networks widely used in the data mining field, mainly because they constitute a dimensionality reduction technique given the fixed grid of neurons associated with the network. In order to properly the partition and visualize the SOM network, the various methods available in the literature must be applied in a post-processing stage, that consists of inferring, through its neurons, relevant characteristics of the data set. In general, such processing applied to the network neurons, instead of the entire database, reduces the computational costs due to vector quantization. This work proposes a post-processing of the SOM neurons in the input and output spaces, combining visualization techniques with algorithms based on gravitational forces and the search for the shortest path with the greatest reward. Such methods take into account the connection strength between neighbouring neurons and characteristics of pattern density and distances among neurons, both associated with the position that the neurons occupy in the data space after training the network. Thus, the goal consists of defining more clearly the arrangement of the clusters present in the data. Experiments were carried out so as to evaluate the proposed methods using various artificially generated data sets, as well as real world data sets. The results obtained were compared with those from a number of well-known methods existent in the literature

Cartografia temporal e análise geoambiental da dinâmica da foz do Rio Piranhas-Açu, regiao de Macau-RN, com base em imagens Landast 5-MT

This work embraces the application of Landsat 5-TM digital images, comprising August 2 1989 and September 22 1998, for temporal mapping and geoenvironmental analysis of the dynamic of Piranhas-Açu river mouth, situated in the Macau (RN) region. After treatment using several digital processing techniques (e.g. colour composition in RGB, ratio of bands, principal component analysis, index methods, among others), it was possible to generate several image products and multitemporal maps of the coastal morphodynamics of the studied area. Using the image products it was possible the identification and characterization of the principal elements of interest (vegetation, soil, geology and water) in the surface of the studied area, associating the spectral characteristics of these elements to that presented by the image products resulting of the digital processing. Thus, it was possible to define different types of soils: Amd, AQd6, SK1 and LVe4; vegetation grouping: open arboreal-shrubby caatinga, closed arborealshrubby caatinga, closed arboreal caatinga, mangrove vegetation, dune vegetation and areas predominately constituted by juremas; geological units: quaternary units beach sediments, sand banks, dune flats, barrier island, mobile dunes, fixed dunes, alluvium, tidal and inundation flats, and sandy facies of the Potengi Formation; tertiary-quaternary units Barreiras Formation grouped to the clayey facies of the Potengi Formation, Macau Formation grouped to the sediments of the Tibau Formation; Cretaceous units Jandaíra Formation; moreover it was to identify the sea/land limit, shallow submersed areas and suspended sediments. The multitemporal maps of the coastal morphodynamics allowed the identification and a semi-quantitative evoluation of regions which were submitted to erosive and constructive processes in the last decade. This semi-quantitative evoluation in association with an geoenvironmental characterization of the studied area are important data to the elaboration of actions that may minimize the possible/probable impacts caused by the implantation of the Polo Gas/Sal and to the monitoring of areas explorated by the petroleum and salt industries

Estudo da dinâmica costeira da região da foz do rio Piranhas-Açu para geração de mapas de sensibilidade do litoral ao derramamento de óleo

The study area is located in the northern coast of Rio Grande do Norte State comprising the mouth of Açu-Piranhas river including the cities of Porto do Mangue e Areia Branca. The local geological setting comprises Cretaceous, Tertiary and Quaternary geological units of the Potiguar Basin. One is about a region of high morphologic instability due to action of the rigorous dynamic coastal processes, beyond the intense human activities mainly for the performance of the petroliferous industry, salt farms and tanks of shrimp industry.For the accomplishment of this work Landsat 5 TM and Landsat 7 ETM + from four distinct dates were used as cartographic base, in which one applied techniques of digital processing to elaborate thematic maps of the existing natural resources to support the geologic and geomorphologic characterization and the soil and landuse maps. The strategy applied was the interpretation of multitemporal images from aerial and orbital remote sensors alIied to the terrain truth recognition, integrated through a Geographic Information System. These activities had alIowed the production of Sensitivity Maps of the Coast to Oil Spilling for the area, on the basis of the Coastal Sensibility Index. Taking into account the seasons were created maps to distinct datas: July 2003 represents the winter months that presented a sensibility lower when compared with the month of December 2003. For the summer months greater sensitivity is due to the hydrodynamic data that suggest a lesser capacity of natural cleanness of the oil and its derivatives in spilling case.These outcomes are an important and useful database to support an assessment to a risk situation and to taking decision in the face of an environmental disaster with oil spilling in coastal area, alIowing a complete visualization of the area and identifying all portions in the area with thei environmental units and respective Coastal Sensibility Index.