Columbia University scientists developed an algorithm that improves extreme weather prediction. Algorithms consider cloud organization, unlike traditional climate models. Cloud structure determines rain amount and frequency.

The Columbia University 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 can handle two scales of cloud organization: those that climate models can resolve and those that cannot.

PNAS, a prestigious scientific publication, published their crucial discoveries. Predicting the weather is crucial as global warming increases extreme weather events.

In order to accurately predict severe weather, scientists have developed an algorithm.

Climate models underestimate natural rain variance and favor light rain. This makes it difficult to estimate rainfall, especially during intense occurrences.

The study delighted Columbia University geophysics professor Pierre Gentine. “Our findings are especially exciting because, for many years, the scientific community has debated whether to include cloud organization in climate models,” he added.

“Our work gives an answer to the debate and a new way to include organization, showing that this information can greatly improve our ability to predict precipitation intensity and variability,” he added.

Gentine’s PhD student Sarah Shamekh developed a neural network technique that improves predictions using machine learning. This program learns how small-scale cloud organization affects rainfall.

The system employs cloud organization, measured by how near clouds are, to enhance rain forecasts. Shamekh trained the program on a high-resolution moisture field showing small-scale structure.

“We found that our organization metric explains almost all of the differences in precipitation and could replace a stochastic parameterization in climate models,” stated research leader Sarah Shamekh. She said that incorporating this information considerably improved precipitation estimates, making extremes and spatial variation more accurate.

This groundbreaking breakthrough improves weather forecasts and provides new avenues of investigation. The researchers are studying “precipitation memory,” the theory that the environment remembers past weather and utilizes it to alter future weather.

This study might be applied beyond weather prediction. It could model ocean and ice sheets. Adding cloud structure to climate models helps scientists understand and reduce climate change-induced extreme weather occurrences.