Vessel Detection: Monitoring China’s Maritime Militia


The South China Sea is home to significant mineral deposits, oil & gas reserves, and major shipping lanes, making it one of the world’s geopolitical hotspots.

With much at stake, China has been locked in a years-long contest with its neighbors over competing claims of sovereignty.

One of the techniques to assert de facto control has involved hundreds of Chinese fishing boats, which cluster in large groups as a display of force, comprising an unconventional naval force, known as the People’s Armed Forces Maritime Militia (PAFMM).

There is a strong need to keep tabs on China’s maritime militia, but that’s a difficult task given the number and size of boats involved. China’s fishing fleet is the world’s largest, while fishing boats tend to be smaller than other commercial and military vessels. Additionally, the South China Sea is notoriously cloudy, causing difficulty in obtaining frequent optical collections.


Ursa Space offers an automated, multi-site, maritime militia monitoring service that works in any weather conditions, day or night.

Our industry-leading network consists of 35 cutting-edge synthetic aperture radar (SAR) satellites and AIS data.

Seamlessly fused into a single workflow, this backbone enables us to pinpoint fishing vessels in specific harbors and swiftly detect any fluctuations in their numbers, potentially indicating strategic deployments.

Stay one step ahead with our compelling solution that revolutionizes how you monitor and respond to critical maritime activities.


Brief History:

The PAFMM was founded shortly after the end of the Chinese Civil War in the 1940s and was originally used in coastal waters near the Chinese mainland.

However, since the 1970s, it has been used in an increasingly aggressive manner further away from mainland China. In 1974, the PAFMM provided logistical support to the Chinese military during the Battle of the Paracel Islands, which resulted in the Chinese seizure of the islands from South Vietnam.

China has continued to seize territory using this strategy on numerous reefs and islands in the South China Sea, and throughout the 2010s the militia was at the front line of China’s expansion of its de facto territory, further into the South China Sea, and has proven itself as a crucial part of China’s naval force projection strategy.



The benefit of using SAR is the 24/7 coverage that the technology provides. Unlike optical imagery, SAR works regardless of clouds or darkness.

This makes it more reliable than optical, especially over the South China Sea, which is covered in clouds for approximately half of the year. Trying to monitor this region while being forced to wait for clear skies to collect a satellite image is unacceptable.

SAR can easily detect a fishing vessel because the steel hull produces a strong radar return. The reliable consistency allows Ursa to create an automated and scalable monitoring service tailor-made for these vessels.

Moreover, fishing vessels cluster within specific areas of Chinese harbors, or have entire harbors solely dedicated to them.

AIS data confirms that fishing vessels often travel between the fishing harbors and South China Sea hotspots. This heat map shows AIS data over a 19-month period that was broadcast by more than 150 ships that researchers found were likely part of China’s maritime militia.

However, AIS data can be unreliable for several reasons. Ship captains can turn off the AIS transponder (going “dark”) or falsify their location and/or identity (“spoofing”). In addition, fishing boats usually carry a Class B transponder (2 watts), versus the stronger Class A transponder (12.5 watts) carried by larger vessels.


Monitoring Service

Our SAR-based solution entails monitoring the harbors we identified above. When there is a significant decrease in the number of boats present, this could indicate that they have been deployed to the South China Sea to fulfill their maritime militia duties.

This approach can take two forms, depending on resolution type.

Our virtual constellation includes access to cutting-edge, high-resolution SAR imagery from Umbra, a leader in advanced space radar technology, which we can use to automatically detect even small vessels.

High-resolution SAR imagery is required to delineate individual boats because of the manner in which they anchor and raft together in a “vessel chain.”

Another proprietary algorithm is designed to detect individual vessels and then measure its length & width, and determine position and bearing.

We cross-correlate vessel detections with AIS data to determine which ships were “dark” (i.e., not broadcasting AIS signals), providing valuable context for a range of scenarios.

Going dark may indicate that a vessel is hiding its location and identity to conceal illicit activities, such as sanction evasions, or in this case, cluster in disputed waters.

Some of China’s maritime militia vessels are equipped with Class B transceivers, which results in spottier coverage via commercially-available AIS platforms.

A SAR-based solution offers a more comprehensive approach, detecting fishing boats that would not appear in AIS-only data services, and functions regardless of the weather and ambient light conditions that would limit optical imagery.

For example, the image below shows Umbra’s SAR imagery (50-centimeter resolution) collected May 16, 2023 outside Pag-asa Island (Thitu Island). We automatically detected nine vessels which were almost certainly maritime militia, six of which were suspected to be “dark.”

This is a scalable approach that can encompass multiple harbors. Historical data will reveal patterns-of-life that can signal a likely deployment of ships.

To demonstrate this point, we utilized archived imagery collected over a single harbor. However, because the imagery is lower resolution, we could not use the same vessel detection algorithm because the fishing boats are too small to differentiate.



Let’s look at one example from 2017 involving Puyuzhou Bay, a fishing harbor that doubles as a maritime militia base, courtesy of the European Space Agency’s Sentinel-1A satellite.

We can still observe there was a change in the number of fishing boats present, as seen in the example below.

Even without being able to count individual ships, we can capture trends with lower resolution SAR imagery by creating an index.

The starting point is an algorithm that sets a threshold for the amount of SAR energy in each pixel, which approximates whether a boat is present or not. The more pixels above the threshold corresponds with more ships, and vice versa.

We then convert this pixel count into an index to normalize the data.

This graph shows the index over Puyuzhou Bay starting in 2017. There are several instances when the drops were significant enough to suggest a large, coordinated ship deployment, including in late 2022 (circled).

The benefits of Sentinel-1A imagery is that it’s free and open source. However, an average revisit rate of 12 days in Asia may be too infrequent for time sensitive insights, and better suited for strategic, long-term trend analysis.

That said, both SAR-based options offer an automated solution for monitoring a pressing challenge of interest to a wide range of professionals in the security and commercial sectors.





Live count of Ursa Space’s SAR + Optical imagery/data catalog

Learn more about our imagery as a service.