INVESTIGASION OF PALM OIL PLANTATION USING MULTIALGORITHM AND MULTIRESOLUTION-SPECTRAL OF OPTICAL IMAGERY (Case Study: Asahan Regency, North Sumatra Province)

Abstract

The purpose of this study was to find the best classification method for identifying oil palm plantations using optical images to be used as a map of oil palm plantation land cover in Asahan Regency, North Sumatra Province. Starting from the collection of Landsat 8 OLI (Operational Land Imager) satellite image data 2017, Sentinel-2A 2017 and SPOT 6 2017 and data on oil palm plantation blocks in 2017. The next stage is radiometric and geometric correction of optical images used for the classification process and identification of oil palm plantations with the help of oil palm plantation block data and Google Earth. The classification method used is the Parallelepipe, Maximum Likelihood, and Support Vector Machine algorithms. The results in this study indicate that Support Vector Machine (SVM) using channel 412 on SPOT 6 images is the best algorithm to find out the distribution map of oil palm plantations which produces an accuracy of 92,32% and kappa 0,85.