Título:	Mapeamento de cicatrizes de áreas queimadas no Cerrado usando imagens Landsat, Google Earth Engine e Deep Learning
Autor(es):	Arruda, Vera Laísa da Silva
Orientador(es):	Pereira, Reginaldo Sérgio
Coorientador(es):	Matricardi, Eraldo Aparecido Trondoli
Assunto:	Bioma Cerrado Fogo - Cerrados Área queimada Google Earth Engine Deep Learning Fogo - comportamento
Data de publicação:	26-jul-2021
Data de defesa:	1-abr-2021
Referência:	ARRUDA, Vera Laísa da Silva. Mapeamento de cicatrizes de áreas queimadas no Cerrado usando imagens Landsat, Google Earth Engine e Deep Learning. 2021,61 f., il. Dissertação (Mestrado em Ciências Florestais)—Universidade de Brasília, Brasília, 2021.
Abstract:	The Cerrado biome is enhanced by a mosaic of different vegetation types with one of the highest levels of biodiversity in the world. Fires have historically contributed to shaping natural vegetation and are now used in the management of agricultural systems and pastures. The fire regime has changed in recent decades, consequently, mapping the spatial and dynamic distribution of fire is crucial to assess its impacts on ecosystems and to define, control, and prevent fire measures. In this study, we developed a methodology for mapping burnt areas in the Cerrado biome, using Landsat images and a Deep Learning algorithm; implemented in the Google Earth Engine platform and Google Cloud Storage platform. We obtained 97% accuracy from validation evolving 2,200 points. Based on the efficiency of the methodology for mapping scars in 2017, the methodology was extended from 2000 to 2019, to better understand the behavior of fire in the Cerrado. From the historical series, it was estimated that 41% of the Cerrado were already burned at some point in the time, with the average area burned representing 109,138 km2 (5.4% of the Biome). Regarding the recurrence of fire in the same area, we observed that 270,601 km2 burned 2 to 3 times, 181,612 km2 burned 4 to 10 times, and 13,085 km2 burned more than 10 times, showing how the changes in land use have altered the fire regime to more frequent intervals. Regarding the types of surface on which burned areas are recorded, burning areas over natural vegetation (forest, savanna and grassland) predominates over anthropic areas. Proportionally to the areas of the biome's land categories, fire predominates in protected areas (indigenous lands, conservation units and quilombola territories). The analyzes demonstrate the potential of remote sensing products, such as those Landsat time series used by the MapBiomas project, to explore the spatio-tempora variations of fire occurrences. The results also highlight the importance of adopting management measures and policies aimed at the conservation of the Cerrado. In the last decades, the occurrence of fire has increased, consequently, CO2 emissions also has increased. Thus, decision makers of public policies and management plans need to consider future scenarios of climate change and greater risks of fire occurrences.
Unidade Acadêmica:	Faculdade de Tecnologia (FT) Departamento de Engenharia Florestal (FT EFL)
Informações adicionais:	Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Florestal, Programa de Pós-Graduação em Ciências Florestais, 2021.
Programa de pós-graduação:	Programa de Pós-Graduação em Ciências Florestais
Licença:	A concessão da licença deste item refere-se ao termo de autorização impresso assinado pelo autor com as seguintes condições: Na qualidade de titular dos direitos de autor da publicação, autorizo a Universidade de Brasília e o IBICT a disponibilizar por meio dos sites www.bce.unb.br, www.ibict.br, http://hercules.vtls.com/cgi-bin/ndltd/chameleon?lng=pt&skin=ndltd sem ressarcimento dos direitos autorais, de acordo com a Lei nº 9610/98, o texto integral da obra disponibilizada, conforme permissões assinaladas, para fins de leitura, impressão e/ou download, a título de divulgação da produção científica brasileira, a partir desta data.
Aparece nas coleções:	Teses, dissertações e produtos pós-doutorado

Os itens no repositório estão protegidos por copyright, com todos os direitos reservados, salvo quando é indicado o contrário.