UP42 Announces Thetaspace as Copernicus Masters Challenge Winner with Solution to Cloud Cover in EO Data

The Gilching, Germany, firm has won EUR 100,000 in awards for development of its CloudlessEO algorithm which leverages Deep Learning (DL) to fill data gaps created by clouds and their shadows in optical satellite imagery.

“We congratulate Thetaspace for devising a practical solution to the cloud-cover problem that has plagued optical satellite imaging since its inception,” said UP42 CEO Sean Wiid. “The CloudlessEO algorithm will help address environmental and sustainability issues in frequently cloudy areas, such as equatorial regions where deforestation is a major concern.”

Now in a beta version, the CloudlessEO algorithm uses a DL technology called Generative Adversarial Networks (GANs) to extract geometric and spectral information from Synthetic Aperture Radar (SAR) satellite data. This process generates highly accurate synthetic optical data to replace pixels obscured by clouds or shadows in the optical satellite image.

The CloudlessEO algorithm will dramatically enhance the cost-effectiveness of satellite imaging by increasing the usability of imagery once considered too cloud covered for practical use. The Thetaspace solution will also save significant time for geospatial end users who may spend days or weeks waiting for a cloud-free image to be acquired over their area of interest.

“We are excited about the many opportunities the UP42 Copernicus Masters Challenge offered us,” said Zayd Mahmoud Hamdi, CEO of Thetaspace. “Not only did it validate our original concept, but it also allowed us to deploy a SAAS solution that can help solve critical issues such as wildfires, floods, deforestation and other severe phenomena that continue to affect our planet.”