%0 Conference Proceedings
%4 sid.inpe.br/marte2/2017/10.27.13.48.48
%2 sid.inpe.br/marte2/2017/10.27.13.48.49
%@isbn 978-85-17-00088-1
%F 59307
%T Uso de análise de imagens baseada em objetos (OBIA) e classificação não-supervisionada para identificação de envelopes bioclimáticos no bioma amazônico
%D 2017
%A Silva, Thiago Sanna Freire,
%A Cordeiro, Carlos Leandro de Oliveira,
%@electronicmailaddress tsfsilva@rc.unesp.br
%E Gherardi, Douglas Francisco Marcolino,
%E Aragão, Luiz Eduardo Oliveira e Cruz de,
%B Simpósio Brasileiro de Sensoriamento Remoto, 18 (SBSR)
%C Santos
%8 28-31 maio 2017
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 4443-4450
%S Anais
%1 Instituto Nacional de Pesquisas Espaciais (INPE)
%X Mapping bioclimatic envelopes onto geographical space is an important process for predicting and understanding species range sizes and distribution limits This mapping depends on the availability of relevant environmental datasets at the appropriate scale, several which are commonly derived from remote sensing sources. Object-based image analysis (OBIA) is a relatively new approach in remote sensing, and together with machine learning algorithms, has been used successfully to map land cover classes in highly heterogeneous satellite and aerial images. Its framework, however, is equally applicable to other problems involving the delineation and classification of spatially homogeneous regions. Therefore, we investigated the applicability of OBIA and unsupervised clustering to detect possible bioclimatic envelopes and support biogeographical studies. Our selected study area was a portion of the Amazon basin, corresponding to interfluve between the Negro and Branco rivers. We combined topographic data from the SRTM mission, precipitation data from the CHIRPS dataset, radar backscattering from the ALOS/PALSAR sensor and spatially explicit estimates of canopy height and vegetation biomass, which was supplied as input data for image segmentation and posterior clustering, based on mean and standard deviation attributes. Overall, several bioclimatic envelopes could be mapped, some of them coinciding to land features usually associated with vicariance events, such as the Branco River channel. Our results emphasize the applicability of modern image analysis methods for identifying bioclimatic envelopes using spatially-explicit environmental data, and also the potential role of bioclimatic discontinuities, as well as vicariance, in explaining current distribution patterns for amazonian species.
%9 Monitoramento e modelagem ambiental
%@language pt
%3 59307.pdf