Climate change is causing an increase in extreme weather events around the world, with devastating impacts on people, property, and businesses. To ensure efficient risk management and accurate predictions, accurate models of precipitation intensity are needed. So far, traditional climate models have failed to precisely predict precipitation intensity, usually underestimating extreme rainfall. To tackle this problem, researchers from Columbia Engineering have developed a machine-learning algorithm to capture cloud structure and help accurately predict precipitation intensity.
The new algorithm, developed by Pierre Gentine, director of the Learning the Earth with Artificial Intelligence and Physics (LEAP) Center, utilizes global storm-resolving simulations and machine learning to identify sites of small-scale organization of clouds, which is usually not captured by traditional climate models. Sarah Shamekh, a PhD student working with Gentine, created a deep-learning algorithm which learns the intricacies of how small-scale cloud organization affects precipitation intensity and variance. The algorithm takes in a high-resolution moisture field to train itself, encoding the degree of small-scale organization. The researchers found that their organization metric was able to clarify almost entirely the variability of precipitation, providing a novel solution to predict precipitation intensity with greater accuracy.
This new approach provides an answer to the research question of whether cloud organization needs to be included in climate modeling and will enable better predictions of extreme weather events. Moreover, this method may have applications beyond modeling precipitation, such as modeling of the ocean surface and the ice sheet.
Columbia Engineering is an esteemed Ivy League school located in New York City. The school is the country’s first collegiate engineering school and has world-renowned faculty and innovative, cutting-edge research initiatives.
Pierre Gentine is the Maurice Ewing and J. Lamar Worzel Professor of Geophysics at Columbia University. He is the director of the Learning the Earth with Artificial Intelligence and Physics (LEAP) Center and has several notable awards to his name. Gentine’s research focuses on developing methods to better understand, and predict, Earth’s climate system.
