Prioritizing unmatched vessel detections when you've got one patrol boat

Several small fishing boats scattered far apart across a calm sea under a dawn sky.

Most MCS units don't have a patrol asset problem so much as a prioritization problem. You get a list of twelve, twenty, sometimes forty unmatched detections for the day, and you have one patrol boat, maybe an aircraft if the weather cooperates and the budget allows. Chasing all of them isn't an option. Picking the wrong three is how a real incursion sits unbothered for a week while a crew burns fuel on a vessel that turns out to be a legitimate transit with a broken transponder.

The triage question isn't "is this detection real." Optical confirmation already answered that. The question is which of the real, unmatched detections deserves the boat today.

What actually moves a detection up the list

Four things matter more than the rest, and none of them require modeling.

Distance to the line. A detection sitting two miles inside your EEZ boundary or an MPA buffer outranks one forty miles into open water, even if the second one is a bigger vessel. Boundary proximity is the single fastest way to cut a long list down to a short one.

Persistence at the same spot. One unmatched detection in a given area on a given day could be a transit vessel with its AIS off for routine maintenance. The same position showing up unmatched on day two and day three starts looking like a vessel that's working a ground, not passing through. Run your daily exports against the prior few days before you rank anything. A single hit and a three-day pattern don't belong in the same tier.

Size class relative to what fishes that water. A 15-meter detection in a zone where the licensed fleet runs 25-meter trawlers is worth a second look on its own. Size class from the detection record, cross-referenced against what your licensing office says should be out there, does a lot of the filtering for you before you even think about dispatch.

Heading and loitering behavior, where you have it. A detection on a straight transit line reads differently than one that's been sitting still or circling. If your optical pass gives you any sense of motion, or if a follow-up pass a few hours later shows the vessel hasn't moved, that's a stronger pull on the boat than a vessel that looks like it's just passing through on its way somewhere else.

Building the ranked list without new software

You don't need a scoring algorithm. A watch desk can do this with the daily CSV and a whiteboard. Tag each unmatched detection with a letter for boundary distance (A for inside five miles of a line, B for five to fifteen, C beyond that), a number for how many consecutive days it's shown up unmatched in roughly the same position, and a flag if the size class is out of pattern for the area. Sort on those three columns and the dispatch order mostly writes itself. That sort turns forty detections into the three worth fuel and crew hours, and it takes under ten minutes at shift handover once the tagging habit is in place.

This is also where a clean daily feed earns its keep. If your unmatched list already comes pre-correlated, with position, size class, and timestamp sitting in one row instead of scattered across a radar log and an AIS printout you reconciled by hand, the triage step above takes minutes instead of most of a shift. Illegal Fishing Detection builds that feed as a daily export so the desk starts the ranking exercise with a clean list instead of a raw correlation job.

What this doesn't replace

None of this tells you what's actually happening on deck. A boundary-proximity, persistence, and size-class ranking gets you a dispatch order, not a verdict. The crew that boards still makes the call on the water, and the watch desk still logs what came back so tomorrow's ranking carries yesterday's outcome. The ranked list sets the day's dispatch priority; the patrol asset and the log afterward are where the actual enforcement work happens.

If your desk is still building that ranked list by hand from separate AIS and detection logs, it's worth seeing what a pre-matched daily export looks like before the next shift starts.

← Back to the blog