%0 Conference Proceedings
%4 sid.inpe.br/marte2/2017/10.27.13.19.23
%2 sid.inpe.br/marte2/2017/10.27.13.19.24
%@isbn 978-85-17-00088-1
%F 60066
%T Comparação entre classificações de imagem RapidEye para o cálculo CN de bacia hidrográfica urbana: estudo de caso do Arroio Pepino (Pelotas/RS)
%D 2017
%A Nagel, Gustavo Willy,
%A Terra, Fabrício Silva,
%A Oliveira, Jade Silva de,
%A Aragona, Márcio Pagano,
%@electronicmailaddress fabricio.terra@ufpel.edu.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 3822-3829
%S Anais
%1 Instituto Nacional de Pesquisas Espaciais (INPE)
%X The runoff curve-number (CN) is an empirical parameter used for predicting direct runoff from rainfall excess, and it depends on land use and cover changes. High spatial resolution images have been important to identify these changes. This research aimed to compare effects of different land use and cover maps produced from K-means, MaxVer, and SAM classifications of high spatial resolution orbital image on calculation of CN value in the urban watershed of Arroio Pepino (Pelotas/RS). Our hypothesis was that different classification algorithms have produced divergent maps that in turn have affect the CN value of an urban watershed. A RapidEye image was classified in order to map the surface, and the following 10 classes were identified: water, asphalt, dirt road, vegetation (three types), roofs (three types), and building shade. The CN value of each class was obtained by comparing to corresponding tabulated values, and the total CN value was calculated taking into account the proportional area of each class. The MaxVer was the best-performed classifier (global accuracy: 64.89 % and kappa index: 0.59). The three CN values based on the distinct maps had different intensities where values calculated from K-means (CNtotal: 88.91 %) and SAM (CNtotal: 88.88 %) classifications were similar to each other and different of the value from MaxVer (90.71 %). Differences on proportions of land use and cover classes obtained from different classifiers affect the CN value of this urban watershed where its quality is highly dependent on accuracy of the classified image.
%9 Hidrologia
%@language pt
%3 60066.pdf