ESA WorldCover¶

ESA WorldCover 10 m 2020 v100 / 2021 v200 (global 10 m land cover maps from Sentinel-1 and Sentinel-2, 11 classes)

Discontinued

At a glance¶

Note on period: Two maps: 2020 (v100, released October 2021) and 2021 (v200, released 28 October 2022). They were made with different algorithm versions, so their difference mixes real change with algorithm effects and they should not be treated as a change-detection time series.

Overview¶

The first global land cover maps at 10 m resolution, produced from Sentinel-1 radar and Sentinel-2 optical imagery by a VITO-led consortium funded by ESA. They give 11 classes — tree cover, shrubland, grassland, cropland, built-up, bare/sparse vegetation, snow and ice, permanent water bodies, herbaceous wetland, mangroves, and moss and lichen — for 2020 and 2021, with independently validated overall accuracies of 74.4% and 76.7%. The detail resolves urban, cropland and river boundaries that 300 m and 500 m products miss, which makes WorldCover popular for regional land surface model inputs, ecology and hydrology studies, and as a reference for checking other maps.

What changes with the successor¶

WorldCover 2020 v100 → WorldCover 2021 v200¶

Transition: 2022-10

The 2021 map was produced from scratch with an improved algorithm (v200); the 2020 map was not reprocessed with v200. Format and legend are the same, but the two maps are different versions.

Sources: ESA WorldCover Product User Manual v2.0 · ESA WorldCover — Data access (version notes)

ESA WorldCover (2020–2021) → Copernicus LCFM 10 m Global Land Cover (2020–2026)¶

Transition: 2026-02

The ESA demonstration project has ended, and 10 m global land cover is now an operational product of the EU Copernicus Land Monitoring Service (CLMS) LCFM project, planned yearly for 2020–2026. As of 2026-10-13 only the 2020 LCFM 10 m map has been released.

Sources: ESA WorldCover news (2026-02-25, Copernicus LCFM operational successor) · CLMS — LCFM: a new chapter in global land cover monitoring (2025-06-26) · CDSE forum — answer on LCM 10 m availability from 2021 onward (October 2025)

Available variables¶

Based on the WorldCover Product User Manual (PUM v2.0). Files are Cloud Optimized GeoTIFFs named ESA_WorldCover_10m__.tif (e.g. ESA_WorldCover_10m_2021_v200_N36E126_Map.tif). Map and InputQuality below are the layer names at the end of the file name; InputQuality is a three-band GeoTIFF, so each row's band number is given in its note.

instantaneous mean accumulated maximum minimum static — instantaneous: value at the time stamp · mean/accumulated/maximum/minimum: over the indicated interval · static: does not change with time

Land cover map (10 m, classification for the year)¶

Input quality (60 m, 3 bands)¶

The full variable list is in the official documentation.

How to get the data¶

• Public AWS S3 bucket (esa-worldcover) — S3 / HTTPS (no login)
  • Tiles are under s3://esa-worldcover/v100/2020/map/ and s3://esa-worldcover/v200/2021/map/ (eu-central-1); use --no-sign-request. The HTTPS root is https://esa-worldcover.s3.eu-central-1.amazonaws.com/... and the tile list is in esa_worldcover_grid.geojson. Only the Map layer is offered here.
• WorldCover viewer / Terrascope (VITO) — Web viewer download / terracatalogueclient (Python)
  • Requires free Terrascope registration. The InputQuality layer is available only through this route (viewer or Terrascope).
• Zenodo (2020 v100, 2021 v200) — HTTPS download (zipped 60°×60° macrotiles)
  • The 2020 v100 record is https://zenodo.org/record/5571936.
• Google Earth Engine — Earth Engine assets ESA/WorldCover/v100 and ESA/WorldCover/v200
  • Convenient for aggregation and visualization over large areas.

Example code¶

Reads the Cloud Optimized GeoTIFF directly from AWS with rasterio, without logging in, and counts class shares around Seoul. Tile names give the lower-left corner; tile N36E126 covers 36–39°N, 126–129°E.

