Environment and Climate Change Canada's Canadian Centre for Climate Modelling and Analysis (CCCma) has developed an AI system to improve seasonal and decadal climate predictions. The system uses advanced neural network models and conditional variational autoencoders to post-process climate forecasts, enhance prediction accuracy, and better capture extreme weather events. This system is currently in development and is used internally by Government of Canada employees to support climate forecasting operations.
The AI system does not collect or process personal information. Instead, it works with climate model outputs and meteorological data to improve the quality and reliability of long-term climate predictions. By using machine learning techniques, the system can identify patterns that help meteorologists better understand and forecast seasonal and decadal climate variations, ultimately supporting better climate information for Canadians.
The system employs both deterministic and generative AI models developed in-house, combining traditional climate science with modern machine learning capabilities. All development and deployment is based on open-source frameworks and principles, ensuring transparency and reproducibility in climate prediction enhancement.
