Summary
EO has a unique role in monitoring the Polar Regions by providing information that is consistent, repeatable, year-round, and covers the extensive area. In particular, Synthetic Aperture Radar (SAR) missions, such as Sentinel-1, have for many years proven useful in high resolution monitoring due to their capability of acquiring data independent of cloud cover and polar night. However, the size of the polar regions means that relying on human analysis of the large volume of data available is not practical.
The polar user communities have been early adopters of AI/ML applied to EO data. This project addresses the need for a ML platform to better serve the polar user communities. The work will implement MLflow, a well-proven ML platform, augmented by DVC to manage training data.
The operation and benefits of MLflow within Polar TEP will be validated and illustrated through the following showcases:
- Showcase 1: Nature-Based Solutions for Flood and Erosion Protection, and
- Showcase 2: Polar Voyage Planning and Support
Showcase 1 addresses the need for resilient infrastructures to mitigate increased coastal and riverine flood and erosion potential as a consequence of climate change. Employing nature-based methods to mitigate flood and erosion hazards is an environmentally friendly solution.
Showcase 2 responds to the needs of ships and people operating in the polar regions for past, present, and future environmental information, especially concerning sea ice. These regions are experiencing climate change at a rate that is up to five times greater than the rest of the planet. ML applied to environmental data can help mitigate the impact of these changes by providing better information with which to plan and conduct polar operations and thus provide improved safety for people, infrastructure, and the environment.
