# Global Human Settlement Layer

Raster maps of the whole world: how much is built up and how many people live there, per 100-metre or 1-kilometre cell, for 1975, 1980 and so on up to 2030. Made from satellite images and censuses, the same way for every country.

- Publisher: Joint Research Centre (Europese Commissie), Copernicus
- Area: World · Finest detail: grid cell (100 metres)
- Period: 1975 – 2030 · Refreshed: yearly
- Access: Open · Licence: CC BY 4.0
- Join keys: rastercel (Mollweide, EPSG:54009), landcode (GADM, NLD)
- Status: ok (2026-08-24) · nieuwste map: EMC_BUILT_GLOBE_R2025A · 55 GHS-producten
- URL: https://human-settlement.emergency.copernicus.eu

## What can you do with it?
- **Fifty years of building in the Netherlands.** According to the GHSL country statistics (2024 release), built-up area in the Netherlands grew from 955 km² in 1975 to 1,906 km² in 2020: twice as much. Population grew from 13.7 to 17.4 million over the same period.
- **More space per resident.** Divide built-up area by population. In the Netherlands that was 70 m² of built-up area per resident in 1975 and 109 m² in 2020 (own calculation from the country statistics). Compare with Belgium and Germany.
- **Where did the Randstad grow?.** Overlay the 1975 and 2020 grids and colour the cells that became built up. You will see where new towns and districts appeared, such as Almere (founded in 1976).
- **Urban or rural, by European rules.** The GHSL classifies every cell by the Degree of Urbanisation: urban centre, urban cluster or rural. That lets you compare urbanisation fairly between countries, regardless of municipal borders.

## How to process it
1. **Start with the country statistics.** A small zip file (1.5 MB) with an Excel table: area, population and built-up surface per country and class (UC = urban centre, UCL = urban cluster, RUR = rural), for 1975 to 2030. Add up the three classes for the national total.
   `https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/GHSL/GHS_COUNTRY_STATS_MT_GLOBE_R2024A/V1-0/GHS_COUNTRY_STATS_MT_GLOBE_R2024A.zip`
2. **Choose product, year and resolution.** Each folder in the download directory is one product and release, e.g. GHS_POP_GLOBE_R2023A (population) or GHS_BUILT_S_GLOBE_R2023A (built-up surface). Below that, one folder per year (E1975 … E2030) and resolution (54009_100 = 100 metres in Mollweide, 4326_3ss = 3 arc seconds).
   `https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/GHSL/GHS_POP_GLOBE_R2023A/`
3. **Fetch only the tiles for the Netherlands.** The global population file for 2020 at 100 metres is 4.7 GB. The tiles/ folder holds the same map in tiles of a few tens of MB. Use the download page to find which tiles cover the Netherlands.
   `https://human-settlement.emergency.copernicus.eu/download.php`
4. **Convert to your own area.** The grids are in Mollweide (EPSG:54009). Reproject to RD (EPSG:28992) or sum cells per municipality with zonal statistics in QGIS or rasterstats (Python).

## Interesting combinations
- Does the model match the real counts? → [cbs-kwb](https://www.opendataindex.nl/en/bron/cbs-kwb.md) via rastercellen opgeteld per buurt of gemeente. Sum GHSL population per municipality and compare it with the CBS population count. Where does the model deviate most?
- Does the satellite see what was built? → [pdok-bag](https://www.opendataindex.nl/en/bron/pdok-bag.md) via locatie (bouwjaar per pand). Compare GHSL growth between 2000 and 2020 with the buildings that the BAG says were built in those years.

## Similar sources
- [esa-worldcover](https://www.opendataindex.nl/en/bron/esa-worldcover.md): Land cover at 10 metres for 2020 and 2021, without population and without a series back in time.
- [Vierkantstatistieken (CBS)](https://www.cbs.nl/nl-nl/dossier/nederland-regionaal/geografische-data/kaart-van-100-meter-bij-100-meter-met-statistieken): Actual counts per 100 metres, only for the Netherlands. Since 2015 only squares with at least 5 residents or 5 dwellings.
- [WorldPop](https://www.worldpop.org): Also population per grid cell worldwide, with a different model. Useful to put two estimates side by side.

## What to watch out for
- **A model, not a count.** Population per cell is estimated from satellite images and censuses. For the Netherlands itself, CBS grid statistics are more precise.
- **2025 and 2030 are projections.** The R2023A release (and the R2024A country statistics) was made in 2023 and 2024. The years 2025 and 2030 are therefore projections, not observations.
- **Do not mix releases.** Each release (R2019A, R2022A, R2023A …) recalculates the whole series. Only compare years within one release.
- **Large files, unusual projection.** Global files run to gigabytes and use Mollweide. Work with tiles and reproject before joining with Dutch maps.

