TerraMesh is a planetary-scale, multimodal dataset for Earth-Observation foundation models. It merges data from Sentinel-1 SAR, Sentinel-2 optical, Copernicus DEM, NDVI, and land-cover sources into more than 9 million co-registered patches. The dataset was created by ibm-esa-geospatial and was last updated on March 24, -2026.
Use Cases
- Training multimodal Earth-Observation foundation models based on co-registered SAR, optical, DEM, and land-cover data.
- Developing algorithms for planetary-scale analysis based on the dataset's global coverage.
- Conducting representation learning experiments based on the more than 9 million prepared patches.
- Benchmarking geospatial AI models based on the analysis-ready, co-registered data structure.
Strengths
- Contains more than 9 million co-registered data patches.
- Integrates data from multiple sources: Sentinel-1 SAR, Sentinel-2 optical, Copernicus DEM, NDVI, and land-cover.
- Designed as analysis-ready for large-scale representation learning.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Last updated 2026-03-24 07:08:07; freshness should be verified.
Provenance
- Source
- ibm-esa-geospatial
- Collection Method
- Merges data from Sentinel-1 SAR, Sentinel-2 optical, Copernicus DEM, NDVI, and land-cover sources.
- Time Range
- null
- Freshness
- Last updated 2026-03-24 07:08:07.
- Geography
- Planetary-scale