Neural network-based approach to evaluate convective hazards frequency shift under climate change

This study addresses challenges in predicting extreme weather events, including intense rainfall, hail, and strong winds.Such events pose significant threats to infrastructure and human life, and their frequency has been increasing due to climate change.We show, that integrating climate forecasts with modern machine learning techniques improves prediction accuracy and helps identify regions where these events may become more frequent and dangerous.To achieve reliable predictions, we propose a robust neural network architecture that outperforms several common baselines in accuracy and reliability.Our model leverages problem-specific physics encapsulated in Coupled Model Intercomparison Project data.The analysis also highlight the impact of rugged terrain on the risk distribution of extreme weather events.The