How Competitive Intel Product Teams Use Marine Transportation Data

Event data is the tier worth paying for, because a position alone cannot tell you whether a call was planned. MarineTraffic's contracted REST API exposes /portcalls, /berth-calls, /vesselevents, /voyageforecast and /port-congestion alongside /expectedarrivals and /predictive-arrivals, returning paired ARR_TIMESTAMP_UTC and DEP_TIMESTAMP_UTC values with PORT_UNLOCODE and BERTH_NAME fields - the documented sample shows QUEEN MARY 2 arriving at Southampton (GBSOU), Berth 101 of the City Cruise Terminal, at 2020-10-14 05:26:00 UTC. Historical tracks reach back to January 2015 at up to 2-minute resolution. Access is credit-metered and each key is technically capped at 100 total requests per minute, so define the watchlist before wiring alerts.

Two archives cover the replay job, and neither invoices you. MarineCadastre AIS Vessel Traffic Data (USCG Nationwide AIS) - the slice's only quality-score-10 record - republishes US Coast Guard Nationwide AIS point records for 2009 through 2025 under commercial delivery terms, downsampled to one position per vessel per minute across 17 documented fields including vessel_name, imo, draft, vessel_type codes and cargo code. Daily Zstd-compressed CSV starts in 2015 (monthly by UTM zone before that) and 2024 GeoParquet files reach about 325 MB each. One gap shapes model design: 2010-2014 files encrypt MMSI numbers and drop ship names at USCG request, so clean per-vessel histories begin in 2015.

Europe's equivalent trades granularity for coverage: EMODnet compresses satellite AIS into 1x1 km monthly vessel-density rasters expressed as hours per square kilometre, split into 13 ship-type layers from 2017 to 2024 under commercial delivery terms 4.0. Route-level intensity answers whether a competitor's cargo actually sails the corridor it claims - with the publisher's caveat that satellite data loss from June onward biases 2024 densities low.

Movements show intent; statistics show outcome. BTS Maritime Data and Port Performance Freight Statistics ranks the top 25 US container ports by TEU - the 2023-vintage table puts the Port Authority of New York and New Jersey at 5,352,134 import TEUs against 1,252,065 export TEUs - and adds monthly container and tanker dwell times plus weekly counts of container ships waiting to dock (July 2021 through September 2023), all through a keyless Socrata SODA API in the US commercial delivery terms. When a terminal operator blames congestion for a missed window, the waiting-ships series is the receipt.

Eurostat Maritime Transport Statistics (mar_go / mar_pa / mar_tf) mirrors the benchmark for Europe: gross weight handled and container TEUs per port collected under Directive 2009/42/EC, annual series reaching back to 1997, released within 10 months of the reference quarter - confirmation, not tripwire. USACE Waterborne Commerce Statistics Center (WCSC) digs below port level: TOWS trip and cargo files for 2004 through 2023 split short tons into receipts, shipments, intra-port and through movements by WCSC/SITC Rev 3 commodity code, so a rival's claimed commodity growth is checkable harbor by harbor.

UNCTADstat closes the international frame: the quarterly Liner Shipping Connectivity Index covers about 200 economies, and its AIS-derived port-call tables report median time in port by vessel type - built on data supplied by MarineTraffic, which tells you where even the UN gets its operational signal.

IMO GISIS - Global Integrated Shipping Information System handles counterparty risk. After free registration, its 20+ public modules expose member-state-reported marine casualties, piracy and armed robbery reports and ship/company particulars keyed by IMO number - the safety-record check before chartering an unfamiliar operator. The NGA World Port Index (Pub 150) supplies the reference join table: 3,801 ports across 109 attribute columns covering channel depths, pilotage and container, oil-terminal and LNG facility flags, refreshed monthly as a roughly 3.5 MB CSV.

When a thesis spans decades, UK Maritime and Shipping Statistics (DfT) reaches port freight tonnage back to 1965 - long enough to document structural change such as Felixstowe going from absent in the 1965 column to 6,536 thousand tonnes lifted by 1980, while Clyde shrank from 15,304 to 8,845 over the same years.

Ranked by how much competitor behaviour each record exposes per dollar and per engineering hour, 10 records make the cut. The top four cover event attribution (MarineTraffic), counterparty structure (VesselFinder), historical replay (MarineCadastre) and live detection (AISStream.io) - three of the four free. The statistical tier below trades latency for authority: BTS, Eurostat, USACE and UNCTADstat are all free official releases, while Global Fishing Watch and IMO GISIS close the diligence gaps a movement feed cannot.

Sequenced by how soon each step pays off, using only the sources named above:

1. Backfill baselines. Load MarineCadastre daily CSV or GeoParquet for 2015 onward (earlier years encrypt MMSIs) and compute per-port call counts and port-stay durations for watched fleets.

1. Wire the live layer. Stream AISStream.io for whole-world positions and mid-voyage destination changes; add BarentsWatch historic tracks by MMSI when Norwegian waters matter.

1. Attribute events. Where budget exists, log MarineTraffic port-call, berth-call and voyage-forecast records per rival to separate scheduled calls from unscheduled ones.

1. Verify quarterly. Diff BTS and Eurostat throughput, USACE commodity flows and UNCTADstat connectivity against your stored baseline, and run new counterparties through GISIS casualty and piracy checks before signing anything.