Product Overview: Satlas
Introduction
Satlas is an innovative software platform developed by the Allen Institute for AI (AI2) that leverages artificial intelligence (AI) and satellite imagery to generate and provide high-accuracy global geospatial data products. This platform is designed to revolutionize various fields such as agriculture, urban planning, and disaster relief by offering detailed and up-to-date geospatial information.
Key Features
Data Sources and Analysis
Satlas utilizes images from the European Space Agency’s Sentinel-2 satellites. These images are analyzed using deep-learning computer models to identify and map various features, including offshore platforms, solar farms, onshore wind turbines, and tree cover. The models have been extensively trained, with manual labeling of thousands of data points, such as 36,000 wind turbines, 4,000 solar farms, and 7,000 offshore platforms.
Geospatial Data Products
The platform offers a range of geospatial data products that are updated monthly since January 2016. These products include:
- Positions of offshore wind turbines and platforms
- Locations of solar farms and onshore wind turbines
- Percentage of tree canopy coverage for 100-square-meter plots of land
- Future expansions are planned to include mapping tools for urban land use, agricultural crop types, and land cover.
Visualization and Download
Satlas allows users to visualize these geospatial data products on a zoomable online map, enabling the tracking of changes over time. All data displayed on the platform can be downloaded for offline analysis, facilitating further research and application in various geospatial domains.
Super-Resolution Capability
One of the notable features of Satlas is its ability to sharpen the resolution of satellite imagery using AI. The platform employs deep learning models to combine information from multiple low-resolution images captured at different times, resulting in high-resolution images with improved detail from 10 meters to 2.5 meters per pixel.
Community Engagement and Applications
Satlas is designed to support earth and environmental scientists, as well as organizations in various geospatial sectors. The platform’s data has been used in several sample applications, such as monitoring commercial logging impacts on tree cover and tracking investments in solar farms. Additionally, Satlas’ marine infrastructure data is being integrated into other AI2 projects, like Skylight, to improve vessel tracking and combat illegal fishing operations.
Model Accessibility
The AI models used in Satlas, including the SatlasPretrain models, are available on GitHub, allowing for transparency and community involvement in model development and optimization. These foundation models are trained on multiple image modalities and are designed to improve performance in downstream remote sensing applications.
Conclusion
Satlas is a powerful tool that harnesses the potential of AI and satellite imagery to provide accurate and actionable geospatial data. Its key features, including high-accuracy data products, super-resolution capabilities, and downloadable data, make it an invaluable resource for environmental monitoring, urban planning, and other geospatial applications. As the platform continues to expand its dataset and improve its models, it is poised to play a significant role in addressing global environmental and developmental challenges.