PhD Student
Nilas – a new approach to sea ice forecasting
Nilas - A new approach to sea ice forecasting
Nilas is an artificial intelligence model that produces high resolution sea ice forecasts at one to 15 day range. It’s developed by the Polar Artificial Intelligence team at British Antarctic Survey in collaboration with the University of Cambridge, and is a continuation of the IceNet project, which developed longer range, lower resolution forecasting.
Nilas is a state-of-the-art medium-range (1-15 day), high-resolution (6.25 km) probabilistic sea ice forecasting system developed by the Polar AI Lab at the British Antarctic Survey in collaboration with the University of Cambridge.
Nilas combines AMSR2-derived sea ice concentration with ECMWF AIFS-ENS weather forecasts using machine learning to create an ensemble forecast of possible sea ice conditions over the next fifteen days.
A regional forecasting model, Nilas outperforms all physics-based and AI-based forecasting systems over the Barents Sea in the Arctic and Weddell Sea region in the Antarctic. Future work will adapt the model to other seas throughout the polar regions.
Field trials aboard the RRS Sir David Attenborough in December 2025 demonstrated promising opportunities for Nilas to inform science and operations across strategic and tactical timescales.
More Information to come soon …
-
PolarSim – AI forecasting and modelling
Read more of: PolarSim – AI forecasting and modellingPolarSim delivers AI forecasting, modelling, downscaling and emulation tools for the polar regions
-
Logist – AI decision support tools
Read more of: Logist – AI decision support toolsLogist develops decision support tools that use environmental data to help experts make the best routing and planning decisions.