School of Meteorology Ph.D. Student Publishes Machine Learning Paper in AMS Journal
[Image: Fronts detected by humans, numerical frontal analysis (NFA), and a convolutional neural network (CNN). [a] Predictors and human fronts. The colour fill is wet-bulb temperature (°C), and the grey barbs are wind velocity, at 1000 mb. The light blue triangles are cold fronts, and dark blue circles are warm fronts, drawn by human meteorologists. [b] Fronts detected by NFA. Dark blue lines are cold fronts; dark red lines are warm fronts; and corresponding frontal zones are shaded in light colours. [c] Fronts detected by a CNN. As in panel b, except that shading shows probabilities. [d-f] As in panels a-c but for a different time step.]
Machine Learning is an ever-growing part of the scientific world, and important work in this sector is being done by Ph.D. student Ryan Lagerquist. This week, Lagerquist’s paper “Deep Learning for Spatially Explicit Prediction of Synoptic-scale Fronts” was published in the American Meteorological Society’s (AMS) journal, Weather and Forecasting. This paper examines the use of convolutional neural nets (CNN) to identify fronts in gridded data. Since Synoptic-scale fronts are often associated with extreme weather in the mid-latitudes, this research can help to identify those fronts across a large amount of data.
Lagerquist had two co-authors; Dr. Amy McGovern, Professor in the OU School of Computer Science and Adjunct Professor in the School of Meteorology, and Dr. David Gagne II, Machine Learning Scientist at the National Center for Atmospheric Research and OU alumnus.