Scientists at Columbia University have developed an algorithm that improves the accuracy of predicting extreme weather events. The algorithm addresses the issue of cloud organization, which has been lacking in traditional climate models. Cloud organization plays a crucial role in predicting precipitation intensity and variability.
The research team, led by Pierre Gentine, director of the Learning the Earth with Artificial Intelligence and Physics (LEAP) Center at Columbia University, utilized global storm-resolving simulations and machine learning techniques to create an algorithm capable of handling two distinct scales of cloud organization: those that can be resolved by climate models and those that are too small to be resolved.
Their groundbreaking findings have been published in the prestigious journal Proceedings of the National Academy of Sciences (PNAS). Accurate weather predictions have become increasingly important in light of the rising frequency of extreme weather events caused by global warming.
While precipitation in nature exhibits significant variability, climate models tend to underestimate this variability and often bias towards light rain. Consequently, accurately predicting precipitation intensity, particularly during extreme events, has proven challenging.
Pierre Gentine, a professor of Geophysics at Columbia University, expressed excitement about the study’s results, stating, “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 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.”
To achieve these improved predictions, Sarah Shamekh, a PhD student working with Gentine, developed a neural network algorithm that harnesses the power of machine learning. This algorithm learns the relationship between fine-scale cloud organization and precipitation.
The algorithm autonomously measures the clustering of clouds, a key metric of cloud organization, and employs this metric to enhance precipitation predictions. Shamekh trained the algorithm using a high-resolution moisture field that encodes the level of small-scale organization.
The study’s lead author, Sarah Shamekh, explained, “We discovered that our organization metric explains precipitation variability almost entirely and could replace a stochastic parameterization in climate models.” She further highlighted that including this information significantly improved precipitation predictions, accurately forecasting extremes and spatial variability.
This groundbreaking research not only improves weather prediction accuracy but also opens up new avenues of investigation. The researchers are now exploring the concept of precipitation memory, wherein the atmosphere retains information about recent weather conditions, influencing future atmospheric conditions within the climate system.
The implications of this research extend beyond weather prediction, with potential applications in modeling ice sheets and ocean surfaces. By incorporating cloud organization into climate models, scientists are taking a significant step towards better understanding and mitigating the impacts of extreme weather events driven by climate change.