Data sources
Every provider behind the score and the globe, its license, and the difference between a live signal and a bundled sample.
Score domains
| Domain | Provider | License |
|---|---|---|
| Land cover | GLAD / Hansen Global Forest Change (VCF5KYR) | CC BY 4.0 |
| Wildfire | NASA FIRMS | Open |
| Air | CAMS, the Copernicus Atmosphere Monitoring Service | Copernicus open license, attribution required |
| Ocean warming | NOAA OISST V2.1 and Coral Reef Watch | US Government work, open |
| Ocean acidification | No provider is assigned. The domain is open for proposals. | Not applicable yet |
| Cryosphere | WGMS is cited for the glacier dataset. No feed is assigned to the domain. | Open, attribution |
| Biodiversity | GBIF occurrence signal | CC BY-NC for the derived clustered points |
| Anthroposphere | Global Human Modification index | CC BY 4.0 |
| Conservation | IUCN and WDPA, coverage signal only | WDPA terms |
Two rows above name no provider. Ocean acidification and cryosphere each carry a weight of 0.10 and neither has a feed assigned to the domain. They are listed here rather than left out because a gap that is 20 percent of the methodology weight should be visible on the sources page, not only in the code.
Copernicus attribution
CAMS requires its attribution to be reproduced. Any use of the air domain carries it:
Generated using Copernicus Atmosphere Monitoring Service information 2026Annotation layers on the globe
The globe carries point layers beyond the score domains. Each one ships with its provider's license.
| Layer | Provider | License |
|---|---|---|
| Power plants | WRI Global Power Plant Database v1.3.0 | CC BY 4.0 |
| Extraction sites | Maus and others, global mining polygons v2 (PANGAEA) | CC BY 4.0 |
| Industrial sites and top emitters | Climate TRACE | CC BY 4.0 |
| Datacenters | PeeringDB facility records | Open, attribution required |
| Carbon offset projects | Berkeley Carbon Trading Project voluntary registry | CC BY 4.0 |
| Protected areas | Wikidata protected areas, WDPA identifiers retained | CC0 |
| Environmental nonprofits, US | IRS Exempt Organizations Business Master File, placed by US Census ZCTA centroids | US Government work, public domain |
| Base map | Natural Earth | Public domain |
What ships offline is not what the provider published
Every bundled layer carries a representative offline sample so the viewer never blocks and never crashes, and upgrades to the live object when that object is deployed. A representative sample is a coarsened or derived snapshot, and for several layers it is an authored analytic field rather than a reduction of upstream data. The bundled asset says which it is in its own metadata.
So a provider name in the table above tells you where the live signal comes from. It does not tell you that the bytes a given reader is looking at right now came from that provider. The viewer labels a fallback when it uses one, and limits covers what that means for the score.
Identity is not in the data
Every layer is aggregated and identity free by design. No callsigns, vessel names, tail numbers, registrations or personal identifiers, and no precise sensitive locations. Identity suppression is a property of the data rather than a display-time filter, which means it cannot be turned off by a client. Nonprofit points are aggregated to coarse postal areas and name organizations, never people.