aliasfox's picture
Upload README.md with huggingface_hub
7c82532 verified
---
license: other
license_name: copernicus
license_link: https://copernicus.eu/user-license
task_categories:
- other
tags:
- elevation
- dem
- copernicus
- terrain
- geospatial
size_categories:
- 100K<n<1M
---
# Copernicus GLO-30 DEM Chunks
Global Digital Elevation Model (DEM) tiles from the Copernicus GLO-30 dataset,
pre-processed into 256x256 chunks in OZCHNK01 merged format for efficient
web serving.
## Source
Copernicus DEM GLO-30 (30-meter resolution) from the European Space Agency.
Original COG tiles downloaded from AWS Open Data:
https://copernicus-dem-30m.s3.eu-central-1.amazonaws.com/
## License
Copernicus Licence (free for most uses). See: https://copernicus.eu/user-license
## Format
Binary merged format (OZCHNK01):
- **Magic**: 8 bytes "OZCHNK01"
- **Version**: 2 (Copernicus = Float32)
- **Grid**: Variable ROWS x COLS (depends on tile dimensions)
- **Index**: Array of [4-byte offset, 4-byte size] per chunk
- **Data**: Concatenated zlib-compressed 256x256 Float32 chunks
### Tile Grid
- Internal COG tiles: 1024x1024 pixels
- Each split into 4x4 = 16 chunks of 256x256
- Edge tiles may have fewer chunks
- Data type: Float32 (4 bytes/pixel)
### File Naming
Files are organized by latitude band:
- Path: `{lat_dir}/{basename}.merged`
- Example: `N00/COPDEM_N00_00_E006_00_DEM.merged`
## Coverage
Global land areas at 30m resolution (~90GB total, ~26,450 tiles).
## Usage
For use with OpenZenith elevation API. Each merged file contains all 256x256
chunks for one Copernicus tile, allowing efficient range requests to fetch
individual chunks.
To read a chunk:
1. Open the merged file
2. Read the 12-byte header (magic + version + rows + cols)
3. Read the index: rows*cols entries of [offset, size]
4. Seek to the chunk offset, read size bytes
5. Decompress with zlib to get 256x256 Float32 pixel data
6. Elevation value = float32 at row*256 + col