Abstract

In this work, a multi-sensor approach to separate oil palm plantations from forest cover using NDFI and a modified Pauli Decomposition technique is presented. The main contribution of this research is the potential to reduce misclas-sification of both classes, in the context of automated-base supervised classification algorithms, to decrease uncertainties derived through the detection and mapping process of forest cover. The hereby proposed method includes the generation of a primary forest map cover defining thresholds from a high resolution multi-spectral satellite image, and then the palm oil plantation will be filtered out from this classification using scattering mechanisms by a Pauli Decomposition approach. Preliminary results shown the capabilities of this approach in order to generate complementary information to separate the oil palm plantations from the forest cover classification.