Bilge Dump System

By Lizwe Mdakane, CSIR

Ships are required by law to dump bilge waste at a port. However, vessels can accidentally or deliberately spill oil in the ocean. Oil spill events are often caused by vessels when dumping oily bilge wastewater at sea. Depending on the amount and location, oil spill events can be very harmful to the sea ecosystem. The risk of bilge dumping increases with the increase in maritime transportation.

BD_Dest overview
BD_DeST stands for Bilge Dump Decision Support Tool. The Dest uses spaceborne Synthetic Aperture Radar (SAR) images (See Figure 1). Automatically detects bilge dumps using state-of-art image processing algorithms published in IEEE journal. It also provides a complete service chain, starting from the acquisition and process of the data to a product ready for the client. An alert on detected bilge dump events is published to response authorities through scheduled reports and live dashboard website.

Figure 1: Sentinel-1A SAR subimages showing the darkspot and limited area (shown in yellow) used to extract physical features.

Why use SAR?
SAR is an active microwave sensor that uses radar return energy to capture Earth’s surface features. It can capture large areas of the Earth’s surface from either airborne or space-borne platforms. Bilge dumps are detected under most weather conditions, day or night. SAR can not only monitor illegal bilge dumping activities but can also identify the vessel responsible.

SAR acquisition
Radasat-2 imagery provided on weekly bases by the South African National Space Agency (SANSA) with zero downtime. Received over 360 images covering the entire South African Exclusive Economic Zone (EEZ), including Marion Island.

SAR processing
BD_DeST uses an automatic image-segmentation-based framework (See Figure 2 and Figure 3) to detect oil slick from moving vessels using SAR images over Southern African oceans. The system uses an automated threshold-based algorithm and a region-based algorithm to achieve a more efficient oil slick detection. First, the threshold-based method detects areas with a high oil slick probability; the region-based method then extracts the full extent of the identified oil slick. The framework is robust to intensity variations, weak boundaries, and is also more computationally efficient.

Figure 2: SAR subimages with segmented linear darkspots image (binary image).

Figure 3: Input image (left), segmented linear darkspots image (middle) and final discrimination image (right).

Reporting – Scheduled reporting to use

Figure 4 contains an example of a bilge dump report.

Figure 4: Bilge Dump Report.

Web viewer – historical events and recent events
Figure 5 contains an example of the bilge dump system dashboard.

Figure 5: Bilge Dump System Dashboard.