Mapeamento das áreas de ocorrência de infestação do Eragrostis Plana Nees (capim annoni) com sensoriamento remoto: estudo de caso em Santana do Livramento/RS/ Brasil

• Ano de defesa: 2017
• Autor(a) principal: Cicconet, Nelson
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• dARK ID: ark:/26339/0013000003fx4
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Geografia; UFSM; Programa de Pós-Graduação em Geografia; Centro de Ciências Naturais e Exatas
• Palavras-chave em Português: Capim Annoni; Gramíneas nativas; Sensoriamento remoto; Annoni grass; Native grasses; Remote sensing; CNPQ::CIENCIAS HUMANAS::GEOGRAFIA
• Link de acesso: http://repositorio.ufsm.br/handle/1/13534

Eragrostis Plana Nees (Capim annoni) is an invasive species originating from the African continent that has caused numerous environmental and socioeconomic damages, mainly in parts of the Bioma Pampa. Several studies have been and are being developed, aiming at annoni grass. Remote sensing and geoprocessing techniques were studied and described to better understand and apply in this study. In this sense, the objective of this study is to identify, analyze and differentiate the spatial distribution of Eragrostis Plana Ness (grass annoni) from other native pastures, through the technique of digital classification of high and medium spatial resolution images, applying the Support Vector Machine classifier SVM), analyzing four different orbital images (spectral, spatial and temporal differences) of the North / Northeast part of Cerro Chato Wind Park, Santana do Livramento - RS, as part of the pilot project to map spatial distribution of annoni grass in the area For further mapping in other areas and scales. The methodology used was the digital classification of four satellite images - two RAPIDEYE, one IKONOS and one KOMPSAT of different dates and resolutions. The methodology was supervised classification, with data collected from field visits to obtain the areas of interest. The results were placed in the form of use maps, charts and tables. Seven classes were identified: annoni grass, native grasses, transition areas between annoni grass and native grasses, water bodies, wetlands, tree vegetation and exposed soil. Confusion matrices and Kappa index were generated for each image to evaluate which image would yield the best result. Therefore, the IKONOS image was the one that best suited this work, generating a Kappa index higher than 0.8 and a classification accuracy index higher than 90%. The main conclusions regarding this dissertation show us the maps of use and coverage of part of the Cerro Chato Wind Farm and with this methodology it is possible, through this and future studies, to extrapolate the results to all the municipality of Santana do Livramento and to others Where the invasive species can be found.