Probabilistic forecasts of extreme heatwaves using convolutional neural networks in a regime of lack of data

Forecasting extreme climate events, for instance heat waves, is key society and a scientific challenge. In this paper we propose novel machine learning approach that successfully forecasts waves up to 45 days before the end of event. The allows dynamical process studies. A message optimal require large amount data. image shows temperature (colors) geopotential height (lines) anomalies typical atmospheric situation.