# pip install rasterio numpy
import numpy as np
import rasterio
from rasterio.windows import from_bounds

year, version, tile = 2021, "v200", "N36E126"   # use "v100" for 2020
url = (
    "https://esa-worldcover.s3.eu-central-1.amazonaws.com/"
    f"{version}/{year}/map/ESA_WorldCover_10m_{year}_{version}_{tile}_Map.tif"
)
with rasterio.open(url) as src:
    win = from_bounds(126.8, 37.4, 127.2, 37.7, src.transform)   # around Seoul (W, S, E, N)
    lc = src.read(1, window=win)

names = {10: "Tree cover", 20: "Shrubland", 30: "Grassland", 40: "Cropland", 50: "Built-up",
         60: "Bare/sparse", 70: "Snow and ice", 80: "Water", 90: "Herbaceous wetland",
         95: "Mangroves", 100: "Moss and lichen"}
vals, counts = np.unique(lc, return_counts=True)
for v, c in zip(vals, counts):
    print(names.get(int(v), v), f"{100 * c / lc.size:.1f}%")

Practical tips¶

• Do not difference the 2020 (v100) and 2021 (v200) maps to estimate land cover change. The official manual says most of the difference is due to the algorithm change. Pick one year, or use another dataset for change studies.
• When moving the 10 m classes to a coarser grid, use mode aggregation or per-class fractions; averaging or bilinear interpolation produces meaningless codes.
• On a latitude–longitude grid the east–west width of a pixel shrinks toward the poles. To compute areas, weight pixels by their latitude-dependent area rather than counting them.
• Known issues: cloud artefacts in persistently cloudy regions, straight-line seams along Sentinel-2 orbit or processing-block borders, glacier ablation zones and mountain shadows classified as water, and confusion between irrigated cropland and herbaceous wetland.
• Downloads from AWS, Zenodo and ESA contain only the Map layer; the InputQuality layer with observation counts is available only from the WorldCover viewer or Terrascope.
• The successor Copernicus LCFM 10 m map is a separate product. There is no official statement that it shares the WorldCover legend or algorithm, so check legends and definitions before joining the two.

Notices and change history¶

References for your paper¶

Dataset citation¶

WorldCover 2021 v200 — DOI 10.5281/zenodo.7254221

@misc{zanaga2022worldcover,
  author    = {Zanaga, Daniele and Van De Kerchove, Ruben and Daems, Dirk and De Keersmaecker, Wanda and Brockmann, Carsten and Kirches, Grit and Wevers, Jan and Cartus, Oliver and Santoro, Maurizio and Fritz, Steffen and Lesiv, Myroslava and Herold, Martin and Tsendbazar, Nandin-Erdene and Xu, Panpan and Ramoino, Fabrizio and Arino, Olivier},
  title     = {{ESA WorldCover} 10 m 2021 v200},
  year      = {2022},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.7254221}
}

WorldCover 2020 v100 — DOI 10.5281/zenodo.5571936

@misc{zanaga2021worldcover,
  author    = {Zanaga, Daniele and Van De Kerchove, Ruben and De Keersmaecker, Wanda and Souverijns, Niels and Brockmann, Carsten and Quast, Ralf and Wevers, Jan and Grosu, Alex and Paccini, Audrey and Vergnaud, Sylvain and Cartus, Oliver and Santoro, Maurizio and Fritz, Steffen and Georgieva, Ivelina and Lesiv, Myroslava and Carter, Sarah and Herold, Martin and Li, Linlin and Tsendbazar, Nandin-Erdene and Ramoino, Fabrizio and Arino, Olivier},
  title     = {{ESA WorldCover} 10 m 2020 v100},
  year      = {2021},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.5571936}
}

Reference papers¶

WorldCover Product User Manual v2.0 — no DOI

@techreport{worldcover2022pum,
  author      = {{ESA WorldCover consortium}},
  title       = {{ESA WorldCover} Product User Manual, Version 2.0},
  institution = {VITO},
  number      = {WorldCover\_PUM\_v2.0},
  year        = {2022},
  url         = {https://esa-worldcover.s3.eu-central-1.amazonaws.com/v200/2021/docs/WorldCover_PUM_V2.0.pdf}
}

Acknowledgements¶

Cite the Zenodo DOI of the year you used (the official website says a reference publication is in preparation). On maps and figures, credit the source with the matching year, e.g. "© ESA WorldCover project 2021 / Contains modified Copernicus Sentinel data (2021) processed by ESA WorldCover consortium".

Information on the datasets used in this study was compiled with the help of the Korean Climate Data Navigator (KCDN) (Sim, 2026; https://doi.org/10.5281/zenodo.23300923).

Links¶

• ESA WorldCover website
• WorldCover data access and citation
• Product User Manual v2.0 (2021 v200)
• WorldCover viewer
• Copernicus CLMS — Global Dynamic Land Cover (including LCFM)