4.1.6 — Maritime Intelligence — maturity: live
Multi-Source Maritime Fusion
Cross-modal fusion algorithms that combine AIS, SAR, RF and optical detections into a single, deduplicated, identity-resolved vessel track.
Weaving AIS, SAR imagery, RF geolocation, and optical feeds into a single sovereign picture of every vessel in your exclusive economic zone — before a crisis demands it.
Each maritime sensor sees a different shadow of the same vessel. AIS gives an identity but no presence guarantee. SAR gives a hard hull detection but no identity. RF geolocation gives a position estimate but no class. Optical confirms class but only in good light. Multi-source fusion is the layer that takes these heterogeneous, non-aligned, partially-overlapping detections and produces a single unified track per vessel — with an identity, a confidence score, and a behaviour history.
This is increasingly where competitive advantage in maritime intelligence lives. Once the underlying sensors become commodity (and they are), the differentiation is in the fusion: can the system correctly merge the AIS broadcast at 14:02 with the SAR detection at 14:11 and the RF emission at 14:23 into one vessel, with high confidence, while keeping a thousand other near-by detections separate? The leading providers (Windward, Spire's Maritime AI, Kpler, Lloyd's List Intelligence, EMSA's IMS) compete heavily on fusion quality. The technique stack draws from track-association theory developed for air-defence radar fusion, modern probabilistic graph models, and increasingly LLM-based reasoning over unstructured records (port logs, ownership filings, news).
Frequently asked
What exactly is 'fusion' in this context, and why can't we just use AIS alone?
AIS provides self-reported identity and position, but any vessel wishing to evade detection can simply switch it off or spoof coordinates. Fusion layers in independent sources — SAR radar imagery, optical satellite imagery, VHF/RF geolocation, and LRIT — so that a vessel showing no AIS signal but detected by SAR can still be tracked, identified, and flagged. A sovereign fusion capability means your analysts see the complete picture, not just the vessels that want to be seen.
How many satellites does a credible sovereign maritime-fusion constellation actually require?
A minimum viable constellation for a medium-sized EEZ (roughly 1–2 million km²) needs approximately 6–12 SAR microsatellites for sub-4-hour revisit, complemented by 15–20 nanosatellites carrying AIS receivers and an RF-geolocation payload cluster of at least 3 satellites. That is achievable within a five-year programme and a capital budget under $400 million — well within reach of mid-income maritime nations.
How does this differ from simply subscribing to a commercial maritime intelligence service like MarineTraffic or Windward?
Commercial services aggregate and sell access to data collected by third-party satellites; you see what the vendor chooses to share, on pricing and licence terms they set, and your access can be suspended under sanctions regimes, vendor M&A, or geopolitical pressure. A sovereign programme means raw data is downlinked to a ground station you own, processed on infrastructure you control, and classified at sensitivity levels you determine — with no third party able to revoke access the morning a crisis begins.
Can a small island developing state realistically afford its own maritime-fusion satellites?
Not necessarily alone. The most practical path for SIDS is a regional constellation shared among neighbouring states — similar to the Pacific Island Countries' work with the Pacific Community (SPC) — where a joint programme amortises satellite and ground-segment costs while each nation retains sovereign access to data covering its own EEZ. ITU filing and orbital slot registration can be handled collectively without diluting individual sovereignty.
What happens to the data feed during a geomagnetic storm or satellite outage?
A well-designed sovereign architecture maintains at least two independent observation modalities (e.g., SAR plus RF geolocation) so that the loss of one satellite or payload type does not create a complete blind spot. Additionally, coastal HF radar networks — operated by agencies such as national coast guards — provide a terrestrial backstop with 200–300 km range that remains unaffected by orbital anomalies.
How quickly can a fused picture be updated after a vessel of interest is detected?
With current commercial-grade LEO SAR constellations (ICEYE, Capella, Umbra), tasking to image delivery runs 30–90 minutes. Fusing that image with an existing AIS track and RF geolocation fix adds another 5–15 minutes of automated processing. A sovereign system replicating this architecture would target a full-cycle latency under 60 minutes from anomaly detection to authoritative track update and alert generation.
What international legal framework governs acting on fused maritime intelligence — for example, boarding a dark vessel?
UNCLOS Articles 73, 110, and 111 govern enforcement rights within the EEZ and on the high seas, including the right of hot pursuit and boarding in cases of illegal fishing or stateless vessel suspicion. Satellite-derived evidence is increasingly accepted in coastal-state courts and by regional fisheries bodies such as CCAMLR and WCPFC, but evidence-handling chains must be documented meticulously — timestamped, georeferenced imagery with unbroken custody logs — to survive legal challenge.
How do we handle the firehose of data — won't we need an AI/ML layer?
Yes. At the scale of a national EEZ, manual correlation of AIS, SAR, and RF tracks is operationally impossible. Proven approaches include anomaly-detection models trained on historical AIS trajectories (as deployed by Global Fishing Watch and Windward), SAR-to-AIS correlation algorithms, and vessel-type classifiers applied to SAR imagery. The critical sovereignty point is that these models must run on infrastructure you own and must be auditable by your own analysts — black-box commercial AI creates a second layer of dependency on top of the data-dependency problem you are already trying to solve.