Avaliação do método SVM (Support Vector Machine) para o mapeamento multitemporal do municipio de São Gonçalo do Amarante - CE

• Ano de defesa: 2021
• Autor(a) principal: Damasceno, Aline Moreira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Aprendizado de máquina; Mapeamento; Acurácia
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/64383

Identifying the changes and alterations that occur on the Earth’s surface related to its resources and interactions between natural and anthropic phenomena promotes an improvement in the understanding and comprehension of the processes of environmental modification, providing responsible means in the decision making processes related to the environmental planning of a region. For this, several tools can be used to capture these changes, such as remote sensing data, which are indispensable and extensively used sources, to assist in the detection of changes occurring over the years, and mainly through multitemporal analysis. In this way, this work used remote sensing techniques combined with machine learning to trace a diagnosis of the changes in the municipality of São Gonçalo do Amarante, located in the State of Ceará, which has been suffering from intense changes over the years. The study evaluated Landsat satellite images from 1991, 1999, 2007 and 2018 to detect environmental changes, from a joint analysis between land use maps and environmental maps, as well as mapping data and reports of studies already carried out in the region. The information collected and generated were associated in a GIS environment (Georeferenced Information System) and the results generated were important to diagnose the changes detected in the region through the land use maps, as well as to identify the most fragile areas, emphasizing mainly the intense pressure suffered over the years, on the environmental units of the area, before and after the implementation of the Industrial and Port Complex of Pecém (CIPP).