Marine Disaster Detection Using the Geostationary Ocean Color Imager (GOCI)

Abstract

Recently, harmful algae (e.g., red tide) has damaged human and marine ecosystems. To address this, a response system should be developed to quickly cope with these ocean disasters. However, it is difficult to simultaneously monitor the vast ocean areas. Here, a marine disaster detection system can be developed through a convergence between the satellite-based ocean color remote sensing and the marine sensor network. The system architecture is divided into two steps: first, the system detects ocean anomalies in realtime using the satellite-based techniques, and secondly, the detected disaster information is transferred to the ships via the marine sensor networks. In this paper, we only focused on the first step and the second step is reserved for future work. Although the polar orbit satellite-based ocean color sensor platforms (e.g., MODIS, MERIS, and SeaWifs) can be used to simultaneously monitor the vast ocean areas, they are unsuitable for capturing subtle changes on a geographically equivalent area. On the other hand, the Geostationary Ocean Color Imager (GOCI), the world’s first ocean color remote sensor platform operated on a geostationary orbit, receives ocean color data around the Northeast Asia region every hour, eight times a day. Therefore, GOCI can be more effectively utilized to observe subtle changes and to detect anomalies in ocean environments in real-time. In this paper, we attempted to build a system to monitor marine disasters by detecting ocean anomalies using the ocean color data derived from GOCI. This system directly compares the test spectrum vectors (i.e. anomaly candidates) to a predefined reference spectrum vector (i.e. a target anomaly) through the cosine similarity. The experimental result showed that the proposed system could efficiently detect the disasters (e.g., the red tide) on the ocean environments.