IntenseStormNet

The IntenseStormNet model uses GOES ABI and GLM data to detect the most intense regions of convection probabilistically. The model is a convolutional neural network that can be run in real-time on CONUS or mesoscale sectors, and is extensible to other domains with GOES satellite coverage. IntenseStormNet output is used as a predictor in ProbSevere v3 models.

See this blog post for a succinct introduction to IntenseStormNet. Other related posts can be found on the CIMSS satellite blog and GOES-R HWT blog.

Real-time data viewer

Publications and presentations

• A Deep-Learning Model for Automated Detection of Intense Midlatitude Convection Using Geostationary Satellite Images, WAF (2020)
• A Deep-Learning Approach to Severe Convective Weather Prediction Using Geostationary Satellite Data, AMS satellite conference (2019)