Mangroves Change Detection using Support Vector Machine Algorithm on Google Earth Engine (A Case Study in Part of Gulf of Bone, South Sulawesi, Indonesia)

摘要

Remote sensing data have been proven to be efficient as data source for mangrove mapping and monitoring to support decision making and policy related to mangrove management. One of the key advantages of remote sensing is the temporal availability of the data which allow monitoring of mangrove status from different time period. In line with this advantage, the recent development of Google Earth Engine (GEE) has open wider possibility to work with large image datasets in an online platform for mangrove monitoring. This study aims to develop a method to monitor mangrove cover changes at some parts of Gulf of Bone, South Sulawesi, Indonesia from 2014 to 2018 using a combination of GEE and Support Vector Machine (SVM) algorithm applied to Landsat 8 OLI (30 m pixel size). We used region of interest (ROI) technique to distinguish mangroves, non-mangroves, open area, water bodies, and cloud objects. The result of five classes ROI was for defining all the dataset for data model. The algorithm implementation result shows that the mangrove cover from 2014 to 2015 had decreased significantly along the beach and in several side of fishponds. However, from 2016 to 2018 the mangrove cover had increased especially in the south side of the study area. This change pattern shows the dynamic of mangrove cover in the study area, mainly caused by the development of fish or shrimp ponds and some mangrove restoration efforts. The result shows the potential of SVM and GEE for spatio-temporal data analysis based on Landsat 8 OLI to monitor the mangrove cover changes over the time. Nevertheless, the spectral characteristics of mangroves which is influenced by water bodies or unconsolidated sediment background make the identification of mangroves or non-mangroves area remains challenging.