OpenLandMap-soilDB 30m (OpenGeoHub)
Machine-learning-mapped global soil at 30m — finer-grained than SoilGrids.
What this is
OpenGeoHub's soilDB maps global soil properties from harmonized legacy samples using machine learning. The 2026 release reaches 30m and covers 2000–2022+. Available as cloud-optimized GeoTIFFs (COGs) through a SpatioTemporal Asset Catalog (STAC) at stac.openlandmap.org.
Why we surface it
SoilGrids runs at 250m. For contexts where that grain is too coarse, OpenLandMap soilDB is the next step, at 30m.
What you can do with it
Shown for commercial use. Personal, research, and nonprofit use is often more permissive — switch the verdict lens on the catalog.
- Internal / individual use
- Free to use for any purpose, including commercial and AI training. Nothing's triggered until you share it.
- Redistribute the data
- Fine to use and redistribute, including commercially. One string attached: credit the original source wherever it shows up.
- Republish a derivative
- Build on it and republish freely, commercial or not. You must credit the original source in what you ship.
- Build into a private product
- Fine to build into a closed or commercial product. Attribution is still owed wherever the data is actually exposed to users.
This decoder is a starting point, not legal advice. For commercial deployment, run the actual license terms past someone who reads them for a living.
Good to know
The CC-BY-4.0 license applies to the soilDB data product specifically. The broader OpenLandMap portal hosts layers from multiple sources, some carrying ODbL instead — check the license on the specific STAC record before use.
○ Planned
Not in the map yet — this source is on the V1 backlog. Schema, sample, and citation slots fill in once the layer ships.