Application of Electro Optical (EO) in Starboard

As part of the Using Satellite Data for Analysis in Starboard series, this article uses practical case studies and technical insights to further explore how Electro Optical (EO) data can be integrated for maritime domain awareness (MDA) in Starboard.

In an MDA context, optical imagery generally provides the clearest and most intuitive form of satellite imagery for detecting vessels and objects on the ocean surface—provided it is collected during daylight hours and under clear-sky conditions. However, there are important trade-offs to consider between different providers and imaging capabilities.

For example, freely available low-resolution imagers can typically detect only vessels larger than approximately 30 metres in length, but offer reliable global coverage. Detecting smaller vessels generally requires sub-metre resolution from commercial satellites, which comes at higher financial cost and reduced footprint size.

A small pleasure craft near a reef, captured using high-resolution EO imagery.

The core technological difference between publicly-accessible sources such as the European Space Agency’s (ESA) Sentinel-2, and at-cost commercial imagers, often lies in their imaging method. Sentinel-2 uses a pushbroom sensor, which collects data line-by-line as the satellite moves along a predetermined orbit. This enables wide-swath coverage (often exceeding 100,000nm²) with consistent revisit rates, but is generally limited to coastal and land area imaging (rather than open ocean).

Try it out: Download the latest acquisition plan KML from Sentinel-2, and drop the file onto the Starboard map to view the satellite’s planned footprints in the coming days.

In contrast, most commercial satellites employ a frame-based sensor, capturing snapshots of smaller areas (~200 nm²) with very high spatial resolution—under 1 metre. This enables finer detail but relies on bespoke tasking requests in highly specific locations, and is subject to satellite availability at the time of collection. See how this works in practice by visiting our Youtube Masterclass.

This high resolution acquisition consists of two subsequent scans which are stitched together in Starboard to create a larger footprint.

Another key distinction is that some commercial satellites also capture high-resolution panchromatic imagery, i.e., broadband grayscale images with finer spatial detail than the multispectral bands. Panchromatic data is particularly valuable for detecting small objects with its sub-metre resolution.

A common use of EO in maritime surveillance is to task high resolution (small area) commercial satellite imagery based on uncorrelated low-resolution (wider area) detections from public data providers like ESA. This workflow is referred to as tip-and-cue, where low-resolution detections “tip” a location of interest, which then “cues” a higher-resolution acquisition of that same area.

When searching for sanctioned tanker vessels that are trying to evade detection on AIS, EO data integration can be a valuable source of validation. Sanctioned oil tankers will often use AIS handshakes to mask their AIS activity during Ship-to-Ship (STS) transfers, as detailed by C4ADS in their article Oil and Water.

The following Sentinel-2 image, which is fused with AIS and global sanction lists in Starboard, shows a sanctioned vessel conducting an AIS handshake. There are two different vessels (both sanctioned) transmitting in the same location on AIS, but with only one physical vessel present in the imagery. This indicates potential spoofing activity in support of smuggling operations.

Another method of AIS obfuscation used by sanctioned vessels is through the use of scrapped vessel identities, dubbed “zombie ships”. Zombie ships are those using the identity of a previously scrapped vessel in order to disguise illegal activities (and true identity). These vessels can be easily identified in Starboard by isolating IMO numbers from the Scrapped Vessel List and analysing their track history against satellite imagery.

Sentinel-2 imagery combined with AIS in Starboard shows EM LONGEVITY in 2023 (left) just before being scrapped, and a different vessel using the IMO two years later (inset, right).

One of the downfalls of using Copernicus data for this type of analysis is that the constellation orbits on a set schedule, which can result in a gap of several days between images of the same area – And often only available in coastal areas. For time-sensitive monitoring work which requires higher persistence, it may warrant the procurement of additional satellite feeds from commercial providers who can produce on-demand imagery on a case-by-case basis. An example of this type of work is provided in the next case study.

Contact support@starboard.nz if you would like to order Copernicus data for your organisation in Starboard.

Smuggling activity by merchant vessels is a growing challenge for maritime authorities, with traffickers using deceptive tactics like identity obfuscation and mothershipping to evade detection. These ships can appear legitimate against the backdrop of routine traffic, but when offloading contraband, real-time monitoring becomes essential.

In 2024, a Tanzania-flagged cargo vessel named MV NAMI (broadcasting on AIS as “POLO”) was monitored by law enforcement as it traveled from Taiwan to Australia, suspected of smuggling illegal tobacco. With insufficient evidence to detain it in port, authorities monitored the vessel as it loitered off the coast of Sydney for three weeks before returning to Taiwan.

Try it out: See the filterset that was used identify potential smuggling vessels entering Australia's EEZ.

The following image shows open-source Sentinel-2 imagery of the MV NAMI at the time of loitering (correlated through AIS), but due to the low resolution of the image, as well as the small stature of these type of smuggling vessels (~60m), it makes it almost impossible to derive any useful information other than its general presence. A higher resolution is required.

Sentinel-2 imagery of the vessel at the time of loitering inside the Australian EEZ. The image resolution is 10m and the vessel length is 60m, which has resulted in a visual blending effect with the background ocean.

Many high resolution satellites are not able to capture more than ~200 nm² of ocean per scan (around the size of Singapore), which severely hampers attempts at imaging a moving vessel, and even more so when trying to find dark vessels in open ocean. To mitigate this, two different types of tip-and-cue methods can be employed.

To illustrate how this activity may appear in practice, consider the following high resolution EO acquisition as if it were tipped from the MV NAMI Sentinel-2 detection. While unrelated to that case, the image shows a similarly sized mothership deploying multiple go-fast boats, demonstrating how high-resolution imagery can be used to monitor such activity after a mothership’s location is first identified by lower-resolution, wide-area sensors.