Canasat Project accuracy assessment of sugarcane thematic maps

Canasat Project accuracy assessment of sugarcane thematic maps
Marcio Pupin Mello, Marcos Adami, Bernardo F. T. Rudorff and Daniel Alves Aguiar

National Institute for Space Research (INPE), Remote Sensing Division (DSR) Av. dos Astronautas, 1758 – Jd. Granja – São José dos Campos – SP, 12227-010, Brazil (mello@ieee.org, adami@dsr.inpe.br, bernardo@dsr.inpe.br, daniel@dsr.inpe.br)

Abstract: In order to perform the thematic accuracy assessment of Canasat Project mapping, we used a novel approach to construct the reference dataset which consists of a web platform used to integrate remote sensing images, time series, and ancillary data. Results showed that, although Canasat Project mapping of sugarcane areas has a mean overall accuracy of 98%, the mean thematic error associated with sugarcane area estimates was of –0.7%, since part of the omission errors was compensated by the inclusion errors. Despite assessing only one crop year (2010/2011) we would also expect to obtain very similar results for other crop years due to the consistent method and the careful visual interpretation carried out by the Canasat Project experienced team.

Keywords: Remote sensing, web platform, sampling, Monte Carlo method.

AttachmentSize
MelloAccuracy2012.pdf263.25 KB