Análise de métodos de realce espectral e temporal aplicados à discriminação de classes de uso e cobertura da terra em áreas de Cerrado

• Autor(a) principal: SILVA, Iza Carla de Oliveira e
• Data de Publicação: 2010
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tde/1848
• topic: Cerrado; Sensoriamento remoto da vegetação; Processamento digital de imagem; Realce espectral; 1. Cerrado 2. Sensoriamento remoto da vegetação 3. Processamento digital de imagem 4. Realce espectral; Cerrado; Remote sensing of vegetation; image processing; Spectral enhancement; CNPQ::CIENCIAS HUMANAS::GEOGRAFIA::GEOGRAFIA HUMANA::GEOGRAFIA AGRARIA

The Cerrado biome has the national and international eminence, because it dimensions, around 205 millions ha. in the Brazil, beyond to emphasize like an areas of the global importance, call hotspots. With 9,2% of it is area in the parks and ecological reservation, studs of the mapping of the vegetation indicate a scenery of the conversion estimated between 40 % and 55 %, for areas of the pasture and agricultural. This work has the objective to evaluated the potential of discrimination of the natural and human target from the images with high resolution and high spectral resolution, accentuated amount the spectral dominion like temporal, to identify the better technique and protocols, which results in mapping semi-automatic of the landscape cover in the Cerrado biome, more accurate. It was selected an area test for this analyze, the municipality of Mineiros, the definition of the are was base in the environment importance, because inside of this municipality there is the fountainhead of the Araguaia - Tocantins River basin, the Prata basin and the Paraguai basin, beyond the strategic areas for conservation (Parque Nacional das Emas) and the biodiversity corridors (Cerrado Pantanal). The models of spectral enhancement NDVI, MLME and CP were apply in a LANDASAT TM (orbit/point: 224/73) from march 30 of 2008 (rain season) and august 21 of 2008 (dry season). The statistic analyses show the existence of one spectral confusion more intensify between Campo Sujo and Pasture and Cerrado Sensu Stricto, in all categories of the analyze data. Other results showing that the capacity of discriminate is bigger in the august, the dry season, when there is more distinction in the vegetation. And finally, between the spectral model, which presents better potential to distinction the class was observed by the images, component main, with 96 % of the sample classify correctly. Before the MLME presents results of 91 % in relation to NDVI, with 84 % of the capacity to define the class, this difference is because the number of the patchy in NDVI is less ( i.e NDVI use just two bands, red and infra-red) when compare to MLME and CP. Thus the work consider that application of the spectral enhancement in the image of the original digital values contribution in the distinction potential of the natural and anthropic, concluding that the analyze of the main components increase with the precision of the semi-automatic mapping of use land cover.