Algorithms for improved rainfall predictions
The Mediterranean region (MED) is a hot-spot of anthropogenic climate change. To assure effective near-term planning, decisionmakers in weather-dependent sectors depend on skilful forecasts of precipitation on sub-seasonal to seasonal (S2S) timescales. However, some fundamental challenges prevent reliable predictions beyond approximately 10 days. The EU-funded CausalBoost project will apply an innovative method to improve S2S forecasts of MED rainfall. The approach relies on a combination of innovative causal discovery algorithms from machine learning with operational forecast models. The project will identify central S2S drivers of MED rainfall, systematically assess them with prediction models and produce process-based bias corrections.