World Resources Institute

The World Resources Institute (WRI) publishes free environmental data built for geographic information system (GIS) work, climate analysis, and sustainability research. Its best-known asset is Global Forest Watch, a near-real-time platform that tracks tree cover loss and gain across the world from satellite imagery, but the wider catalogue reaches across climate, fresh water, food, energy, and land use. Its companion platform, Resource Watch, gathers hundreds of curated datasets onto a single interactive map for exploration and comparison.

The data arrives in several shapes. You can browse interactive maps and dashboards, download GIS layers and CSV files, or pull data programmatically through APIs, and much of the geospatial material is also available through Google Earth Engine for large-scale analysis. A lot of it is inherently spatial, for instance forest cover by location and year, so for RAG you will usually summarise a region or a trend into natural-language statements a retriever can match, rather than trying to index raw raster tiles directly.

This suits any RAG project that answers questions about deforestation, emissions, water risk, or climate resilience and wants a credible, well-documented source behind each answer. The accompanying research reports, technical notes, and country briefings are themselves excellent grounding text, often more directly useful to a text pipeline than the underlying geospatial layers. WRI sits comfortably next to sources like NASA Earthdata and the World Bank: it is more curated and issue-focused, where those go deeper on raw observations and headline economic indicators.

On licensing you are in comfortable territory. Most WRI data is released under the Creative Commons Attribution licence (CC BY 4.0), so you may reuse it, including in commercial products, as long as you credit WRI. A few datasets carry different terms, so it is worth checking the licence on each one before you build, but the default position is genuinely open with attribution rather than anything restrictive.