Somalia has roughly three thousand kilometres of coastline, the longest of any mainland African country, and a fisheries resource that several other people’s fleets have found considerably more interesting than the licensing regime would suggest. Any serious attempt to manage that runs into one question early: what is actually out there?

The instinctive answer is vessel tracking data. It is the wrong place to start, and understanding why saves a great deal of money.

What transponder data is good at

Automatic Identification System data is genuinely excellent within its design intent. It gives you identity, position, heading, speed, destination and history — things no image can provide. Coverage of compliant commercial traffic is broad, satellite-collected AIS extends it well beyond coastal receivers, and the data is comparatively cheap.

If your question is “where is the traffic, who are they, and where are they going”, AIS answers it well.

What it is structurally blind to

AIS is a cooperative system. It reports what a vessel chooses to transmit, which produces three gaps that all point the same way.

Transponders get switched off. Not accidentally, and not randomly — they go quiet in the places and at the times somebody would most want to look. A tracking picture built only on AIS is a map of the fleet that has no reason to hide.

Transponders can report fiction. Position and identity are self-declared. Spoofing is well documented globally and requires no sophistication.

Many vessels carry nothing. Small craft, artisanal fleets and a good deal of what operates in Somali waters are simply outside the carriage requirements.

The consequence is uncomfortable but important: the vessels most relevant to an enforcement question are the ones least likely to appear in the dataset most programmes buy first.

What imagery adds

Satellite imagery and radar detect hulls. They do not care what is being transmitted, and radar works through cloud and at night, which is when a great deal of activity happens.

The limitation is the mirror image of AIS: a detection is not an identification. Imagery tells you a vessel of roughly a certain size is at a certain place at a certain time. It does not tell you its name, flag or owner.

The useful method is the difference between the two

Neither dataset alone produces a maritime picture. Cross-referenced, they produce something considerably more useful than either: the set of detections that do not correspond to any broadcasting vessel.

That gap is the actionable output. It is the basis for tasking a patrol at a specific location rather than on a rumour, for establishing whether activity in a zone is consistent with the licences issued for it, and for building an evidence record over a season rather than an anecdote.

It also scales in a way patrolling does not. Nobody can patrol three thousand kilometres of coastline. Observation narrows that to a short list, which is the only way limited assets get used well.

What this supports, realistically

Licensing and revenue. A fishing licence regime is worth operating only if there is some means of knowing who is fishing. Detection of activity inconsistent with issued licences turns an honour system into an enforceable one, and the revenue case is usually what funds the rest.

Patrol tasking. A detection with a position and a timestamp is a defensible reason to send an asset. It is also defensible afterwards, which matters when the decision is questioned.

Pattern over time. Single detections are interesting. Repeated activity in the same zone, at the same season, is evidence — and it builds a case that a one-off image cannot.

Coastal change. The same imagery answers slower questions: port and anchorage use, shoreline retreat, infrastructure change along the coast.

What we would tell you before you commissioned anything

Maritime domain awareness is frequently sold as a finished national system. It is worth being clear about the boundary.

Ground segment, communications to vessels, imagery acquisition, detection and analysis are infrastructure and data problems, and they can be delivered from inside Somalia rather than from a foreign hub. Sensors on aircraft and vessels, patrol assets, prosecution capability and the legal machinery around enforcement are not that, and any provider offering all of it under one heading is either partnering quietly or overreaching.

We would rather name the boundary at the first meeting than have a programme discover it at the third. A capability that delivers detections to an authority with no means to act on them is not a maritime programme — it is a subscription.

Where to start

Define the zone, the revisit rate the decision needs, and what happens when something is found. If the answer to the last one is not yet clear, that is the piece to solve first — the observation is comparatively easy.