Skip to content

LICENSE-DATA.md

Latest commit

 

History

History
88 lines (69 loc) · 5.8 KB

LICENSE-DATA.md

File metadata and controls

88 lines (69 loc) · 5.8 KB
Raw

MARIS-Forecast — Data license and per-file license mapping

This file covers the dataset (the trajectory tables and context artefacts distributed on Zenodo, DOI 10.5281/zenodo.21224009, and mirrored on the Hugging Face Hub). The source code in this GitHub repository is separate and is released under the MIT License.

MARIS-Forecast is a composite release: different files, and different columns within a single table, are derived from different upstream sources and therefore carry different licences. The mapping below is authoritative. When you redistribute all or part of the data, apply the licence that governs the specific files/columns you use, preserve the attribution notices in NOTICE.md, and — for any OpenStreetMap-derived content — comply with the ODbL share-alike obligation described in Section 3.

1. Upstream AIS source licences (trajectory / kinematic content)

The vessel-trajectory tables (per-split part-000.csv.gz) and the social-context records are derived only from the AIS feed of their region and follow that feed's terms:

Region Directory prefix Source & terms
Denmark (DMA) *dma* Danish public-sector information — released by the Danish Maritime Authority under the Danish Act no. 596 of 24 June 2005 on the further use of public-sector information (the PSI Act). Free to download and reuse, with no warranty ("The Danish Maritime Authority does not guarantee the correctness of the AIS data … and is not liable for any damages"), and on the condition that the data are not combined so as to identify persons without a permit. This is not a Creative Commons licence. Source: https://www.dma.dk/safety-at-sea/navigational-information/ais-data
United States (NOAA) *noaa* U.S. Government work in the public domain (no copyright). NOAA/BOEM MarineCadastre. Source: https://marinecadastre.gov/ais/
Greece (Piraeus) *piraeus* CC BY 4.0 — Tritsarolis, Kontoulis & Theodoridis (2022), Zenodo DOI 10.5281/zenodo.6323416.
Norway (Kystverket) *norway* NLOD 2.0 (Norwegian Licence for Open Government Data). Kystverket / Kystdatahuset, https://kystdatahuset.no/.

2. Reanalysis-derived columns (weather & sea state)

The fifteen anchor-time context scalars include six weather columns (met_*) and four sea-state columns (sea_*). These are derived from the ERA5 reanalysis (Copernicus Climate Change Service / ECMWF) and the ECMWF WAM wave model, obtained via the Open-Meteo archive. They are governed by the Copernicus licence / ECMWF terms and Open-Meteo's terms (attribution to Copernicus/ECMWF and Open-Meteo). "Contains modified Copernicus Climate Change Service information."

3. OpenStreetMap-derived content — ODbL (share-alike)

All environmental context artefacts are derivative databases of OpenStreetMap and are distributed under the Open Database License v1.0 (ODbL), © OpenStreetMap contributors. ODbL is share-alike: if you publicly use a Derivative Database that incorporates any of these files/columns, you must license that Derivative Database under ODbL and keep the OpenStreetMap attribution. This obligation is not waived by the "composite" nature of the release.

The following files and columns are ODbL:

Artefact (relative to a leaf) Kind
context_v1/environment/rasters/<split>/masks.npz six-channel OSM land/water/navigable/boundary/barrier raster
context_v1/environment/rasters/<split>/signed_dist_shore.npy shore signed-distance field
context_v1/environment/rasters/<split>/signed_dist_nav.npy navigable-water signed-distance field
context_v1/environment/vectors/<split>/vectors.jsonl.gz clipped OSM shoreline / waterway / pier / quay / breakwater polylines
context_v1/environment/features/<split>/environment_descriptors.csv OSM-derived scene descriptors (scene type, water/navigable ratios, boundary densities, env-quality score)
context_v1_2019osm/** (Piraeus) historical-OSM (2020-01-01) rebuild of the above
osm_temporal_consistency/<split>_flags.csv Stage-17 flags computed from the OSM shore SDF
Inline columns in <split>/part-000.csv.gz: port_nearest_dist_km, port_nearest_name, in_fairway, dist_to_fairway_m, in_tss, osm_temporal_consistent, osm_max_inland_depth_m, osm_n_inland_points, osm_max_consec_inland_run OSM-derived port / fairway / TSS / consistency columns

Historical OSM extracts (e.g. greece-200101.osm.pbf) are courtesy of Geofabrik and are likewise OSM under ODbL.

4. Mixed tables — how the split CSV is licensed by column

Each part-000.csv.gz (and the merged augmented/<split>/part-000.csv.gz) is a mixed table:

  • AIS-derived columns — identifiers, local-metre geometry (hist_*, fut_*), kinematics (*_sog_json, *_cog_*, *_heading_*), quality scalars, split — follow the region's AIS source licence (Section 1).
  • Reanalysis columns — met_*, sea_* — Section 2.
  • OSM-derived columns — the port/fairway/TSS/consistency columns listed in Section 3 — are ODbL.

A Derivative Database that retains the OSM-derived columns or the context tensors must be released under ODbL (Section 3). If you keep only the AIS-derived trajectory columns and drop all OSM-derived content, ODbL does not attach and the region's AIS source licence governs.

5. Attribution

Regardless of which subset you use, preserve the attribution notices in NOTICE.md and cite MARIS-Forecast (CITATION.cff). Map data © OpenStreetMap contributors, ODbL. Contains modified Copernicus Climate Change Service information (ERA5) and ECMWF data via Open-Meteo.