A community data commons for equitable Earth Science

Samapriya Roy, Sayantan Majumdar , Tyson L. Swetnam

Earth observation has entered an era of extraordinary data abundance, yet the analysis-ready, well-documented, reproducible datasets that climate, water, agriculture, and disaster research actually require remain scarce. Every research group that repeats the same preprocessing pays a last-mile cost that falls hardest on the institutions and communities least able to afford it. Here we describe how a grassroots response to this gap, the Google Earth Engine (GEE) Community Catalog, has grown over six years of unfunded, volunteer operation into critical scientific infrastructure: 4,400 curated, cloud-ingested datasets totaling 585 TB across 16 thematic domains, serving nearly 4 million monthly data requests, with cumulative reach spanning 228 countries and territories and demand led by Brazil, the United States, and India. Drawing on this experience, we examine what a peer-produced data commons reveals about equity in global science, including the hosting of Indigenous territory data and datasets built by and for local communities, and about the sustainability paradox facing community infrastructure that the scientific enterprise depends upon but does not fund. We distill five transferable lessons for building community data commons: treat curation as scientific labor, design for findability first, build loops rather than pipelines, meet communities where they are and share stewardship, and plan succession from the start. As Earth observation confronts both accelerating environmental change and accelerating data growth, community data commons are not a convenience; they are essential infrastructure for equitable and anticipatory science.