Algorithm development and optimization

Cloudflight's space data algorithm development and optimization services are designed to improve processing and mission outcomes.

Data has become a key part of the NewSpace sector. But to extract the most value from it, in the most efficient manner, it needs to be processed using effective algorithms.

Algorithm optimization is the process of finding the input parameters or arguments to a function that result in the minimum or maximum output of the function.

One of the main tools to achieve this is machine learning (ML), an approach that enables the efficient processing, organization, labeling, and categorization of information, making it easier to learn from and/or apply to real-world problems.

One example of algorithm optimization in action is Cloudflight’s partnership with GRASP, the world’s most complex algorithm for aerosol detection. GRASP is a scientific research project that uses Earth Observation (EO) data on the effects of aerosols (from sandstorms, volcanos, wildfires, pollution, etc.) on the global climate.

With over 5,000 lines of code developed by Cloudflight’s teams, 5.9PB of storage, and 72 virtual machines, the company has supported GRASP in scaling their algorithm even further, achieving a speedup factor of 140x.

Furthermore, Cloudflight has created a cloud-based development environment and cluster infrastructure for GRASP scientists and implemented a decentralized processing system.

In September 2020, GRASP was applied to Sentinel 3 data for the area of Paraguay and used it to distinguish wildfire aerosols from harmless clouds.