Trimtabist

What is observed and what is merely declared. Position, speed and the elapsed clock are observed — GPS off the transponder, timed by our own receivers. Whether a stationary ship is at a berth or at anchor is inferred from physics, not from the crew-set status field: a ship at anchor swings through a wide arc as wind and tide turn it, a ship on mooring lines cannot move. Destination and draught are typed by the crew and are shown as declarations — measured, not believed.

Why elapsed hours and not a congestion figure. Any aggregate we could publish today is biased toward zero by three separate mechanisms: ships still waiting when the window closes are dropped, and they are the longest waits; berth durations are clipped by the length of the archive; and laid-up hulls pad berth time. All three flatter the ports. Elapsed state is immune, because nothing has to finish. The waiting rate follows once the archive is long enough for the first two to stop mattering.

Coverage boundary. A ship counts as at a gateway only within 25 nm, where the attribution is unambiguous. Further out it appears only if its declared destination names that gateway — our receivers see plenty of traffic that is not coming here at all. Recording reaches roughly 200 nm, so a queue held well offshore is visible, but this is not a complete census of ships bound for the US and must not be read as one.

Two exclusions, 27 hulls in total right now. Government sealift — Ready Reserve Force and USNS hulls pass the merchant filter but hold a berth for months without working cargo; commercial ships with similar names are kept, because the published roster is matched rather than the name pattern. Hulls the registry identifies as something else — bulk carriers, car carriers and ro-ros are AIS cargo type 70–79 too. A ship whose type is simply unknown is kept: dropping traffic we cannot classify would thin the sample and could hide real congestion. Vessels under 150 m are out of scope. Window 36 h.