cff-version: 1.2.0
title: "EnvShip-Bench Cross-Domain Extension: A Multi-Jurisdiction, Multi-Scale, Context-Aware Ship Trajectory Prediction Benchmark"
message: "If you use EnvShip-Bench in your research, please cite this work and the upstream AIS providers."
type: dataset
authors:
  - family-names: Ma
    given-names: Kun
    email: kunma1220@gmail.com
    orcid: ""
abstract: >
  The EnvShip-Bench cross-domain extension augments the original DMA-only
  benchmark with three additional jurisdictions (NOAA, Piraeus, Norway)
  and a long-horizon Track B (30 minute observation, 60 minute prediction).
  The four subsets together provide 330,000 Track A samples and 106,857
  Track B samples curated across four operationally distinct regimes:
  Northern European coastal traffic, US open-ocean transits,
  Mediterranean port and ferry traffic, and Norwegian fjord and coastal
  traffic. Each sample is paired with a 6-channel OpenStreetMap raster,
  2-channel signed-distance field, exact polylines, 14-dimensional scene
  descriptors, and up to ten neighbour vessels within a 3 km radius
  (with CPA / TCPA). For the Piraeus subset (AIS year 2019), we
  additionally release an env / SDF stack rebuilt with the contemporary
  Geofabrik 2020-01-01 OSM snapshot to eliminate temporal mismatch with
  recently-built port infrastructure. A new Stage 17 quantifies and
  flags OSM temporal-consistency per sample.
version: "1.0"
date-released: 2026-06-13
keywords:
  - "ship trajectory prediction"
  - "AIS"
  - "maritime benchmark"
  - "cross-domain"
  - "environmental context"
  - "signed distance field"
  - "social interaction"
  - "long horizon forecasting"
license: CC-BY-4.0
repository-code: "https://github.com/mark000071/envship_v2_datasets"
url: "https://huggingface.co/datasets/mark000071/envship_v2_datasets"
preferred-citation:
  type: conference-paper
  title: "EnvShip-Bench: A Cross-Domain Multi-Scale Benchmark for Context-Aware Ship Trajectory Prediction"
  authors:
    - family-names: Ma
      given-names: Kun
  collection-title: "Proc. IEEE International Conference on Data Engineering (ICDE)"
  year: 2026

# Upstream AIS providers — please cite alongside this work
references:
  - type: dataset
    title: "Danish Maritime Authority AIS Data (aisdk-2025-09)"
    authors:
      - name: "Danish Maritime Authority"
    url: "https://www.dma.dk/safety-at-sea/navigational-information/ais-data"
    license: CC-BY-4.0
  - type: dataset
    title: "NOAA / MarineCadastre AIS Data, March 2025"
    authors:
      - name: "U.S. National Oceanic and Atmospheric Administration"
    url: "https://marinecadastre.gov/"
    notes: "U.S. government work in the public domain."
  - type: dataset
    title: "The Piraeus AIS Dataset for Large-scale Maritime Data Analytics"
    authors:
      - family-names: Tritsarolis
        given-names: Andreas
      - family-names: Kontoulis
        given-names: Yannis
      - family-names: Theodoridis
        given-names: Yannis
    year: 2022
    doi: "10.5281/zenodo.6323416"
    license: CC-BY-4.0
  - type: dataset
    title: "Kystverket / Kystdatahuset AIS positions"
    authors:
      - name: "Kystverket (Norwegian Coastal Administration)"
    url: "https://kystdatahuset.no/"
    license: NLOD-2.0
  - type: dataset
    title: "OpenStreetMap contributors"
    authors:
      - name: "OpenStreetMap contributors"
    url: "https://www.openstreetmap.org/copyright"
    license: ODbL-1.0
    notes: "Cartographic basis for all environmental rasters and signed-distance fields."
  - type: dataset
    title: "Geofabrik OpenStreetMap historical extracts (greece-200101.osm.pbf)"
    authors:
      - name: "Geofabrik GmbH"
    url: "https://download.geofabrik.de/"
    license: ODbL-1.0