%0 Conference Proceedings
%4 dpi.inpe.br/sbsr@80/2008/10.27.19.01
%2 dpi.inpe.br/sbsr@80/2008/10.27.19.01.28
%@isbn 978-85-17-00044-7
%T Avaliação da exatidão do mapeamento da cultura do café no Estado de Minas Gerais
%D 2009
%A Adami, Marcos,
%A Moreira, Mauricio Alves,
%A Barros, Marco Aurélio,
%A Martins, Vagner Azarias,
%A Rudorff, Bernardo Friedrich Theodor,
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@affiliation GEOAMBIENTE Sensoriamento Remoto Ltda
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@affiliation Instituto Nacional de Pesquisas Espaciais (INPE)
%@electronicmailaddress adami@dsr.inpe.br
%@electronicmailaddress mauricio@dsr.inpe.br
%@electronicmailaddress marcoaurelio.barros@geoambiente.com.br
%@electronicmailaddress vagner@dsr.inpe.br
%@electronicmailaddress bernardo@dsr.inpe.br
%E Epiphanio, José Carlos Neves,
%E Galvão, Lênio Soares,
%B Simpósio Brasileiro de Sensoriamento Remoto, 14 (SBSR)
%C Natal
%8 25-30 abr. 2009
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 1-8
%S Anais
%1 Instituto Nacional de Pesquisas Espaciais (INPE)
%K remote sensing, sampling, error matrix, sensoriamento remoto, amostragem, matriz de erros.
%X Accurate and updated agricultural statistics on coffee crop area estimation, based on an objective method, is not available. Remote sensing images are an important source of information to estimate crop area, especially when crop identification procedure is carried out in an effective way. The objective of the present work is to evaluate a coffee crop map obtained from remote sensing images in the state of Minas Gerais. In the adopted method the samples were randomly selected within strata obtained in a two stage stratification. In the first stage five strata were obtained based on both percent of coffee area and regional crop management characteristics. In the second stage the strata were divided into areas with and without coffee crop. For each stratum 52 samples were randomly selected. Prior to the field work the samples were visually identified on both high spatial resolution images available in Google Earth and recent Landsat images used to perform the map. Several samples could be clearly identified as coffee or non-coffee reducing drastically the field work. Confusion matrix was used to provide the global accuracy index, the consumer accuracy and the producer accuracy for each stratum. Overall mapping accuracy for strata 1, 2, 3, 4 and 5 were 99%, 90%, 95%, 86% and 83%, respectively. Global mapping accuracy for Minas Gerais was 95%. It was observed that the mapping accuracy is related to: a) regional vegetation that may cause confusion with coffee; b) cultural practices; c) crop management; and d) relief.
%9 Agricultura
%@language pt
%3 1-8.pdf