Friday 13th January – 2-3pm – Online
Speaker – Zack Labe, Princeton
The popularity of machine learning methods, such as neural networks, is rapidly expanding in nearly all areas of Earth science. The interest in these tools also coincides with a growing influx of big data and the need for high efficiency in solving a range of tasks. In climate science, we often consider signal-to-noise or detection and attribution problems to help disentangle external climate forcing from natural variability.
In this seminar, Zack will show examples of how relatively simple classification problems can be combined with explainable artificial intelligence methods to improve our understanding of historical and future climate projections. We find to make these predictions that the neural networks are often leveraging regional patterns of forced change within observations and climate model large ensembles. These same explainable neural networks can be easily adapted for a wide variety of applications in understanding climate variability and climate change.
Dr. Zachary Labe is a postdoc at NOAA’s Geophysical Fluid Dynamics Laboratory and the Atmospheric and Oceanic Sciences Program at Princeton University. His current research interests explore the intersection of climate variability, extreme events, decadal prediction, and explainable machine learning methods. In addition to academic research, he is very passionate about improving science communication, accessibility, and outreach through engaging climate change data visualizations.