Predicting Aircraft Turbulence with a Deep Learning Model Applied to Geostationary Infrared Imagery

Aircraft turbulence often occurs near particular wave patterns in geostationary water vapor imagery. These patterns are complex and difficult for forecasters to discern rapidly and reliably in order to issue warnings to aircraft. Here we apply a deep learning convolutional neural network model trained on geostationary satellite data to identify these patterns and produce a probabilistic clear air turbulence forecasting product. We also discuss the implications of this work to general innovations in aviation hazard detection.