Satlas

Satlas is an image analysis tool. Satlas from Ai2 is an open geospatial platform that combines Sentinel-2 satellite imagery with AI annotations to show monthly changes in marine infrastructure, renewable energy sites and tree cover, plus a super resolution tool for sharper global imagery.

Satlas is an open geospatial platform from Ai2 that publishes monthly, AI annotated satellite data showing how the Earth's surface is changing. Three data themes anchor the project: marine infrastructure (offshore wind turbines and offshore platforms), renewable energy infrastructure (solar farms and onshore wind turbines), and tree cover. Rather than selling imagery, Satlas pairs Sentinel-2 satellite captures with machine learning annotations that identify and label these features, then presents the result as a browsable, time indexed map. The map experience is built around comparison. A date picker lets you set two points in time, and the layout can switch between a single map, a side by side view, or a change over time animation. Layer toggles control which infrastructure types and which tree cover density bands are drawn, and a heatmap mode emphasises concentration. It is the kind of interface that makes a decade of land use change legible in a couple of clicks, which is why the showcase examples lean on stories: the rapid build out of Chinese offshore wind, a new wind farm near Denmark, Amazon deforestation, commercial logging in Washington State, oil platforms in the Gulf of Mexico, the Sinop dam in Brazil, and solar development in Virginia and Brazil. A separate super resolution tool takes low resolution Sentinel-2 captures and runs AI models over them to produce higher resolution imagery at global scale, with before and after examples in places such as Nakuru in Kenya, Sydney, New Orleans and Los Angeles. That is useful when the raw pixel size of Sentinel-2 is too coarse to read built structures. The audience is researchers, journalists, policy analysts, students, and GIS practitioners who need evidence of physical change without building their own detection pipeline. It fits alongside commercial Earth observation platforms and raw imagery services: Satlas gives away the annotations and the exploration layer, and its data is published openly rather than held behind a licence.