Open data, checked by hand
Open data you can read directly from a public host — Zenodo, Hugging Face, an S3 bucket, a SPARQL endpoint or a plain GitHub repository. Forest and protected-area layers to work with, Spain's protected areas in full, plus the sources behind the historical atlas tree. Every field comes from the dataset's own record, and where a record does not say something, the card says so instead of guessing.
S2B_34UFD_20240828_0_L2A, 0.05% cloud, over Białowieża. © European Union, Copernicus Sentinel-2AWS S3
The Sentinel-2 Level-2A archive rewritten as Cloud-Optimized GeoTIFFs, with a STAC API in front of it. Atmospherically corrected surface reflectance, so you can pull a single band over one forest instead of a whole granule.
s3://sentinel-cogs
us-west-2 · open bucket, no requester-pays
AWS S3
Eleven land-cover classes for the whole planet at 10 m, built from Sentinel-1 radar and Sentinel-2 optical data. The tree-cover class is the quickest honest forest mask there is.
s3://esa-worldcover/v200/2021/map/
eu-central-1 · e.g. ESA_WorldCover_10m_2021_v200_N00E006_Map.tif
Attribution required when mapped: “© ESA WorldCover project 2020 / Contains modified Copernicus Sentinel data (2020) processed by ESA WorldCover consortium”.
AWS S3
Above-ground biomass density in Mg/ha for every GEDI laser footprint, with the prediction error for each one. Sampled shots along the ISS ground track rather than a wall-to-wall raster, which is exactly why it pairs well with a drawn area.
s3://ornl-cumulus-prod-protected/gedi/GEDI_L4A_AGB_Density_V2_1/data
us-west-2 · protected bucket, needs a free NASA Earthdata login
Hugging Face
Monospecific forest patches labelled by species, each with both very-high-resolution aerial imagery and a dense Lidar point cloud over the same ground. A rare case of a species benchmark where the two modalities actually line up.
hf://datasets/IGNF/PureForest
Also loadable with datasets.load_dataset("IGNF/PureForest")
Hugging Face
Tree species labels for 60 m plots in Lower Saxony, each carrying a full year of Sentinel-1 and Sentinel-2 observations plus a 0.2 m aerial patch. Built for models that need to watch a stand change through a season.
hf://datasets/IGNF/TreeSatAI-Time-Series
Licence: repo metadata says CC BY 4.0, the card text says CC BY-SA 4.0 — check before redistributing
Zenodo
A single global canopy height layer for 2020, fused from GEDI and ICESat-2 returns with Sentinel-1 and Sentinel-2 imagery in Earth Engine. Heights are stored in centimetres, so remember to divide.
10.5281/zenodo.7643403
Nie, Zhu, Wang & Xi, 18 February 2023
Zenodo
Nearly four decades of European forest disturbance from the Landsat archive: when each patch was hit, how hard, how many times, and whether the cause was wind and bark beetle, fire or harvest. Country by country, so you can take just the one you work in.
10.5281/zenodo.13333034
Viana-Soto & Senf, TU Munich, 20 August 2024
EEA
The boundaries of Europe’s protected-area network: every Special Area of Conservation and Special Protection Area, with the habitats and species each site was designated for. The layer to clip against when the question is “is this inside a protected site?”
https://sdi.eea.europa.eu/data/91357f39-7866-41ce-b447-43905c364ec8
Compiled by the European Commission, published through the EEA
world_1914.geojson, with nothing underneath them. aourednik/historical-basemaps, GPL-3.0GitHub
Georeferenced boundaries of world countries and cultural regions, one GeoJSON per year, made for putting historical data on a map. This is the layer the atlas draws when you scrub the timeline: the 1914 world, then the 1938 one.
github.com/aourednik/historical-basemaps
The atlas embeds 18 of these snapshots. Author's own caveat: this is work in progress, historical boundaries are more disputed than modern ones, and you should check the maps against other sources before using them in academic work.
Base layer of the historical atlas
SPARQL
The free knowledge base, queried live rather than downloaded. Ask for every battle, castle, monastery or observatory with coordinates and a date, and you get a themed layer back — which is exactly how the atlas builds most of what you can switch on.
https://query.wikidata.org/sparql
Point-in-time events keep one year; things that persist get a [start, end] lifespan, which is what lets a layer be filtered to the year under the playhead.
Conflict, disaster, built-heritage and institution layers in the historical atlas
start_date, pulled through Overpass. The clusters are where the project has been mapped so far, not where history happened. © OpenHistoricalMap contributors, CC BY-SAOverpass API
OpenStreetMap with a time dimension: the same editing model and tags, plus start_date and end_date, so a feature can appear and disappear as you move through history. Real geometry, not dots — historical boundaries come back as polygons and routes as lines.
https://overpass-turbo.openhistoricalmap.org
Planet dumps at planet.openhistoricalmap.org if you would rather hold the whole thing than query it.
Dated OSM features in the historical atlas
Direct download
The community gazetteer of the ancient world: settlements, roads, temples and regions from classical antiquity, each with a stable URI, a location and a date range. The reference layer for anything happening before the map had countries.
https://pleiades.stoa.org/downloads
Take the numbered quarterly release if you need a citable version; the daily exports move under you.
Ancient-places layer in the historical atlas
SPARQL
Linked open data for numismatics, pooled from museums and collections across the world: coins, hoards, dies and monograms, tied to the mints and places they came from. Plot coin types at their mints and you get trade and power drawn on a map, deep into BCE.
https://nomisma.org/sparql
There is no single licence for the whole of Nomisma. Contributing institutions keep their own terms, which run from ODbL through CC BY and CC BY-SA to CC BY-NC-SA — check the dataset you actually query.
