Backscatter multiespectral como ferramenta para classificação da geodiversidade do fundo marinho

• Ano de defesa: 2023
• Autor(a) principal: Menandro, Pedro Smith
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Doutorado em Oceanografia Ambiental; Centro de Ciências Humanas e Naturais; UFES; Programa de Pós-Graduação em Oceanografia Ambiental
• Palavras-chave em Português: Backscatter multiespectral; Classificação do fundo marinho; Mapeamento do fundo marinho; Multifeixe; Ciências Ambientais
• Link de acesso: http://repositorio.ufes.br/handle/10/12783

The scientific field in which seabed mapping is inserted has been assuming increasingly greater importance in recent decades, following the demands associated with numerous uses and purposes in society, as well as the relevant evolution of technology. Currently, we are coming up to the halfway point of the Ocean Science Decade, but still a long way from achieving one of its priority targets, which encompasses an atlas of the oceans, not just with bathymetric information, but also with understandable thematic maps with multiple variables that can be used for several purposes. In this context, increasingly robust and complex analysis techniques used in this interdisciplinary field of seabed mapping have contributed to improving our knowledge about the ocean based on different variables, most notably those derived from depth and backscatter. Acoustic backscatter, although still somewhat underused on many occasions, is basically defined as the acoustic energy that returns to the equipment after a number of physical interactions and processes (with the water column and the seabed). It has yielded outstanding results as indicative of the seabed's physical properties and can be analysed using different approaches - image-based analysis (mosaics) and angular response-based analysis. The acoustic backscatter is collected synchronously with the bathymetric data, and has played a key role in various classification schemes, inputting predictive models and contributing to the interpretation of the marine landscape and its geodiversity. More recently (around 8 years ago), some studies involving multi-frequency backscatter have begun to be published in the scientific literature, based on the same thinking used in terrestrial remote sensing - that multiple bands allow for greater discrimination of the surface being analysed. This work, therefore, explores data collected with a multibeam multispectral backscatter echo sounder (frequencies of 170 kHz, 280 kHz, 400 kHz, and 700 kHz) on different seabed types, aiming to understand how the acoustic response behaves according to frequency and seabed, and to improve seabed classification by applying different analysis approaches and classification models. The structure of the thesis sets out to initially present the concept and potential of backscatter for seafloor classification (chapters 1 and 2), applications in different seabed types considering morphology and substrate (sand, mud, rhodolith beds) using different approaches and methods to classify the seabed and indicate possible advantages and disadvantages (chapters 3 and 4), in addition to providing an overview of the state of the art regarding the scientific development of backscatter as a tool for seabed classification. This work addresses one of the key themes - multispectral backscatter - conventionally defined for backscatter studies by diversified groups around the world, as well as other topics such as combining complementary approaches to analysing backscatter, machine-learning techniques applied to the study of the oceans, mapping and classifying habitats for multiple purposes, mapping rhodoliths and mapping geodiversity in the state of Espírito Santo.