NGA World Port Index (Pub 150)

One reference table quietly underwrites almost every port-level analysis in shipping. The NGA World Port Index (Pub 150) is the National Geospatial-Intelligence Agency's global directory of maritime ports: about 3,801 records, one per port, harbor or oil terminal, each carrying 109 attribute columns. Where a vessel-tracking feed tells you where ships went, the WPI tells you what waits for them — whether the harbor is a Very Small river anchorage or a large coastal port, how deep the channel runs, whether a pilot is compulsory, and whether the piers can take containers, a Ro-Ro ramp, liquid bulk or an LNG carrier.

Attributes fall into families a data team can reason about. Identity and placement: WPI number, published and alternate names, UN/LOCODE, country, sailing-direction region, water body and a WGS84 position in decimal degrees. Physical character: harbor size, type, use and shelter afforded, tidal range, entrance width. Depths and dimensional limits: channel, anchorage and cargo-pier depths in metres, and the largest vessel the port takes — length, beam and draft, alongside and offshore. Services and handling: pilotage indicators, quarantine and security flags, communications, container / Ro-Ro / solid and liquid bulk / oil and LNG terminal flags, mooring types and crane capacities. Provisions: fuel oil, diesel, potable water, provisions, repairs, dry dock and railway.

Inside Datadory's marine transportation slice of 15 primary records it is the spatial spine — the table every AIS track, tonnage series and connectivity score joins onto. Get a sample of this dataset cut to the regions and port lists you track.

One row per port, flattened and analysis-ready — the rows below are drawn from the verified sample:

# One row per port/harbor/oil terminal — key columns shown of 109
World Port Index Number   7950              # unique WPI identifier
Main Port Name            Maurer
Country Code              United States
Harbor Size               Very Small
Harbor Type               River (Natural)
Latitude                  40.533333
Longitude                 -74.250000
# ...102 further columns on this row: depths, pilotage,
# facility flags, supplies...

World Port Index Number   47620             # Iharana, Madagascar
Harbor Size               Very Small
Harbor Type               Coastal (Natural)
Latitude / Longitude      -13.350000 / 50.000000

World Port Index Number   47020             # Chake Chake, Tanzania
Harbor Size               Small
Harbor Type               Coastal (Natural)
Latitude / Longitude      -5.250000 / 39.766667

Three things worth noticing. First, World Port Index Number is a clean integer key — the join column every port-level table wants, immune to the spelling drift that plagues name matching. Second, Harbor Size and Harbor Type are categorical, so "every coastal-natural harbor above small in the western Indian Ocean" is one filter pass rather than a research project. Third, the position arrives in decimal degrees, so the rows drop straight into any GIS layer or geocoder with no parsing step. The spread here — a US river port, two Madagascan roadsteads, a Tanzanian coastal harbor — is the point: coverage runs to the long tail of anchorages that commercial directories skip.

The 109 columns organize into the families below. Two do the heavy lifting in most workflows: the position pair keys every spatial join, and the depth-plus-maximum-vessel cluster answers the question chartering desks ask first. Definitions come from the publisher's own field documentation and were verified during research, so the dictionary is shape-exact rather than inferred; per-column naming for your pipeline is locked when your sample is prepared.

• Geography: global — all maritime nations, about 3,801 ports. The frame reaches down to very small river and coastal anchorages that commercial port directories never bother to list, which is exactly where route studies and risk models spring leaks.
• Temporal: a snapshot-of-record. This is the canonical current edition — one authoritative state of the world's ports, not a time series. Historical and forward-looking depth comes from the freight-statistics companions in the same slice, which is the correct division of labor: reference now, flows over time.
• Granularity: one record per port, harbor or oil terminal. Sub-facility detail stops at the port boundary — you know a port handles LNG, not how many jetties serve it. Oil terminals that stand apart from their parent port do earn their own rows.

On Datadory's 0–10 quality rubric the record scores 9, against a 7.81 average across the 1,744 cataloged datasets — verified field documentation and global breadth at port grain, discounted only for being a snapshot rather than a series.

API, files, or your warehouse. Daily, weekly, or hourly.

You pick the channel and the cadence; the field dictionary above travels unchanged across all three. Tabular files suit overnight warehouse loads, structured payloads suit event-driven pipelines, and the flat port table lands in spreadsheet- and SQL-friendly shapes for analysts who live in a BI tool. Because the grain never changes — one row per port, always — switching cadence later is a settings conversation, not a re-integration project.

• Canonical port master table — reconcile a messy internal port list against WPI numbers, alternate names and UN/LOCODEs, and walk away with one deduplicated reference keyed for everything else you join.
• Feasibility and network screening — filter by harbor size, channel depth and maximum vessel dimensions before any routing analysis; the LNG-terminal flag alone collapses days of manual port research into one query.
• Joining AIS tracks to places — land vessel positions onto named ports so movement data resolves to geography; the MarineCadastre AIS pairing is the textbook version.
• Tonnage benchmarking — give regional freight statistics a complete denominator: national series cover their own ports, the WPI covers everyone's.
• Port-services market sizing — enumerate every port with the facility profile your product needs, from bunkering to ship-repair, before the first sales call.
• Cited reference work — anchor journalism and academic papers to an authoritative government-published port inventory with clean attribution.

Data scientists (tagged use case: a 3,800-port global database as bulk files) treat the WPI as the static spine behind dynamic vessel data — enrichment that never changes schema mid-project; workflow notes in data scientists use cases. Developers & builders (tagged: simple tabular/GIS joins) wire it in as the first table of any shipping data model because the integer key makes every downstream merge trivial; wiring notes in developers builders use cases. Market researchers (tagged: canonical global port reference) size port-adjacent markets against a complete enumeration instead of a vendor's customer list; patterns in market researchers use cases. Journalists, academics & students (tagged: authoritative reference) cite it where commercial directories offer nothing citable; see journalists academics use cases.

• MarineCadastre AIS Vessel Traffic Data (USCG Nationwide AIS) — the movement half of the canonical pair: tens of billions of position reports for US waters, 2009–2025, resolving to these port rows (MarineCadastre).
• USACE Waterborne Commerce Statistics Center (WCSC) — US waterborne tonnage by port, waterway and commodity for 2004–2023; domestic flows that gain a global frame from this table (WCSC).
• BTS Maritime Data and Port Performance Freight Statistics — annual top-25 US container ports by TEU plus dwell times; the WPI widens the lens from 25 ports to about 3,801 (BTS).
• Eurostat Maritime Transport Statistics — EU port gross-weight and TEU series reaching back to 1997; the European flow counterpart (Eurostat).
• Natural Earth World Ports (1:10m Cultural Vectors) — a generalized 1,081-port cartographic cousin; lighter for maps, thinner by roughly two-thirds for analysis (Natural Earth).
• Glossary notes — what the World Port Index is and where the Liner Shipping Connectivity Index fits alongside it.

Two honesty notes before you request a sample. The 3,801 count is the figure in the edition verified at research time; the count moves slightly between editions, so confirm the exact row count against delivered rows rather than hard-coding it. And the agency itself states the WPI does not replace detailed charts and publications for operational planning — screen and join with it, then go to chart-level sources before anything sails. Full slice ranking sits on the marine transportation data hub, with the ranked view at best marine-transportation datasets.