IDENTIFICATION OF EUCALYPTUS PLANTATION IN VALE DO PARAÍBA -SP USING FRACTION IMAGES AND SPECTRAL INDEX AS DATASET TO RANDOM FOREST

This study presents a method for assessing eucalyptus and pine plantation areas in the Vale do Paraíba, São Paulo, using annual time series of Landsat-8 Operational Land Imager (OLI) images from 2015 to 2024. By applying a linear spectral mixing model, fraction images were generated to improve land use and land cover classification. A random forest algorithm was employed for classification, achieving an overall accuracy of 87.5%. The analysis effectively identified phenological cycles of eucalyptus plantations, highlighting the importance of remote sensing techniques for environmental management and policy development in forest plantation areas.