Latest Method Predicts Extreme Weather Events More Accurately
This story was originally published by Columbia Engineering.
With the rise of utmost weather events, which have gotten more frequent in our warming climate, accurate predictions have gotten more critical for all of us, from farmers to city-dwellers to businesses all over the world. Up to now, climate models have didn’t accurately predict precipitation intensity, particularly extremes. While in nature, precipitation might be very varied, with many extremes of precipitation, climate models predict a smaller variance in precipitation with a bias toward light rain.
The Missing Piece in Current Algorithms: Cloud Organization
Researchers have been working to develop algorithms that can improve prediction accuracy but, as Columbia Engineering climate scientists report, there was a missing piece of data in traditional climate model parameterizations—a solution to describe cloud structure and organization that’s so fine-scale it just isn’t captured on the computational grid getting used.
These organization measurements affect predictions of each precipitation intensity and its stochasticity—the variability of random fluctuations in precipitation intensity. Thus far, there has not been an efficient, accurate solution to measure cloud structure and quantify its impact.
A recent study from a team led by Pierre Gentine, director of the Learning the Earth with Artificial Intelligence and Physics (LEAP) Center, used global storm-resolving simulations and machine learning to create an algorithm that may deal individually with two different scales of cloud organization: those resolved by a climate model, and people that can’t be resolved as they’re too small. This recent approach addresses the missing piece of data in traditional climate model parameterizations and provides a solution to predict precipitation intensity and variability more precisely.
“Our findings are especially exciting because, for a few years, the scientific community has debated whether to incorporate cloud organization in climate models,” said Gentine, Maurice Ewing and J. Lamar Worzel Professor of Geophysics within the Departments of Earth and Environmental Engineering and Earth Environmental Sciences and a member of the Data Science Institute. “Our work provides a solution to the controversy and a novel solution for including organization, showing that including this information can significantly improve our prediction of precipitation intensity and variability.”
Using AI to Design a Neural Network Algorithm
Sarah Shamekh, a PhD student working with Gentine, developed a neural network algorithm that learns the relevant information concerning the role of fine-scale cloud organization (unresolved scales) on precipitation. Because Shamekh didn’t define a metric or formula upfront, the model learns implicitly—by itself—how you can measure the clustering of clouds, a metric of organization, after which uses this metric to enhance the prediction of precipitation. Shamekh trained the algorithm on a high-resolution moisture field, encoding the degree of small-scale organization.
“We discovered that our organization metric explains precipitation variability almost entirely and will replace a stochastic parameterization in climate models,” said Shamekh, lead creator of the study, published May 8, 2023, by PNAS. “Including this information significantly improved precipitation prediction at the dimensions relevant to climate models, accurately predicting precipitation extremes and spatial variability.”
Future Projections
The researchers at the moment are using their machine-learning approach, which implicitly learns the sub-grid cloud organization metric, in climate models. This could significantly improve the prediction of precipitation intensity and variability, including extreme precipitation events, and enable scientists to higher project future changes within the water cycle and extreme weather patterns in a warming climate.
This research also opens up recent avenues for investigation, equivalent to exploring the potential of precipitation creating memory, where the atmosphere retains details about recent weather conditions, which in turn influences atmospheric conditions in a while, within the climate system. This recent approach could have wide-ranging applications beyond just precipitation modeling, including higher modeling of the ice sheet and ocean surface.
