Identification of Oil Palm Plantation on Multiscatter and Resolution of SAR Data Using Variety of Classifications Algorithm (Case Study: Asahan District, North Sumatera Province)

Abstract

The purpose of this study was to find the best classification algorithm for determining the classification of oil palm plantation area on SAR data. This research begins with collecting primary data and secondary data, where primary data in the form of radar images include Sentinel-1, ALOS PALSAR, and TerraSAR-X images, while secondary data of palm oil age block data are used as a references derived from plantation agency of Indonesia and high resolution imagery, after that radiometric and geometric corrections, classification and accuracy assessment. The classification methods to be used are Parallelepiped, Maximum Likelihood, and Support Vector Machine. The results of this study is the Support Vector Machine algorithm has the highest overall accuracy value among the three methods used with 77.57% for TerraSAR-X, 86.18% for Sentinel-1 and 91.17% for ALOS PALSAR, The other results on this study are map of land cover for oil palm plantations in the district of Asahan, North Sumatra Province.