Accurate forecasts are becoming more important for all of us, from farmers to city inhabitants to enterprises worldwide, as extreme weather occurrences increase in frequency as a result of global warming.
Climate models have so far been unable to correctly estimate precipitation intensity, especially extremes. Climate models forecast a lesser variance in precipitation with a bias toward light rain, however, precipitation can be quite variable in nature and can occur in various extremes.
Missing piece in current algorithms: cloud organization
While scientists have been working to create algorithms that will increase the accuracy of predictions, traditional climate model parameterizations have been lacking a way to describe cloud structure and organization that is so fine-scale that it cannot be captured on the computational grid being used, according to climate scientists at Columbia Engineering.
Predictions of precipitation intensity and its stochasticity the variability of arbitrary fluctuations in precipitation intensity are impacted by these organization measures. There hasn’t been a reliable, efficient approach to gauge cloud structure and estimate its influence until now.
An algorithm that can deal separately with two different scales of cloud organization those resolved by a climate model and those that cannot be resolved because they are too small was developed in a recent study by a team led by Pierre Gentine, director of the Learning the Earth with Artificial Intelligence and Physics (LEAP) Center. The team used global storm-resolving simulations and machine learning.
This innovative method fills in the information gap in conventional climate model parameterizations and offers a means of more accurately predicting precipitation intensity and variability.
We discovered that our organization metric explains precipitation variability almost entirely and could replace a stochastic parameterization in climate models. Including this information significantly improved precipitation prediction at the scale relevant to climate models, accurately predicting precipitation extremes and spatial variability.
Sarah Shamekh
“Our findings are especially exciting because, for many years, the scientific community has debated whether to include cloud organization in climate models,” said Gentine, Maurice Ewing and J. Lamar Worzel Professor of Geophysics in the Departments of Earth and Environmental Engineering and Earth Environmental Sciences and a member of the Data Science Institute. “Our work provides an answer to the debate 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 neural network algorithm
Sarah Shamekh, a Ph.D. student working with Gentine, developed a neural network algorithm that learns the relevant information about the role of fine-scale cloud organization (unresolved scales) on precipitation.
Shamekh did not predetermine a metric or formula, thus the model implicitly learns on its own how to assess the clustering of clouds, a metric of organization, and then makes use of this metric to enhance precipitation prediction. On a high-resolution moisture field, Shamekh taught the algorithm to encode the level of small-scale order.
“We discovered that our organization metric explains precipitation variability almost entirely and could replace a stochastic parameterization in climate models,” said Shamekh, lead author of the study, published May 8, 2023, by PNAS. “Including this information significantly improved precipitation prediction at the scale relevant to climate models, accurately predicting precipitation extremes and spatial variability.”
Machine-learning algorithms will improve future projections
The researchers are currently applying their machine-learning strategy to climate models, which implicitly learns the sub-grid cloud organization metric. This could greatly enhance the ability of scientists to forecast future changes in the water cycle and extreme weather patterns in a warming climate. It should also greatly improve the prediction of precipitation intensity and variability, including extreme precipitation events.
Future work
This study also opens up new lines of inquiry, such as looking into the idea of precipitation-producing memory, in which the atmosphere stores information about recent weather patterns and then uses that information to affect atmospheric conditions in the climate system. Beyond only simulating precipitation, this new method may also be used to more accurately model the ice sheet and the ocean’s surface.
