Investigation of Land Cover Classification in Oil Palm Area Based on ALOS PALSAR 2 Image

Abstract

Oil Palm is one of the most productive oil seeds in Indonesia. At present, Indonesia is the largest producer and exporter of oil palm worldwide. In general, oil palm production will be related to the age of the plant. The development of plant age will undergo physical changes in biomass and canopy density, so to identify the growth of planting age of oil palm can be used satellite image analysis from remote sensing. The objective on this study are examining techniques of image classification ALOS PALSAR 2 for land cover mapping of oil palm area. Study areas in oil palm plantations areas of Simpang Empat sub-district, Asahan Regency, North Sumatera. Methodology consisted of data collection of image ALOS PALSAR 2 and data of age of oil palm planting, the processing includes geometric correction, filtering, image cropping, making training sample using region of interest (ROI) tools, band combination, image classification for several methods like minimum distance, mahalanobis distance, maximum likelihood, and support vector machine, and then using confusion matrix for accuracy assessment. The results from this study are ALOS PALSAR 2 image classified with overall accuracy of 85.21% and coefficient kappa 0.6763 with RGB band combinations for support vector machine method as the most effective method for this research.