An object-based classification of mangrove land cover using Support Vector Machine Algorithm

Abstract

Accurate mapping of mangrove is necessary for effective planning and management of ecosystem and resources, due to the function of mangrove as a provider of natural products The use of satellite remote sensing to map mangrove has become widespread as it can provide accurate, effecient, and repeatable assessments. The type of remote sensing that is based on imaging using the pixel method sometimes results in the misclassification of the imaging due to the "salt and pepper effects". The aim of this study to use approach support vector machine (SVM) algorithm to classification mangrove land cover using sentinel-2B and Landsat 8 OLI imagery based on object-based classification method (OBIA). The field observation was done using Unmanned Aerial Vehicle (UAV) at Liong River, Bengkalis, Riau Province. The result by show overall accuracy classification using Sentinel-2B was better than Landsat 8 OLI imagery the value of 78.7% versus 62.7% and them were different significantly 7.23%.