Sentinel-2 Cloud-Optimized GeoTIFF Public Dataset

Abstract:

It's been 5 years since the launch of Sentinel-2A, and over 3 since the launch of Sentinel-2B and in that time these Earth-observing satellites have proven to be an extremely valuable addition to other open data sources such as Landsat and MODIS. Several bands in Sentinel-2 are higher resolution that Landsat and although it lacks the long wave IR bands it has 3 "vegetation" bands - narrow Near IR bands for applications in agriculture and forestry. However, like many open datasets, the distribution format does not lend itself to scaling processing in the cloud and there has been a lack of analysis ready data available outside of Europe. Between a lack of easy discoverability for scenes of interest and files that must be downloaded, the data is not conducive for inclusion in cloud-optimized workflows that leverage programmatic discovery and access, along with optimized remote access.

Recently, Geoscience Australia and Element 84 have created an archive of the Sentinel-2 L2 (Analysis Ready Data) data covering Africa for the Digital Earth Africa project. The Sentinel-2 data has been converted to a cloud optimized format (Cloud-Optimized GeoTIFF), and a SpatioTemporal Asset Catalog (STAC) metadata created to index the scenes, allowing for easy discoverability of the data. The entire archive, and the API, is available for free as part of the AWS Public Datasets program. This has enabled indexing of the data into the Open Data Cube as part of the Digital Earth Africa platform to enable the creation and use of powerful analysis workflows.

Learn about how this archive was created, and kept up to date, along with how to leverage this data using Open-Source software such as sat-search, the Open Data Cube, or the Pangeo ecosystem.