Brigitta Aurelia Putri Suhendi, Waris Marsisno

Integration of Radar Satellite Imagery and AIS Data for Oil Spill Monitoring with a Machine Learning Approach

Introduction

Integration of radar satellite imagery and ais data for oil spill monitoring with a machine learning approach. Monitor Riau Archipelago oil spills using Sentinel-1 radar imagery, AIS data, and XGBoost machine learning. Detects, maps areas, and identifies passing ships.

Abstract

Oil spill cases are routine occurrences in the Riau Archipelago Province, particularly in the northern part of Bintan Island, as one of the consequences of the island's strategic location near Singapore, a global trade and financial hub. The limitations of resources and data pose obstacles for the government in addressing this issue. Therefore, in this study, the author offers a solution by detecting oil spill areas using Sentinel-1 radar data and wind speed using the adaptive thresholding method, the best classification model from 7 machine learning algorithms and 2 deep learning algorithms. Based on the accuracy, precision, recall, and F1-score values of these 9 algorithms, the best algorithm obtained is XGBoost (Extreme Gradient Boosting). Followed by mapping and estimating the area of the oil spill at 177.54 ha, as well as identifying ships passing through the oil spill area using AIS (Automatic Identification System) data and wind direction, with the result being 2 ships passing through the area.

Full Text

Comments

You need to be logged in to post a comment.