GPU-Accelerated Deep Learning for Weather, Climate, and Space

We'll demonstrate how to use deep learning to tackle important challenges in weather forecasting, climate modeling, and the processing of satellite observations. We'll present recent results from ongoing research collaborations with the National Oceanic and Atmospheric Administration, NASA, and various universities, and explain how accurate results were achieved. We'll show how to automate feature detection to identify threats from severe weather, solar storms, and near-earth objects, and we'll discuss how to accelerate weather/climate models and data assimilation techniques to produce more accurate predictions. Finally, we'll illustrate how to better use satellite observations by enhancing, transforming, interpolating, fusing, and repairing multispectral data.