Microsoft parks Landsat and Sentinel satellite data in Azure's orbit

NASA's Earth-watching archives find new home in Redmond's cloud, complete with Copilot hype

Microsoft has made NASA's Harmonized Landsat and Sentinel-2 (HLS) dataset available on Azure via the Windows giant's Planetary Computer platform.

It seems an excellent use for all that Azure capacity – petabytes of global environmental data can be accessed through APIs or directly via Azure storage. "This flexible scientific environment allows users to answer questions about the data, and both build applications and use applications on top of the platform," Microsoft said.

The HLS dataset is a vast archive of data from Earth-monitoring spacecraft, in this case NASA's Landsat 8 and 9 and the European Space Agency's Sentinel-2 satellites. It's an invaluable tool for researchers looking into climate change, changes in land use, agricultural applications, and so on.

Landsat 8 is getting a little long in the tooth, having been launched in 2013. It was joined in 2021 by Landsat 9. Whether a follow-up – dubbed Landsat Next – will ever happen depends on the final budget for NASA science. ESA has launched Sentinel-2A, B, and C so far. Sentinel-2C reached orbit in 2024 and Sentinel-2D will launch in the next few years. Combined with 2C, it will replace the aging Sentinel-2A and B spacecraft.

The HLS project was a major outcome of the 2016 Satellite Needs Working Group (SNWG). According to NASA, the Landsats can collect observations with 30-meter spatial resolution over a 16-day repeat period. The Sentinels provide 10 to 20-meter spatial resolution with a five-day repeat period. Combined through HLS, observations can be acquired at a 30-meter spatial resolution every two to three days, with the satellite data appearing as one collection.

NASA funds the HLS dataset. The agency is currently subject to the US government shutdown, and it is unclear whether that funding will continue until budget proposals, some of which include substantial cuts to science programs, are finalized. Microsoft did not respond to questions about whether it might step in to cover any gaps.

Considering the amount of compute devoted to AI applications, this seems a good use of Azure resources. However, Microsoft, being Microsoft, had to weigh in with some Copilot and AI suggestions for researchers.

The Windows giant suggested that perhaps researchers might like to use the Azure OpenAI Service "to create intelligent applications that enhance Earth observation analysis."

Or perhaps the prototype NASA Earth Copilot and natural language queries could be used to generate insights into the geospatial data.

Microsoft came up with a list of ways that HLS data and Azure AI could be used. These include automation of land classification and vegetation monitoring, deforestation trends, and the prediction of environmental patterns.

Oddly, as a subject for research, it did not mention the environmental impact of all the datacenters devoted to crunching data for AI applications. We can't imagine why. ®