≈10 k coin types at their mints in the historical atlas
SPARQL
Wikipedia's infoboxes turned into a queryable graph. Useful as a second opinion: ask a second endpoint the same GeoSPARQL question you asked the first, and see whether the two agree on where and when something happened.
https://dbpedia.org/sparql
The public endpoint does not carry every DBpedia dataset — for the rest, pull from the Databus.
Conflicts and power stations in the historical atlas
SPARQL
A Wikibase for historical research, run by historians rather than by a general-purpose project. Same technology as Wikidata, its own property space, and records that go deeper into archival detail than a general encyclopedia would ever carry.
https://database.factgrid.de/query/
Because the property space is its own, queries written for Wikidata do not transfer unchanged.
An independent history graph in the historical atlas
REST API
The authoritative place-name map of Antarctica, pooling the gazetteers of every nation that names features there — so one mountain can carry several official names at once, each with its own country and citation.
https://api.bas.ac.uk/locations/v1/gazetteer/cga
Names outnumber features because the same feature is named independently by different countries.
Part of the Antarctica set in the historical atlas
Direct download
Every protected area declared in Spain, in one file: national parks, natural parks, nature reserves, natural monuments, protected landscapes and marine protected areas. Ley 42/2007 defines five figures at state level; the autonomous communities add more than forty of their own, and all of them are in here.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=enp.zip
A zipped shapefile, so it is a conversion away from the canvas: ogr2ogr -f Parquet enp.parquet enp.shp gives you something the map reads in place. The yearly review is the point — boundaries move, new declarations land, and the file says which year it speaks for.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
Spain's own cut of the Natura 2000 network: the sites of community importance and special areas of conservation from the Habitats Directive, plus the special protection areas for birds. Spain contributes more of this network than any other member state, so this is the file behind a great deal of European conservation mapping.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=rn2000.zip
The European card above covers the whole EU from the EEA. Take this one when you want what Spain itself submitted, on Spain's own update cycle, with the national codes on each site.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
The national wetland register: the marshes, lagoons, deltas and peatlands the state tracks so their condition can be followed over time and written into river basin plans. Small file, and the one to clip against when the question is whether something drained.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=iezh.zip
At 4 MB this is the easiest of the Spanish set to convert and keep on the canvas permanently.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
The Spanish wetlands listed under the Ramsar Convention as internationally important: marshes, estuaries, deltas and lagoons, each one a commitment the state made in a treaty rather than a line drawn at home.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=humedalesramsar.zip
Overlaps the national wetland inventory rather than duplicating it: Ramsar is the international designation, the IEZH is the domestic register.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
The terrestrial and coastal areas UNESCO recognises under Man and the Biosphere. Spain holds one of the largest networks in the world, and unlike a park boundary these are zoned for people living and working inside them.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=reservasbio.zip
Worth pairing with the protected-areas file: a biosphere reserve usually contains several of them, and the interesting question is what sits in the buffer between.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
The drovers' roads: the public network of cañadas, cordeles and veredas that moved livestock across Spain for centuries and is still, legally, public land. Read as ecology rather than history it is a corridor network — long, thin, continuous, and connecting habitats that roads and fields have otherwise cut apart.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=VPecuarias.zip
Component 5f of the Inventario Español del Patrimonio Natural y la Biodiversidad. Scales vary because the source is each region's own classification work.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
The biogeographic regions Spain is divided into for the Habitats Directive: Atlantic, Mediterranean, Alpine and Macaronesian. The layer that explains why the same habitat code means something different in Galicia and in Almería, and the one conservation status is assessed within.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=regionesbio.zip
Small and slow-changing, which makes it a good permanent backdrop to clip other Spanish layers against.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
Direct download
The Important Bird and Biodiversity Areas identified in Spain: sites that matter for bird populations on ornithological criteria rather than administrative ones. They are not a legal designation, which is exactly why they are useful — the gap between an IBA and the nearest ZEPA is a question worth asking.
https://gis.miteco.gob.es/descargas/app/DescargaFichero?f=iba.zip
Published through the Banco de Datos de la Naturaleza; the underlying inventory is SEO/BirdLife's work, so cite both when you use it.
Attribution required when mapped: “Fuente: © Ministerio para la Transición Ecológica y el Reto Demográfico”.
WMS service
Spain's national base cartography, served as a Web Map Service. Paste the URL with a layers= parameter and the canvas tiles it directly — no download, no conversion, and no proxy: the server renders each tile and MapLibre asks for them as you pan.
https://www.ign.es/wms-inspire/ign-base?service=WMS&layers=IGNBaseTodo
A WMS gives you pixels, not features — you can see it and place things against it, but you cannot query or analyse what is in it. That is the trade for reaching data nobody has converted.
Attribution required when mapped: “Instituto Geográfico Nacional de España”.
WMS service
The whole OSM base map, rendered by terrestris and served as WMS. Useful as a backdrop where the built-in basemaps are too dark or too sparse, and as the simplest thing to test a WMS paste against — it answers in Web Mercator with open CORS, which is what makes a service usable from a browser at all.
https://ows.terrestris.de/osm/service?service=WMS&layers=OSM-WMS
Both of these were pasted into the canvas and tiled before they were written down here, which is the only way to know a service answers a browser rather than only a desktop GIS.
Nothing matches that.
Two of these look at the same forest. The Sentinel-2 quicklook and the Natura 2000 card both show Białowieża, one of the last large lowland primary forests in Europe, which is a decent test of whether a dataset can tell old growth from plantation.
Get early access to the catalog