Monitoramento do desmatamento nas terras indígena do sul do amazonas: uma abordagem entre a vetorização manual e a classificação automática
2333 palavras
10 páginas
Anais XV Simpósio Brasileiro de Sensoriamento Remoto - SBSR, Curitiba, PR, Brasil, 30 de abril a 05 de maio de 2011, INPE p.2920Monitoramento do desmatamento nas Terras Indígena do sul do Amazonas: uma abordagem entre a vetorização manual e a classificação automática
Manoel Ricardo Dourado Correia1,0
Daiane Cardoso Lopes Batista1,0
Rutenio Luiz Castro de Araujo2,0
1
Sistema de Proteção da Amazônia
Av. do Turismo, N° 1350, Tarumã, 69049-630 - Manaus - AM, Brasil manoel.correia@sipam.gov.br, dayane.capes@hotmail.com
2
Universidade Federal do Amazonas (UFAM)
Av. General Rodrigo Octávio, 3000,
Campus Universitário Coroado I, 69.077-000,Manaus-AM rutenioa@bol.com.br Abstract. This article aims to map the occurrence of major deforestation in the Indigenous Lands (TIs) from southern Amazonas, and perform a comparison between manual vectorization methodology with automatic sorting (supervised classification). Images from the Landsat / TM 5, acquired during t he months from June to
October 2009. The applications Envi 4.5 and ArcGIS 9.2 were used to georeferencerecord, mosaic, vectorization, supervised classification and to quantify the data. Once calculated the values of deforestation in manual and automatic, these data could then be compared. The res ults indicated that: (a) all the scenes had satisfactory results regarding the registration and mosaic. (B) The manual vectorization, it was noted that indigenous lands Tenharim Gleba B and Nove de Janeiro stood out by having a larger cleared area, with over
1,400 ha. (C) In supervised classification, attempts have shown that better results were in separate units. As a result, decided to choose two indigenous physiognomy of different composition, including: Diahui which achieved an accuracy of 99.79% and kappa index of 0.9599, ie, a result with little confusion between field s and forest; the unit of Tenharim Igarapé Preto , the kappa value was 0.8074 and displayed an accuracy of 98.52%,