CLASSIFYING INDONESIAN PLANTATION AND NATURAL FOREST COVER AND MEASURING CHANGES WITH C- AND L-BAND SAR DATA

摘要

Tropical coverage by Envisat is sparse in space and timernand has limited value for monitoring deforestation. Thernonly available APG multi-temporal dataset over Riaurnprovince, Indonesia (9 images in a single year), is usedrnto distinguish and monitor tropical plantations and theirrndynamics and is compared with L-band PALSAR data.rnFor Envisat APG data, both VV and VH are importantrnin discriminating different types of forest cover, while atrnL-band most of the relevant information is in the crosspolarisedrnchannel. Whether the underlying soil is peat orrnnon-peat in acacia plantations has important effects onrnbackscatter and classification. Supervised classificationrnof the C-band data gave overall accuracies of 86.2% andrnkappa coefficient of 0.78 by comparison with land coverrnmaps derived from optical data. Classifications fromrnseparate phases in the C-band time series allow thernchanges occurring in acacia plantations due tornmanagement to be tracked.