## API (Bestanden (open downloadmap met HTTP-index; GeoTIFF in zip, Excel in zip))

- Base: https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/GHSL
- Access: No key or account needed (tested)
- Formats: zip, geotiff, xlsx
- Size: 59 folders in the download directory (product releases plus test and archive folders), 16 Oct 2026 (tested)
- Filtering: No query language: you pick product, release, year and resolution in the path, and a tile for your area. You do the computation (zonal statistics) yourself.; 

### Endpoints
- `GET /?C=M;O=D`: All product folders, newest first (Apache index; also C=N name and C=S size). Newest on 16 Oct 2026: open_urban_typologies, COG_test, EMC_BUILT_GLOBE_R2025A.
- `GET /{PRODUCT}_GLOBE_{uitgave}/`: Per product and release one folder per year and resolution, e.g. GHS_POP_E2020_GLOBE_R2023A_54009_100 (population 2020, Mollweide, 100 m).
- `GET /{…}/V1-0/{naam}_V1_0.zip`: The global file as a zip with a GeoTIFF (population 2020 at 100 m: 4.7 GB).
- `GET /{…}/V1-0/tiles/{naam}_V1_0_R{rij}_C{kolom}.zip`: The same map in 376 tiles (population 2020, 100 m), from 1 to ~90 MB per tile.
- `GET /GHS_COUNTRY_STATS_MT_GLOBE_R2024A/V1-0/GHS_COUNTRY_STATS_MT_GLOBE_R2024A.zip`: Country statistics 1975–2030 per country and DegUrba class (UC, UCL, RUR) as Excel in a zip: 1.5 MB, Last-Modified 12 Nov 2024.
- `GET /GHS_WUP_COUNTRY_STATS_GLOBE_R2025A/V1-0/GHS_WUP_COUNTRY_STATS_GLOBE_R2025A_V1_0.zip`: Newer country statistics from the WUP series R2025A (5.9 MB, 4 Dec 2025); not opened by us.

### Parameters
- `product`: GHS_POP (population), GHS_BUILT_S (built-up surface), GHS_BUILT_V (volume), GHS_BUILT_H (height), GHS_SMOD (urban/rural, DegUrba), GHS_UCDB (urban centre database), GHS_DUC, GHS_LAND, GHS_AGE, GHS_OBAT … (tested)
- `uitgave`: R2022A, R2023A, R2024A, R2025A …: each release recalculates the whole series. (experience)
- `jaar`: Every 5 years; 2025 and 2030 are projections. (tested)
- `resolutie`: 54009_100 and 54009_1000 (Mollweide, 100 m and 1 km), 4326_3ss and 4326_30ss (WGS84, 3 and 30 arc seconds). (tested)

### What the files contain (selection) (4)
`GHS_POP` raster float, `GHS_BUILT_S` raster, `GHS_SMOD` raster klasse, `AREA_KM2_L1, POP_L1, BU_S_km2_L1, BUCAP_m2_L1` werkblad

### Limits and pitfalls
- No key and no documented limit. The server supports Range requests (accept-ranges: bytes), useful to resume a download. (tested)
- The COG_test folder holds a trial with a 20 GB Cloud-Optimized GeoTIFF (Oct 2026); the regular products are still zips with a GeoTIFF. So you can only inspect a raster after downloading and unzipping. (tested)
- The country statistics zip is an xlsx inside a zip; readable without an Excel package via the zip's central directory and the sheet XML (sharedStrings for t='s'). (experience)
- No robots.txt on jeodpp.jrc.ec.europa.eu and human-settlement.emergency.copernicus.eu (404). (tested)

### Example
```
https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/GHSL/GHS_COUNTRY_STATS_MT_GLOBE_R2024A/V1-0/
```


## Examples with real figures (2)

For each example the finding, the reference date and the call that fetches exactly these figures.

- **Population of the Netherlands by type of area: in urban centres is the largest part, 49% in 2020** (2020). Table GHS-COUNTRY-STATS_MT_GLOBE_R2024_V1_0.xlsx (POP_L1). [Example](https://www.opendataindex.nl/en/voorbeeld/ghsl-inwoners-stad-platteland)
  - `https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/GHSL/GHS_COUNTRY_STATS_MT_GLOBE_R2024A/V1-0/GHS_COUNTRY_STATS_MT_GLOBE_R2024A.zip`
- **Built-up surface in the Netherlands from satellite imagery: 1,906 square kilometres in 2020, 2 times as much as in 1975** (2020). Table GHS-COUNTRY-STATS_MT_GLOBE_R2024_V1_0.xlsx (BU_S_km2_L1). [Example](https://www.opendataindex.nl/en/voorbeeld/ghsl-bebouwd-oppervlak)
  - `https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/GHSL/GHS_COUNTRY_STATS_MT_GLOBE_R2024A/V1-0/GHS_COUNTRY_STATS_MT_GLOBE_R2024A.zip`